W2manual42 Part2 Theory Rev1

CE-QUAL-W2: A Two-Dimensional, Later-
ally Averaged, Hydrodynamic and Water

Quality Model, Version 4.2
User Manual Part 2: Hydrodynamic and Wa-

ter Quality Model Theory
Edited by
Scott A. Wells
Professor
Department of Civil and Environmental Engineering
Portland State University
Portland, OR 97207-0751
Department of Civil and Environmental Engineering

Portland State University
Portland, OR 97207-0751
July 2019
CONTENTS
Contents
Contents ......................................................................................................................................... ii
List of Figures ...............................................................................................................................vii
List of Tables ................................................................................................................................. xi
Preface ...........................................................................................................................................xii
1. Introduction .......................................................................................................................... 1
2. Hydrodynamics Modeling .................................................................................................. 1
Coordinate System.................................................................................................................... 1
Turbulent Time-Averaged Equations...................................................................................... 2
Continuity ................................................................................................................................ 3
x-Momentum Equation.......................................................................................................... 3
y-Momentum Equation.......................................................................................................... 4
z-Momentum Equation.......................................................................................................... 4
Coriolis Effect ......................................................................................................................... 5
Adjusting the Coordinate System ........................................................................................... 5
Governing Equations for General Coordinate System ........................................................ 7
Continuity ................................................................................................................................ 7
x-Momentum Equation.......................................................................................................... 8
y-Momentum Equation.......................................................................................................... 8
z-Momentum Equation.......................................................................................................... 8
Simplification of z-Momentum Equation ................................................................................ 8
Lateral Averaging ...................................................................................................................... 8
Continuity Equation ............................................................................................................... 9
x-Momentum Equation........................................................................................................ 10
Summary of Laterally Averaged Equations ......................................................................... 12
Continuity Equation ............................................................................................................. 12
x-Momentum Equation........................................................................................................ 12
z-Momentum Equation........................................................................................................ 12
Simplification of Pressure Term ............................................................................................ 12
Free Water Surface ................................................................................................................. 14
Equation of State ..................................................................................................................... 16
Summary of Governing Equations ........................................................................................ 16
Linkage of Mainstem Branches with Internal Head Boundary Conditions ..................... 18
Linkage of Tributary or Side Branches................................................................................. 18
Longitudinal Momentum ..................................................................................................... 19
Cross-shear of Tributary Inflow ......................................................................................... 20
Computation of Initial Water Surface Slope and Velocity Field for River
Simulation .............................................................................................................................. 21
Auxiliary Functions .................................................................................................................. 22
Surface Shear Stress .......................................................................................................... 22
Wind Fetch Calculation....................................................................................................... 27
ii
CONTENTS
Bottom Shear Stress ........................................................................................................... 28

Vertical Shear Stress .......................................................................................................... 29
Formulation....................................................................................................................... 32
RNG Turbulent Eddy Viscosity Formulation ................................................................ 34
Nikuradse Formulation.................................................................................................... 35
Parabolic Formulation ..................................................................................................... 36
W2N Formulation............................................................................................................. 37
TKE Formulation .............................................................................................................. 37
Algorithm ....................................................................................................................... 38
Explicit Vertical Convection ........................................................................................ 40
Implicit Vertical Convection ........................................................................................ 42
Boundary and initial conditions for k- model .......................................................... 45
Effect of Vertical Layer Numbers on Vertical Turbulence ......................................... 49
Longitudinal Shear Stress .................................................................................................. 52
Hydraulic Structures ............................................................................................................ 52
Pipes .................................................................................................................................. 52
Internal Weirs ................................................................................................................... 55
Water Level Control or Pumps ...................................................................................... 56
Outlet Structures .............................................................................................................. 56
Spillways/Weirs ................................................................................................................ 59
Gates ................................................................................................................................. 63
Branch Momentum Exchange ........................................................................................... 65
Lateral Inflows ...................................................................................................................... 66
Density .................................................................................................................................. 68
Selective Withdrawal........................................................................................................... 69
Sediment Resuspension .................................................................................................... 70
3. Particle Tracking Algorithm ............................................................................................. 71
4. Temperature and Water Quality Modeling .................................................................... 75
Turbulent Advection-Diffusion Equation .............................................................................. 75
Temperature and Water Quality Transport ......................................................................... 78
Determination of Dz and Dx .................................................................................................... 80
Temperature Auxiliary Functions .......................................................................................... 80
Heat Exchange .................................................................................................................... 80
Surface Heat Exchange.................................................................................................. 80
Evaporation Models for f(W) ...................................................................................... 84
Equilibrium Temperature ............................................................................................ 87
Sediment Heat Exchange .............................................................................................. 88
Dynamic Shading ................................................................................................................ 89
Solar Altitude and Azimuth ............................................................................................. 89
Topographic Shading ...................................................................................................... 91
Vegetative Shading ......................................................................................................... 91
Data Requirements ......................................................................................................... 93
Ice Cover............................................................................................................................... 95
Initial Ice Formation ......................................................................................................... 96
Air-Ice Flux Boundary Condition and Ice Surface Temperature
Approximation................................................................................................................ 96
iii
CONTENTS
Absorbed Solar Radiation by Water Under Ice ........................................................... 97

Ice Melt at Air-Ice Interface ............................................................................................ 97
Ice-Water Flux Boundary Condition Formulation........................................................ 98
Freezing Temperature of Ice ......................................................................................... 98
Gain and Loss of Water .................................................................................................. 99
Overview of Kinetic Source/Sink Term ................................................................................ 99
Generic Constituent .............................................................................................................. 102
Conservative Tracer .......................................................................................................... 104
Coliform Bacteria with decay and photodegradation ................................................... 104
N2 dissolved gas and TDG ............................................................................................... 105
Volatile Generic Constituent ............................................................................................ 109
Water Age or Residence time.......................................................................................... 110
Inorganic Suspended Solids ................................................................................................ 111
Total Dissolved Solids or Salinity ........................................................................................ 112
Labile DOM ............................................................................................................................ 112
Refractory DOM ..................................................................................................................... 114
Labile Particulate Organic Matter ....................................................................................... 114
Refractory Particulate Organic Matter ................................................................................ 116
Carbonaceous Biochemical Oxygen Demand (CBOD) ................................................... 117
Carbonaceous Biochemical Oxygen Demand - Phosphorus (CBODP) ....................... 118
Carbonaceous Biochemical Oxygen Demand - Nitrogen (CBODN) ............................. 119
Algae ....................................................................................................................................... 120
Periphyton/Epiphyton............................................................................................................ 123
Macrophytes ........................................................................................................................... 127
Modeling Frictional Force ................................................................................................. 130
Modeling Porosity .............................................................................................................. 131
Changes to Governing Equations ................................................................................... 131
Zooplankton............................................................................................................................ 133
Phosphorus ............................................................................................................................ 135
Ammonium ............................................................................................................................. 138
Nitrate-Nitrite .......................................................................................................................... 140
Dissolved Silica ...................................................................................................................... 141
Organic Matter Variable Stoichiometry .............................................................................. 142
Labile Dissolved Organic Matter – Phosphorus (LDOM-P) ........................................ 142
Refractory Dissolved Organic Matter – Phosphorus (RDOM-P) ................................ 143
Labile Particulate Organic Matter – Phosphorus (LPOM-P) ....................................... 143
Refractory Particulate Organic Matter – Phosphorus (RPOM-P) .............................. 144
Labile Dissolved Organic Matter – Nitrogen (LDOM-N) .............................................. 145
Refractory Dissolved Organic Matter – Nitrogen (RDOM-N) ...................................... 145
Labile Particulate Organic Matter – Nitrogen (LPOM-N) ............................................. 146
Refractory Particulate Organic Matter – Nitrogen (RPOM-N) .................................... 147
Particulate Biogenic Silica .................................................................................................... 147
Total Iron ................................................................................................................................. 148
Dissolved Oxygen ................................................................................................................. 149
iv
CONTENTS
River Reaeration Equations ............................................................................................. 152

Lake Reaeration Equations ............................................................................................. 154
Estuarine Equations .......................................................................................................... 158
Reaeration Temperature Dependence .......................................................................... 159
Dam Reaeration................................................................................................................. 160
Small Dams or Weirs .................................................................................................... 160
Large Dam Spillways/Gates ........................................................................................ 161
DO Impacts of Spillways .............................................................................................. 162
SYSTDG Algorithm ....................................................................................................... 163
Dissolved Oxygen Saturation Computations............................................................. 167
Sediments............................................................................................................................... 167
Sediment Variable Stoichiometry and Kinetics ................................................................. 169
Sediment Phosphorus ...................................................................................................... 170
Sediment Nitrogen............................................................................................................. 170
Sediment Carbon............................................................................................................... 171
Sediment Diagenesis Model ................................................................................................ 172
Sediment Phosphorus ...................................................................................................... 172
Sediment Production of Methane and Sulfide .............................................................. 175
Consumption of Dissolved Oxygen due to Sediment Resuspension ........................ 177
Wind Induced Resuspension ....................................................................................... 178
Bottom Scour Resuspension ....................................................................................... 178
Mass Balance Equations for Particulate Organic Matter............................................. 179
Dynamic Calculation of Sediment pH and Temperature ............................................. 179
pH..................................................................................................................................... 179
Sediment Total Inorganic Carbon ............................................................................... 179
Sediment Alkalinity ........................................................................................................ 182
Sediment Temperature ..................................................................................................... 184
Metal Complexation and Diagenesis .............................................................................. 187
Ferrous Iron Fe(II) ......................................................................................................... 187
Iron Oxyhydroxide FeOOH(s) ...................................................................................... 190
Manganese Mn(II) ......................................................................................................... 193
Manganese Dioxide MnO2 ........................................................................................... 196
Total Inorganic Carbon ......................................................................................................... 198
Alkalinity .................................................................................................................................. 201
pH and Carbonate Species.................................................................................................. 201
Temperature Rate Multipliers .............................................................................................. 204
5. Numerical Solution.......................................................................................................... 207
Characteristics of the Finite Difference Scheme .............................................................. 207
Hydrodynamic Equations Numerical Solution................................................................... 208
Characteristics of the Hydrodynamics Numerical Solution ......................................... 208
Free-Water Surface Numerical Solution for  ............................................................... 209
Horizontal Momentum Numerical Solution for Horizontal Velocity U ........................ 215
Explicit Solution.............................................................................................................. 215
Implicit Solution .............................................................................................................. 217
Solution of Vertical Velocity W ........................................................................................ 219
Numerical Stability............................................................................................................. 219
v
CONTENTS
Advective-Diffusion Equation Numerical Solution ............................................................ 221

Characteristics of the Constituent Numerical Solution ................................................ 221
Solution of Advective-Diffusion Equation ....................................................................... 221
Advective Transport in x and z ........................................................................................ 223
Non-Uniform Grid QUICKEST Formulation ............................................................... 225
ULTIMATE/QUICKEST Numerical Transport Solution Scheme ............................ 227
Vertical Implicit Transport ............................................................................................. 233
6. References....................................................................................................................... 235
vi
LIST OF FIGURES
List of Figures
Figure 1. Definition sketch of coordinate system for governing equations where x is

oriented east, y is oriented north, and z is oriented upward. .......................................... 2
Figure 2. Definition sketch of turbulent time averaging for velocity. ..................................... 3
Figure 3. Definition sketch of turbulent shear stresses in x-direction. ................................. 4
Figure 4. Sketch of turbulent shear stresses in y-direction. .................................................. 4
Figure 5. Sketch of turbulent shear stresses in z-direction. .................................................. 5
Figure 6. General coordinate system with z-axis compatible with original derivation of
W2 model. .............................................................................................................................. 6
Figure 7. Sketch of channel slope and coordinate system for W2 where the x-axis is
oriented along the channel slope. ....................................................................................... 7
Figure 8. Lateral average and deviation from lateral average components of longitudinal
velocity. ................................................................................................................................... 9
Figure 9. Schematization for simplification of pressure term. .............................................. 13
Figure 10. Coordinate system without channel slope. ......................................................... 14
Figure 11. Coordinate system with channel slope. ............................................................... 15
Figure 12. Definition sketch for channel slope. ..................................................................... 17
Figure 13. Transfer of mass and momentum between branches with unequal vertical
grid spacing. ......................................................................................................................... 18
Figure 14. Linkage of tributary or side branch coming in at an angle to main branch...... 19
Figure 15. Schematic of x and y velocity components where Θmain is the angle of the
main branch segment and Θtrib is the angle of the tributary segment. Ux is the x-
velocity component of the tributary branch into the main segment. Uy is the y-
velocity component of the tributary branch into the main segment. ............................ 20
Figure 16. Shear stress at the air-water surface. .................................................................. 23
Figure 17. Segment orientation. .............................................................................................. 24
Figure 18. Wind orientation. ..................................................................................................... 25
Figure 19. Comparison of current W2 model computation of CD and that recommended
as a lower limit by Wuest and Lorke (2003). ................................................................... 26
Figure 20. Surface shear stress for CE-QUAL-W2 Version 3.6 and later compared to
Version 3.5 and earlier. ...................................................................................................... 27
Figure 21. Illustration of fetch distance calculation. ............................................................... 28
Figure 22. Variation of turbulent vertical eddy viscosity for flow of 2574 m3 s-1 flow down
a channel of length 30 km and width of 100 m at x=15 km. ......................................... 31
Figure 23. Variation of turbulent vertical eddy viscosity for flow of 2574 m3 s-1 flow down
a channel of length 30 km and width of 100 m measured at x=15 km. ...................... 31
Figure 24. Comparison of vertical velocity predictions of W2 model with various eddy
viscosity models compared to theory. .............................................................................. 32
Figure 25. Conceptual diagram of wind induced motion...................................................... 32
vii
LIST OF FIGURES
Figure 26. Mixing length as a function of depth for the Nikuradse formulation. ................ 36
Figure 27. Variation of Az with depth for the parabolic model of Engelund (1978). .......... 37
Figure 28. CE-QUAL-W2 computational grid. Width, density, pressure and water quality
state variables are defined at cell centers. Horizontal velocity, longitudinal eddy
viscosity and diffusivity, and longitudinal shear stress are defined at the right-hand
side of the cell. Vertical velocity and vertical diffusivity are defined at the bottom of
the cell, and the vertical eddy viscosity is defined at the lower right corner of the cell.
............................................................................................................................................... 39
Figure 29. Comparison of CE-QUAL-W2 model predictions of turbulent dissipation with
field data from Nakagawa, Neuzu, and Ueda (1975). ................................................... 48
Figure 30. Comparison of CE-QUAL-W2 model predictions of turbulent kinetic energy
with field data from Nakagawa, Neuzu, and Ueda (1975). ........................................... 48
Figure 31. Variation of Manning's friction factor using formulae from Limerinos (1970)
and Jarrett (1984, 1985) for a channel slope, S, of 0.0005 and 84th pecentile
diameter of the bed material, d84, of 50. .......................................................................... 49
Figure 32. Comparison of vertical velocity predictions with one, three, and seven vertical
layers ..................................................................................................................................... 50
Figure 33. Comparison of elevation drop of W2 model with one, three, and seven
vertical layers with same Manning's friction factor. ........................................................ 51
Figure 34. Schematic of linkage of model segments with a culvert. .................................. 52
Figure 35. Linkage schematic of model segments with a culvert. ...................................... 55
Figure 36. Computed versus observed flow using dynamic culvert model. ...................... 55
Figure 37. Schematic representation of internal weirs. ........................................................ 56
Figure 38. Radial gates and spillway flow. .............................................................................. 57
Figure 39. Flow rate over a spillway or weir for submerged and free flowing conditions.
............................................................................................................................................... 62
Figure 40. Flow at a submerged weir...................................................................................... 62
Figure 41. Flow rate variation with gate opening. ................................................................. 64
Figure 42. Selective withdrawal with outflow connected to a valve with a gate. .............. 65
Figure 43. Schematic of branch connection. ......................................................................... 66
Figure 44. From Goodwin et al. (2001) illustrating the particle transport through the CE-
QUAL-W2 grid. .................................................................................................................... 72
Figure 45. Plan view of x and y coordinates within a layer (K) and segment (I). DLX is
the segment length and B is the segment width at the given layer K. ........................ 73
Figure 46. Side view of z and y coordinates within a layer (K) and segment (I). DZ is the
layer thickness and B is the layer width. .......................................................................... 74
Figure 47. Definition sketch for lateral and longitudinal velocities within the cells.
Withdrawals are always assumed to be on the LHS (left hand side) of the segment.
............................................................................................................................................... 74
Figure 48. Velocity variability with time. .................................................................................. 75
Figure 49. Concentration variability with time. ....................................................................... 76
Figure 50. Lateral average of the velocity field...................................................................... 78
Figure 51. Lateral average of the concentration field. ......................................................... 78
Figure 52. Surface heat exchange term-by-term formulation............................................... 81
Figure 53. Short-wave solar penetration in a water body. .................................................... 84
viii
LIST OF FIGURES
Figure 54. Comparison of the wind speed formualtion for Ryan-Harleman and W2 default
(for Tair=15ºC, Tdew=-5ºC, Tsurface=25ºC). .......................................................................... 86
Figure 55. Concept of the equilibrium temperature, TE, where the net surface heat flux is
defined as zero. ................................................................................................................... 87
Figure 56. Schematic of solar altitude, Ao, and azimuth, AZ. ............................................... 90
Figure 57. Schematic of topographic and vegetative shading, solar altitude (0), and
vegetation height (T) and their effect on shadow length. .............................................. 91
Figure 58. Azimuth angle, A, and stream orientation, 0. .................................................. 92
Figure 59. Relationship between azimuth, stream orientation, and shadow length. ...... 93
Figure 60. Ice formation and melting water balance.............................................................. 99
Figure 61. Internal flux for generic constituent compartment. ............................................ 102
Figure 62. Internal flux for coliform bacteria. ....................................................................... 104
Figure 63. N2 gas exchange. ................................................................................................... 106
Figure 64. Henry's Law constant dependence on temperature. ........................................ 107
Figure 65. Vapor pressure as a function of air temperature and relative humidity. ........ 108
Figure 66. Atmospheric pressure correction as a result of vapor pressure correction
assuming original atmospheric pressure is 1 atm. ....................................................... 109
Figure 67. Water age in Chester Morse Lake, WA. ............................................................. 111
Figure 68. Internal flux for inorganic suspended solids...................................................... 112
Figure 69. Internal flux between labile DOM and other compartments ............................ 113
Figure 70. Internal flux between refractory DOM and other compartments. .................... 114
Figure 71. Internal flux between Labile POM and other compartments. ......................... 115
Figure 72. Internal flux between refractory POM and other compartments. ................... 116
Figure 73. Internal flux between CBOD and other compartments in Version 3.6 and
earlier. ................................................................................................................................. 117
Figure 74. Internal flux between CBOD and other compartments in Version 3.7 and
later...................................................................................................................................... 118
Figure 75. Internal flux between CBODP and other compartments. ................................. 119
Figure 76. Internal flux between CBODN and other compartments. ................................. 119
Figure 77. Internal flux between algae and other compartments. .................................... 120
Figure 78. Internal flux between epiphyton and other compartments. ............................. 124
Figure 79. How periphyton transfer organic matter to the water column at death and to
the sediment during burial. .............................................................................................. 127
Figure 80. Nutrient fluxes for the macrophyte compartment in CE-QUAL-W2............... 128
Figure 81. Zooplankton source/sinks. ................................................................................... 134
Figure 82. Internal flux between phosphorus and other compartments. ......................... 135
Figure 83. Internal flux between ammonium and other compartments. .......................... 138
Figure 84. Internal flux between nitrate + nitrite and other compartments...................... 140
Figure 85. Internal flux between dissolved silica and other compartments. .................... 141
Figure 86. Internal flux between particulate biogenic silica and other compartments ... 147
Figure 87. Internal flux between total iron and other compartments............................... 148
Figure 88. Internal flux between dissolved oxygen and other compartments. ............... 149
Figure 89. Reaeration coefficient as a function of flow rate. ............................................. 154
Figure 90. Variation of wind speed and KL for lake/reservoir equations. ........................ 156
Figure 91. Wind speed of 5 m s-1 and a fetch of 5 km corrected to 10 m as a function of
measuring height on land................................................................................................. 157
ix
LIST OF FIGURES
Figure 92. Wind speed of 5 m s-1 corrected to 10 m as a function of fetch. ..................... 157
Figure 93. Wind corrected to 10 m based on wind measured on land. ........................... 158
Figure 94. Variation of KLT/KL20 as a function of temperature............................................ 160
Figure 95. Internal flux between 0-order sediment compartment and other
compartments. ................................................................................................................... 168
Figure 96. Internal flux between 1st-order sediment compartment and other
compartments. ................................................................................................................... 168
Figure 97. Schematic of sediment phosphate model (DiToro, 2001). .............................. 172
Figure 98. Internal flux between phosphate within the aerobic sediment layer 1 and
other compartments. ......................................................................................................... 173
Figure 99. Internal flux between phosphate within the anaerobic sediment layer 2 and
other compartments. ......................................................................................................... 173
Figure 100. Sources and sinks for methane and hydrogen sulfide. ................................. 176
Figure 101. Schematic of sediment inorganic carbon model. ............................................ 180
Figure 102 Internal flux between inorganic carbon within the aerobic sediment layer 1
and other compartments. ................................................................................................. 180
Figure 103 Internal flux between inorganic carbon within the anaerobic sediment layer 2
and other compartments. ................................................................................................. 181
Figure 104. Sources and sinks of the sediment alkalinity model. ...................................... 183
Figure 105. Schematic of sediment temperature model. .................................................... 185
Figure 106. Schematic of iron model (Ditoro, 2001). ........................................................... 188
Figure 107. Schematic of manganese model (DiToro, 2001). ........................................... 194
Figure 108. Internal flux between inorganic carbon and other compartments. ............. 198
Figure 109. Temperature rate multiplier function. .............................................................. 206
Figure 110. Computational grid variable definitions for no channel slope. ...................... 207
Figure 111. Computational grid variable definitions for arbitrary channel slope. ........... 208
Figure 112. Solution of the water level, . ............................................................................. 214
Figure 113. Grid for computation of a flow boundary in x. U, the longitudinal velocity, at
the flow boundary or at the right-hand side of inactive segment 1 is computed from
the flow rate........................................................................................................................ 214
Figure 114. Grid for computation of a head boundary in x. U, the longitudinal velocity, at
the flow boundary or at the right-hand side of inactive segment 1 is computed from
the head difference between the boundary head (at center of inactive segment 1)
and the head at the first grid cell (segment 2). ............................................................. 215
Figure 115. Solution of vertical velocity, w, from continuity equation. .............................. 219
Figure 116. Horizontal advective and diffusive transport. ................................................... 221
Figure 117. Vertical advective and diffusive transport......................................................... 222
Figure 118. Variable locations in computational grid.......................................................... 224
Figure 119. ULTIMATE schematization for positive flow. .................................................. 228
Figure 120. ULTIMATE schematization for negative flow. ................................................ 228
Figure 121. Definition sketch for monotonic solution domain. .......................................... 230
Figure 122. Comparison of UPWIND, QUICKEST, and ULTIMATE/QUICKEST schemes
for conservative tracer transport. .................................................................................... 231
Figure 123. Definition sketch for variable velocity. ............................................................ 232
x
LIST OF TABLES
List of Tables
Table 1. Governing equations with and without channel slope. ......................................... 17

Table 2. Vertical eddy viscosity formulations. ........................................................................ 29
Table 3. Constants in k-ε model (Rodi, 1993)......................................................................... 39
Table 4. HEC-RAS flow rates through weirs and sluice gates............................................ 58
Table 5. List of weir types (French, 1985; USBR, 2001) ..................................................... 60
Table 6. Typical Evaporation Formulae for Lakes and Reservoirs .................................... 86
Table 7. Criteria for determining sunward bank .................................................................... 94
Table 8. CE-QUAL-W2 Water Quality State Variables ...................................................... 100
Table 9. Comparison of α estimates based on Chamberlin and Mitchell (1978) with
additional values (Bowie et al., 1985) ............................................................................ 103
Table 10. CGKLF input values for hydrocarbons, CO2, and N2. ........................................ 110
Table 11. Parameters and values used for macrophytes in the Columbia Slough model.
............................................................................................................................................. 132
Table 12. Parameters used in the zooplankton model. ....................................................... 134
Table 13. Constituents used for variable stoichiometry of organic matter. ...................... 142
Table 14. River reaeration equations. ................................................................................... 152
Table 15. Lake reaeration equations as a function of wind speed at 20oC. .................... 154
Table 16. Reaeration equations for estuarine waterbody at 20oC.................................... 158
Table 17. Formulae for small dam or weir reaeration effects............................................ 160
Table 18. Equations used in CRiSP model for gas production. ........................................ 161
Table 19. Equations used in CRiSP model for gas production for Columbia and Snake
Dams. .................................................................................................................................. 162
Table 20. Spillways and weirs reaeration ............................................................................. 162
Table 21. List of spillway flow TDG production equations in SYSTDG from Scheider and
Hamilton (2015b). .............................................................................................................. 163
Table 22. Coefficients for Spillway TDG Production in SYSTDG from Scheider and
Hamilton (2015b). .............................................................................................................. 164
Table 23. Spill Pattern Identification with Exponential Coefficient (C) specified in
SYSTDG* ........................................................................................................................... 165
Table 24. List of powerhouse flow entrainment equations in SYSTDG ............................ 166
Table 25. Areal nitrification rates of sediment nitrifiers of Onondaga Lake and the
Seneca River, New York (Pauer and Auer, 2000). ...................................................... 169
xi
Preface
This manual documents the two-dimensional, laterally averaged, hydrodynamic and water quality
model CE-QUAL-W2. As in all complex models, there have been many contributors. This re-write
of the User Manual was based on prior User Manuals: Environmental and Hydraulic Laboratories
(1986), Cole and Buchak (1995) Version 2, and Cole and Wells (2000) Version 3.0 through Cole
and Wells (2019) Version 4.1. Hence, one can think of the primary author as merely an editor of
past documents, rather than reflecting one person’s sole authorship. This updated User Manual
contains numerous corrections, new figures, new sections, additional documentation, and im-
provements in organization and presentation of information compared to Cole and Wells (2019).
This section of the User Manual Part 2 documents the theory for the hydrodynamic and water
quality computations and particle transport.
The other sections of the User Manual are divided into multiple sections for ease of updating and
editing:
• User Manual Part 1: Introduction to CE-QUAL-W2, Model download package, how to run
the model, model versions, changes between model versions
• User Manual Part 2: Theoretical basis for CE-QUAL-W2: hydrodynamics and water quality,
particle transport and numerical scheme
• User Manual Part 3: Model input and output file descriptions and input/output file exam-
ples
• User Manual Part 4: Model examples
• User Manual Part 5: Release notes, bug fixes, differences in model versions, history of bug
fixes, and other user manuals such as for the GUI interface, the Waterbalance algorithm,
and other external codes.
This report should be cited as follows:
Wells, S. A. (2019) "CE-QUAL-W2: A two-dimensional, laterally averaged, hydrodynamic and wa-

ter quality model, version 4.2, user manual part 2, hydrodynamic and water quality model the-
ory," Department of Civil and Environmental Engineering, Portland State University, Portland, OR.
or if a specific section with a different primary author:
Berger, C. (2019) “Macrophytes” in "CE-QUAL-W2: A two-dimensional, laterally averaged, hydro-

dynamic and water quality model, version 4.2, user manual part 1, introduction," ed. by S. Wells,
Department of Civil and Environmental Engineering, Portland State University, Portland, OR.
xii
1. Introduction
This section of the CE-QUAL-W2 User manual outlines the governing equations for hydrodynamics and
water quality that are the basis of the model. These equations include
• fluid continuity equation (the integrated form is the water surface equation),
• x and z momentum equations
• equation of state
• temperature and water quality advective diffusion equations
• source and sink term for temperature and each water quality state variable
Also, auxiliary operations of the model are discussed in this section such as wind shear stress, bottom shear
stress, turbulence closure, particle transport, ice cover dynamics, and sediment resuspension.
The numerical solution of the governing equations is discussed for both hydrodynamics and the advective
diffusion equation for temperature and water quality.
The manual is divided into the following sections:

(1) Hydrodynamics and Transport Modeling
(2) Particle Tracking Algorithm
(3) Temperature and Water Quality Modeling
(4) Numerical Solution
(5) References
2. Hydrodynamics Modeling
Contributors: Scott A. Wells, Chris Berger, Tom M. Cole, Robert Annear, Sam Gould, John Edinger,
Ed Buchak
CE-QUAL-W2 uses the laterally averaged equations of fluid motion derived from the three-dimensional
equations, which consist of six equations and six unknowns. The development of the final set of governing
equations is described below.
Coordinate System
The general coordinate system used in the development of the laterally averaged equations of fluid motion
is shown in Figure 1.
1
earth’s rotation
g

z
y
x

earth’s center equator
meridian
Figure 1. Definition sketch of coordinate system for governing equations where x is oriented
east, y is oriented north, and z is oriented upward.
Based on this coordinate system, we then will use the following definitions:
• x and y are horizontal coordinates

• z is a vertical coordinate assumed to be opposite the gravity force
•  is a vector that represents the angular velocity of the earth spinning on its axis. The rotation of
the coordinate system can result in horizontal acceleration of the fluid. This is usually restricted
to large water bodies such as large lakes and ocean systems. The force that causes horizontal
accelerations because of the spinning coordinate system is termed the Coriolis force.
•  is the latitude of the water body
• The meridian represents the longitude of the water body.
Turbulent Time-Averaged Equations

The governing equations are obtained by performing a mass and a momentum balance of the fluid phase
about a control volume. The resulting equations are the continuity (or conservation of fluid mass) and the
conservation of momentum equations for a rotating coordinate system (Batchelor, 1967; Sabersky et al.,
1989; Cushman-Roisin, 1994). Using the coordinate system in Figure 1, we will assume:
1. Incompressible fluid (it is possible to develop compressible equations of motion for special
cases, see Wells, 2009)
2. Centripetal acceleration is a minor correction to gravity
3. Boussinesq approximation where
2 Hydrodynamics and Transport

4. All velocities (u, v, w) and pressure (p) are the sum of turbulent time averages and deviations from
1 𝑡+𝑇
that average, i.e., 𝑢 = 𝑢̄ + 𝑢′ , where 𝑢̄ = ∫𝑡 𝑢𝑑𝑡 as shown in Figure 2, 𝑣 = 𝑣̄ + 𝑣 ′ ;𝑤 = 𝑤̄ + 𝑤 ′
𝑇
and 𝑝 = 𝑝̄ + 𝑝′ where the overbar represents time averaged and the prime represents deviation
from the temporal average.
u
u
time
t t+T
Figure 2. Definition sketch of turbulent time averaging for velocity.
The governing equations (continuity and x, y, z momentum equations) become after simplification:
Continuity
𝜕𝑢̄ 𝜕𝑣̄ 𝜕𝑤̄
𝜕𝑥
+ 𝜕𝑦 + 𝜕𝑧
=0 (2-1)
where:
u = x-direction velocity
v = y-direction velocity
w = z-direction velocity
x-Momentum Equation
𝜕𝑢̄ 𝜕𝑢̄ 𝜕𝑢̄ 𝜕𝑢̄ 1 𝜕𝑝̄ 𝜇 𝜕2 𝑢̄ 𝜕2 𝑢̄ 𝜕2 𝑢̄
⏟ + 𝑢̄ 𝜕𝑥 + 𝑣̄ 𝜕𝑦 + 𝑤̄ 𝜕𝑧 − 2𝛺
⏟ 𝑧 𝑣̄ + 2𝛺𝑦 𝑤̄ = − 𝜌⏟ + 𝜌 (𝜕𝑥 2 + 𝜕𝑦2 + 𝜕𝑧2 ) +
𝜕𝑡 ⏟ 𝜕𝑥 ⏟
unsteady convective acceleration Coriolis acceleration pressure viscous stresses
acceleration gradient
1 𝜕𝜏𝑥𝑥 𝜕𝜏𝑥𝑦 𝜕𝜏
( 𝜕𝑥 + 𝜕𝑦 + 𝜕𝑧𝑥𝑧 ) (2-2)
⏟
𝜌
turbulent stresses
where:
xx = turbulent shear stress acting in x direction on the x-face of control volume (Figure 3)
xy = turbulent shear stress acting in x direction on the y-face of control volume (Figure 3)
xz = turbulent shear stress acting in x direction on the z-face of control volume (Figure 3)
 = dynamic viscosity
 = component of Coriolis acceleration where:
z = 𝛺𝐸 𝑠𝑖𝑛 𝜑
y = 𝛺𝐸 𝑐𝑜𝑠 𝜑
 = latitude
E = earth’s rotation rate
Hydrodynamics and Transport 3

xz
yy
z xy
x
y
Figure 3. Definition sketch of turbulent shear stresses in x-direction.
y-Momentum Equation
𝜕𝑣̄ 𝜕𝑣̄ 𝜕𝑣̄ 𝜕𝑣̄ 1 𝜕𝑝̄ 𝜇 𝜕2 𝑣̄ 𝜕2 𝑣̄ 𝜕2 𝑣̄ 1 𝜕𝜏𝑦𝑥 𝜕𝜏𝑦𝑦
𝜕𝑡
+ 𝑢̄ 𝜕𝑥
+ 𝑣̄ 𝜕𝑦 + 𝑤̄ 𝜕𝑧 + 2𝛺𝑧 𝑢̄ − 2𝛺𝑥 𝑤̄ = − 𝜌 𝜕𝑦 + 𝜌 (𝜕𝑥 2 + 𝜕𝑦2 + 𝜕𝑧2 ) + 𝜌 ( 𝜕𝑥
+ 𝜕𝑦
+
𝜕𝜏𝑦𝑧
𝜕𝑧
) (2-3)
where:
yx = turbulent shear stress acting in y direction on the x-face of control volume (Figure 4)
yy = turbulent shear stress acting in y direction on the y-face of control volume (Figure 4)
yz = turbulent shear stress acting in y direction on the z-face of control volume (Figure 4)
x = 0
yz
y
x
z y
y
x
y
Figure 4. Sketch of turbulent shear stresses in y-direction.
z-Momentum Equation
𝜕𝑤̄ 𝜕𝑤̄ 𝜕𝑤̄ 𝜕𝑤̄ 1 𝜕𝑝̄ 𝜇 𝜕2 𝑤̄ 𝜕2 𝑤̄ 𝜕2 𝑤̄
𝜕𝑡
+ 𝑢̄ 𝜕𝑥
+ 𝑣̄ 𝜕𝑦
+ 𝑤̄ 𝜕𝑧
− 2𝛺𝑦 𝑢̄ + 2𝛺𝑥 𝑣̄ = −𝑔 − 𝜌 𝜕𝑧 + 𝜌 ( 𝜕𝑥 2 + 𝜕𝑦2 + 𝜕𝑧2 ) +
1 𝜕𝜏𝑧𝑥 𝜕𝜏𝑧𝑦 𝜕𝜏𝑧𝑧
(
𝜌 𝜕𝑥
+ 𝜕𝑦
+ 𝜕𝑧
) (2-4)
where:

zx = turbulent shear stress acting in z direction on the x-face of control volume (Figure 5)
zy = turbulent shear stress acting in z direction on the y-face of control volume (Figure 5)
zz = turbulent shear stress acting in z direction on the z-face of control volume (Figure 5)
x = 0
zz
zx
zy
z
x
y
Figure 5. Sketch of turbulent shear stresses in z-direction.
Note that the turbulent shear stresses are defined as follows:
𝜏𝑥𝑥 = 𝜌𝑢′ 𝑢′
𝜏𝑥𝑦 = 𝜌𝑢′ 𝑣 ′ is the same as 𝜏𝑦𝑥 = 𝜌𝑣 ′ 𝑢′
𝜏𝑥𝑧 = 𝜌𝑢′ 𝑤 ′ is the same as 𝜏𝑧𝑥 = 𝜌𝑤 ′ 𝑢′
𝜏𝑦𝑦 = 𝜌𝑣 ′ 𝑣 ′
𝜏𝑦𝑧 = 𝜌𝑣 ′ 𝑤 ′ is the same as 𝜏𝑧𝑦 = 𝜌𝑤 ′ 𝑣 ′
𝜏𝑧𝑧 = 𝜌𝑤 ′ 𝑤 ′
Coriolis Effect
As noted above, all the x terms are zero and can be eliminated from the y and z-momentum equations. If
one integrates over the y-direction (therefore assuming the net velocity in y is zero) and assumes that the
horizontal length scale is much greater than vertical length scale, it can be shown by using scaling arguments
that the Coriolis acceleration forces are negligible (Cushman-Roisin, 1994). Hence, prior to lateral averaging,
the Coriolis acceleration terms will be neglected.
Adjusting the Coordinate System

The coordinate system is transformed into a form compatible with the original W2 development where the
vertical axis is in the direction of gravity. In addition, as shown in Figure 6, the coordinate system is oriented
along an arbitrary slope.

Coordinate System
h gravity
y

z x
v
u
w
Figure 6. General coordinate system with z-axis compatible with original derivation of W2
model.
The gravity acceleration is a body force that is then represented by a vector:
𝑔⃑ = −𝑔𝛻⃑⃑ℎ (2-5)
where:
h = surface normal from the earth’s surface

g = gravitational acceleration, 9.8 m s-2.
This term can be written as three vector components:
𝜕ℎ
𝑔𝑥 = −𝑔 𝜕𝑥 (2-6)
𝜕ℎ
𝑔𝑦 = −𝑔 𝜕𝑦 (2-7)
𝜕ℎ
𝑔𝑧 = −𝑔 (2-8)
𝜕𝑧
These gravity components can be applied to an arbitrary channel slope as shown in Figure 7.

h
x
z
Figure 7. Sketch of channel slope and coordinate system for W2 where the x-axis is oriented
along the channel slope.
The channel slope, So, can also be incorporated into the definition of the gravity vector if the x-axis is chosen
parallel to the channel slope. The channel slope is defined as:
𝑆𝑜 = 𝑡𝑎𝑛 𝛼 (2-9)
and also in terms of the gravity components:
𝜕ℎ
𝑔𝑥 = −𝑔 = 𝑔 𝑠𝑖𝑛 𝛼 (2-10)
𝜕𝑥
𝜕ℎ
𝑔𝑧 = −𝑔 𝜕𝑧 = 𝑔 𝑐𝑜𝑠 𝛼 (2-11)
𝜕ℎ
The gravity acceleration in y is assumed negligible since
𝜕𝑦
= 0 in the lateral direction of the channel.
Governing Equations for General Coordinate System

After redefining the coordinate system, eliminating Coriolis effects, and neglecting viscous shear stresses
the governing equations become:
Continuity
𝜕𝑢̄ 𝜕𝑣̄ 𝜕𝑤̄
𝜕𝑥
+ 𝜕𝑦 + 𝜕𝑧
=0 (2-12)

x-Momentum Equation
𝜕𝑢̄ 𝜕𝑢̄ 𝜕𝑢̄ 𝜕𝑢̄ 1 𝜕𝑝̄ 1 𝜕𝜏𝑥𝑥 𝜕𝜏𝑥𝑦 𝜕𝜏
⏟ + 𝑢̄ 𝜕𝑥 + 𝑣̄ 𝜕𝑦 + 𝑤̄ 𝜕𝑧 =⏟
𝑔 𝑠𝑖𝑛 𝛼 − + ( + + 𝜕𝑧𝑥𝑧)
𝜕𝑡 ⏟ ⏟
𝜌 𝜕𝑥 ⏟
𝜌 𝜕𝑥 𝜕𝑦
unsteady gravity
convective acceleration pressure turbulent shear stresses
(2-13)
y-Momentum Equation
𝜕𝑣̄ 𝜕𝑣̄ 𝜕𝑣̄ 𝜕𝑣̄ 1 𝜕𝑝̄ 1 𝜕𝜏 𝜕𝜏 𝜕𝜏
⏟ + 𝑢̄ 𝜕𝑥 + 𝑣̄ 𝜕𝑦 + 𝑤̄ 𝜕𝑧 =− + 𝜌 ( 𝜕𝑥𝑦𝑥 + 𝜕𝑦𝑦𝑦 + 𝜕𝑧𝑦𝑧) (2-14)
𝜕𝑡 ⏟ ⏟
𝜌 𝜕𝑦 ⏟
unsteady convective acceleration pressure turbulent shear stresses
z-Momentum Equation
𝜕𝑤̄ 𝜕𝑤̄ 𝜕𝑤̄ 𝜕𝑤̄ 1 𝜕𝑝̄ 1 𝜕𝜏 𝜕𝜏 𝜕𝜏
⏟ + 𝑢̄ 𝜕𝑥 + 𝑣̄ 𝜕𝑦 + 𝑤̄ 𝜕𝑧 =⏟
𝑔 𝑐𝑜𝑠 𝛼 − + 𝜌 ( 𝜕𝑥𝑧𝑥 + 𝜕𝑦𝑧𝑦 + 𝜕𝑧𝑧𝑧 ) (2-15)
𝜕𝑡 ⏟ ⏟
𝜌 𝜕𝑧 ⏟
unsteady gravity
convective acceleration pressure turbulent shear stresses
Simplification of z-Momentum Equation

If the longitudinal length scale is much greater than the vertical length scale, then all vertical velocities <<
horizontal velocities. A result of this assumption is that vertical velocities are very small and vertical accel-
eration and shear stress terms in the z-momentum are negligible such that the z-momentum equation be-
comes the hydrostatic equation:
1 𝜕𝑝̄
𝜌 𝜕𝑧
= 𝑔 𝑐𝑜𝑠 𝛼 (2-16)
This assumption prevents the model from accurately modeling vertical accelerations of the fluid because of
convective cooling at night and other such vertical accelerations.
Lateral Averaging
The governing equations above will be laterally averaged after decomposing all velocities and pressure into
a lateral average and a deviation from the lateral average. The lateral, longitudinal, and vertical velocities
and pressure are defined as follows:
𝑣̅ = 𝑣̿ + 𝑣 ′′ (2-17)
𝑢̅ = 𝑢̿ + 𝑢′′ (2-18)
𝑤
̅ =𝑤
̿ + 𝑤" (2-19)
𝑝̅ = 𝑝̿ + 𝑝″ (2-20)

where:
1 𝑦
𝑢̄̄ = 𝐵 ∫𝑦 2 𝑢̄ 𝑑𝑦
1
B = control volume width, m
y1 = left bank coordinate
y2 = right bank coordinate
The double overbars represent the spatial average of the temporal average quantity. The double prime
represents the deviation from the lateral average and is a function of y. This is shown in Figure 8.
y=y2
y=y1
u y
Figure 8. Lateral average and deviation from lateral average components of longitudinal ve-
locity.
These definitions are substituted into the turbulent time-average governing equations and then laterally
averaged. The y-momentum equation is neglected since the average lateral velocities are zero (𝑣̄̄ = 0) and
cross shear stresses that contribute to vertical mixing will be computed from the analysis of wind stress.
The equations that remain are the continuity, x-momentum, and z-momentum equations.
Continuity Equation
The continuity equation becomes after substituting the above velocity components and laterally averaging
𝜕(𝑢̄̄ +𝑢″ ) 𝜕(𝑣̄̄ +𝑣 ″ ) ̄̄ +𝑤 ″ )

𝜕(𝑤
𝜕𝑥
+ 𝜕𝑦
+ 𝜕𝑧
=0 (2-21)
The lateral average of a double primed variable is by definition zero:
1 𝑦2
𝑢″ = 𝐵 ∫𝑦1 𝑢″ 𝑑𝑦 = 0 (2-22)

Also, note that:
𝑦2 𝑦2
𝜕(𝑣̄̄ +𝑣 ″ ) 1 𝑦2 𝜕(𝑣̄̄ +𝑣 ″ ) (𝑣̄̄ +𝑣 ″ ) 𝑣″
𝜕𝑦
= 𝐵 ∫𝑦1 𝜕𝑦
𝑑𝑦 = 𝐵
| = 𝐵| =𝑞 (2-23)
𝑦1 𝑦1
where q is defined as the net lateral inflow per unit volume of cell [T-1], and:
𝜕(𝑢̄̄ +𝑢″ ) 1 𝑦2 𝜕(𝑢̄̄ +𝑢″ ) 1 𝑦2 𝜕𝑢̄̄ 1 𝑦2 𝜕𝑢″ 1 𝜕 𝑦2 1 𝜕𝐵𝑢̄ ̄

𝜕𝑥
= 𝐵 ∫𝑦1 𝜕𝑥
𝑑𝑦 = 𝐵 ∫𝑦1 𝜕𝑥
𝑑𝑦 + 𝐵 ∫𝑦1 𝜕𝑥
𝑑𝑦 = 𝐵 𝜕𝑥 ∫𝑦1 𝑢̄̄ 𝑑𝑦 = 𝐵 𝜕𝑥 (2-24)
and:
̄̄ +𝑤 ″ )
𝜕(𝑤 1 ̄̄ +𝑤 ″ )
𝑦2 𝜕(𝑤 1 ̄̄
𝑦2 𝜕𝑤 1 𝑦2 𝜕𝑤 ″ 1 𝜕 𝑦2 1 𝜕𝐵𝑤̄ ̄
𝜕𝑧
= 𝐵 ∫𝑦1 𝜕𝑧
𝑑𝑦 = 𝐵 ∫𝑦1 𝜕𝑧
𝑑𝑦 + 𝐵 ∫𝑦1 𝜕𝑧
𝑑𝑦 = 𝐵 𝜕𝑧 ∫𝑦1 𝑤̄̄ 𝑑𝑦 = 𝐵 𝜕𝑧 (2-25)
Combining terms, the continuity equation becomes:
𝜕𝐵𝑢̄̄ 𝜕𝐵𝑤̄̄
𝜕𝑥
+ 𝜕𝑧
= 𝑞𝐵 (2-26)
x-Momentum Equation
The laterally-averaged x-momentum equation is more easily simplified by writing it in conservative form
(this can be verified by using the continuity equation with the x-momentum equation),
𝜕(𝑢̄̄ +𝑢″ ) 𝜕(𝑢̄̄ +𝑢″ )(𝑢̄̄ +𝑢″ ) 𝜕(𝑣̄̄ +𝑣 ″ )(𝑢̄̄ +𝑢″ ) ̄̄ +𝑤 ″ )(𝑢̄̄ +𝑢″ )
𝜕(𝑤 1 𝜕(𝑝̄̄ +𝑝″ )
𝜕𝑡
+ 𝜕𝑥
+ 𝜕𝑦
+ 𝜕𝑧
= 𝑔 𝑠𝑖𝑛 𝛼 − 𝜌 𝜕𝑥
+
1 𝜕𝜏𝑥𝑥 𝜕𝜏𝑥𝑦 𝜕𝜏𝑥𝑧
( + + ) (2-27)
𝜌 𝜕𝑥 𝜕𝑦 𝜕𝑧
Each term in this equation can be simplified as follows (note that the spatial average of any double primed
variable goes to zero by definition).
The unsteady acceleration term:
𝜕(𝑢̄̄ + 𝑢″ ) 1 𝑦2 𝜕(𝑢̄̄ + 𝑢″ )
= ∫ 𝑑𝑦
𝜕𝑡 𝐵 𝑦1 𝜕𝑡
1 𝑦2 𝜕𝑢̄̄ 1 𝑦2 𝜕𝑢″
= ∫ 𝑑𝑦 + ∫ 𝑑𝑦
𝐵 𝑦1 𝜕𝑡 𝐵 𝑦1 𝜕𝑡
1 𝜕 𝑦2 1 𝜕 𝑦2 1 𝜕𝐵𝑢̄ ̄
= 𝐵 𝜕𝑡 ∫𝑦1 𝑢̄̄ 𝑑𝑦 + 𝐵 𝜕𝑡 ∫𝑦1 𝑢″ 𝑑𝑦 = 𝐵 𝜕𝑡 (2-28)
The convective acceleration terms:
𝜕(𝑢̄̄ + 𝑢″ )(𝑢̄̄ + 𝑢″ ) 1 𝑦2 𝜕(𝑢̄̄ + 𝑢″ )(𝑢̄̄ + 𝑢″ )

= ∫ 𝑑𝑦
𝜕𝑥 𝐵 𝑦1 𝜕𝑥
1 𝑦2 𝜕𝑢̄̄ 𝑢̄̄ 1 𝑦2 2𝜕𝑢̄̄ 𝑢″ 1 𝑦2 𝜕𝑢″ 𝑢″
= ∫ 𝑑𝑦 + ∫ 𝑑𝑦 + ∫ 𝑑𝑦
𝐵 𝑦1 𝜕𝑥 𝐵 𝑦1 𝜕𝑥 𝐵 𝑦1 𝜕𝑥

1 𝜕 𝑦2 1 𝜕 𝑦2 ″ ″
= ∫ 𝑢̄̄ 𝑢̄̄ 𝑑𝑦 + ∫ 𝑢 𝑢 𝑑𝑦
𝐵 𝜕𝑥 𝑦1 𝐵 𝜕𝑥 𝑦1
1 𝜕𝐵𝑢̄̄ 𝑢̄̄ 1 𝜕 𝑦2
=𝐵 𝜕𝑥
+⏟ ∫ 𝑢″ 𝑢″ 𝑑𝑦
𝐵 𝜕𝑥 𝑦1
(2-29)
dispersion term
Similarly for the other two convective acceleration terms:
𝜕(𝑢̄̄ +𝑢″ )(𝑤

̄̄ +𝑤 ″ ) 1 𝜕𝐵𝑢̄̄ 𝑤
̄̄ 1 𝜕 𝑦2
𝜕𝑧
=𝐵 𝜕𝑧
+⏟ ∫ 𝑢″ 𝑤 ″ 𝑑𝑦
𝐵 𝜕𝑧 𝑦1
(2-30)
dispersion term
𝜕(𝑢̄̄ +𝑢″ )(𝑣̄̄ +𝑣 ″ )

= 𝑢″ 𝑣 ″ |𝑦2 − 𝑢″ 𝑣 ″ |𝑦1 = 0 (2-31)
𝜕𝑦
The gravity term:
1 𝑦2 1 𝑦2
𝑔 𝑠𝑖𝑛 𝛼 = 𝐵 ∫𝑦1 𝑔 𝑠𝑖𝑛 𝛼 𝑑𝑦 = 𝐵 (𝑔 𝑠𝑖𝑛 𝛼) ∫𝑦1 𝑑𝑦 = 𝑔 𝑠𝑖𝑛 𝛼 (2-32)
The pressure gradient term:
𝜕(𝑝̄̄ + 𝑝″ ) 1 𝑦2 𝜕(𝑝̄̄ + 𝑝″ )
= ∫ 𝑑𝑦
1 𝑦2 𝜕𝑝̄̄ 1 𝑦2 𝜕𝑝″
= ∫ 𝑑𝑦 + ∫ 𝑑𝑦
𝐵 𝑦1 𝜕𝑥 𝐵 𝑦1 𝜕𝑥
1 𝜕 𝑦2 1 𝜕 𝑦2 ″ 1 𝜕𝐵𝑝̄̄
= ∫ 𝑝̄̄ 𝑑𝑦
𝐵 𝜕𝑥 𝑦1
+ ∫ 𝑝 𝑑𝑦
𝐵 𝜕𝑥 𝑦1
=
𝐵 𝜕𝑥
(2-33)
or the above equation can be written, assuming that the derivative of the lateral average pressure gradient
in the x-direction is not a function of y:
𝜕(𝑝̄̄ + 𝑝″ ) 1 𝑦2 𝜕(𝑝̄̄ + 𝑝″ )
= ∫ 𝑑𝑦
1 𝜕𝑝̄̄ 𝑦2 1 𝑦2 𝜕𝑝″
= ∫ 𝑑𝑦 + ∫ 𝑑𝑦
𝐵 𝜕𝑥 𝑦1 𝐵 𝑦1 𝜕𝑥
1 𝜕𝑝̄̄ 1 𝜕 𝑦2 𝜕𝑝̄̄
= 𝐵 𝜕𝑥 𝐵 + 𝐵 𝜕𝑥 ∫𝑦1 𝑝″ 𝑑𝑦 = 𝜕𝑥 (2-34)
The shear stress terms:
𝜕𝜏𝑥𝑥 𝜕𝜏𝑥𝑦𝜕𝜏 1 𝑦2 𝜕𝜏 1 𝑦2 𝜕𝜏 1 𝑦2 𝜕𝜏
( 𝜕𝑥
+ 𝜕𝑦
+ 𝜕𝑧𝑥𝑧 ) = 𝐵 ∫𝑦1 𝜕𝑥𝑥𝑥 𝑑𝑦 + 𝐵 ∫𝑦1 𝜕𝑦𝑥𝑦 𝑑𝑦 + 𝐵 ∫𝑦1 𝜕𝑧𝑥𝑧 𝑑𝑦
1 𝜕 𝑦2 1 𝜕 𝑦2 1 𝜕 𝑦2 1 𝜕𝐵𝜏 𝜕𝐵𝜏𝑥𝑦 𝜕𝐵𝜏
= 𝐵 𝜕𝑥 ∫𝑦1 𝜏𝑥𝑥 𝑑𝑦 + ∫ 𝜏 𝑑𝑦 + 𝐵 𝜕𝑧 ∫𝑦1 𝜏𝑥𝑧 𝑑𝑦 = 𝐵 ( 𝜕𝑥𝑥𝑥 + 𝜕𝑦 + 𝜕𝑧𝑥𝑧 ) =
𝐵 𝜕𝑥 𝑦1 𝑥𝑦
1 𝜕𝐵𝜏𝑥𝑥 𝜕𝐵𝜏𝑥𝑧
𝐵
( 𝜕𝑥 + 𝜕𝑧
) (2-35)

Collecting all terms and neglecting all dispersion terms, the final x-momentum equation is:
𝜕𝐵𝑢̄̄ 𝜕𝐵𝑢̄̄ 𝑢̄̄ 𝜕𝐵𝑢̄̄ 𝑤̄ 𝐵 𝜕𝑝̄̄ 1 𝜕𝐵𝜏𝑥𝑥 𝜕𝐵𝜏𝑥𝑧

+ + = 𝐵𝑔 𝑠𝑖𝑛 𝛼 − + ( + ) (2-36)
𝜕𝑡 𝜕𝑥 𝜕𝑧 𝜌 𝜕𝑥 𝜌 𝜕𝑥 𝜕𝑧
Summary of Laterally Averaged Equations

In the development of CE-QUAL-W2 in Cole and Buchak (1995), the lateral average terms were represented
by uppercase characters, such that 𝑢̄̄ = 𝑈, 𝑤̄̄ = 𝑊, and 𝑝̄̄ = 𝑃. The shear stress terms will be assumed
lateral averages and the double overbars will be dropped for convenience. Making these simplifications,
the governing equations become:
Continuity Equation
 UB  WB
+ = qB (2-37)
x z
x-Momentum Equation
 UB  UUB  WUB B  P 1 B xx 1  B  x z
+ + = gB sin  - + + (2-38)
t x z  x  x  z
z-Momentum Equation
1 𝜕𝑃
= 𝑔 𝑐𝑜𝑠 𝛼 (2-39)
𝜌 𝜕𝑧
Knowing the density through the equation of state, there are now three equations and three unknowns -
U, W, and P.
Simplification of Pressure Term

The z-momentum equation reduces to:
𝑧
𝑃 = 𝑃𝑎 + 𝑔 𝑐𝑜𝑠 𝛼 ∫𝜂 𝜌𝑑𝑧 (2-40)
after integration from a depth z to the water surface defined as z=. Pa is the atmospheric pressure at the
water surface (Figure 9).

g
Pa z=zsurface=
z x
z=h=zbottom
Figure 9. Schematization for simplification of pressure term.
This equation for pressure is now substituted into the x-momentum equation and simplified using Leibnitz
rule. The pressure gradient term in the x-momentum equation then becomes:
1 𝜕𝑃 1 𝜕𝑃𝑎 𝜕𝜂 𝑔 𝑐𝑜𝑠 𝛼 𝑧 𝜕𝜌
− =− + 𝑔 𝑐𝑜𝑠 𝛼 − ∫𝜂 𝑑𝑧 (2-41)
𝜌 𝜕𝑥 𝜌 𝜕𝑥 𝜕𝑥 𝜌 𝜕𝑥
The first term on the RHS is the atmospheric pressure term (accelerations due to atmospheric pressure
changes over the water surface), the second is the barotropic pressure term (accelerations due to water
surface variations), and the third is the baroclinic pressure term (accelerations due to density driven cur-
rents).
Assuming that atmospheric pressure does not vary spatially, the atmospheric pressure gradient term is ne-
glected. This implies that for long water bodies during severe storms the model will not be able to account
for accelerations because of atmospheric pressure changes in x. The pressure term then becomes:
1 𝜕𝑃 𝜕𝜂 𝑔 𝑐𝑜𝑠 𝛼 𝑧 𝜕𝜌
− = 𝑔 𝑐𝑜𝑠 𝛼 − ∫𝜂 𝑑𝑧 (2-42)
𝜌 𝜕𝑥 𝜕𝑥 𝜌 𝜕𝑥
The revised form of the x-momentum equation is then:
 UB  UUB  WUB  g cos B  z
t
+
x
+
z
= gB sin  + g cos B
x
−
  x dz

(2-43)
1 B xx 1  B  x z
+ +
 x  z
Effectively, pressure has been removed from the unknowns by combining the z-momentum and x-momen-
tum equations, but  has been added as an unknown.

Free Water Surface
This equation is a simplification of the continuity equation. The continuity equation integrated over the
depth from the water surface to the bottom is called the free water surface equation. Figure 10 and Figure
11 are definition sketches for the computational grid without and with a channel slope, respectively.
CE-QUAL-W2 coordinate system,  = 0 g
z=0

z=zsurface=
h-
z x
z
x
z=h
Figure 10. Coordinate system without channel slope.

CE-QUAL-W2 coordinate system,  > 0
z=0
g

z=zsurface=
h-
z
z=h x
z
x

Figure 11. Coordinate system with channel slope.
The continuity equation is integrated over the depth as follows:
 UB  WB
h h h
  x dz +   z dz =  qBdz (2-44)
The first term can be expanded as follows using Leibnitz’s rule:
 UB  h 
h h
  x dz = x  UBdz − x UB h + x UB  (2-45)
The integral of the vertical flow rate over z relates to changes in water surface elevation as shown below:
 WB
h


z
dz = WB h − WB  (2-46)

where:
𝜕ℎ 𝜕ℎ
𝑊ℎ = + 𝑈ℎ
𝜕𝑡 𝜕𝑥
𝜕𝜂 𝜕𝜂
𝑊𝜂 = + 𝑈𝜂
𝜕𝑡 𝜕𝑥
Combining these terms together, the free surface equation becomes:
ℎ 𝜕 ℎ 𝜕ℎ 𝜕𝜂 𝜕ℎ 𝜕ℎ 𝜕𝜂 𝜕𝜂
∫𝜂 𝑞𝐵𝑑𝑧 = 𝜕𝑥 ∫𝜂 𝑈𝐵𝑑𝑧 − 𝜕𝑥 𝑈𝐵|ℎ + 𝜕𝑥 𝑈𝐵|𝜂 + 𝑈ℎ 𝐵ℎ 𝜕𝑡 + 𝑈ℎ 𝐵ℎ 𝜕𝑥 − 𝐵𝜂 𝜕𝑡 − 𝐵𝜂 𝑈𝜂 𝜕𝑥
(2-47)
Canceling out terms and applying the no-slip boundary condition that Uh is zero:
𝜕 ℎ 𝜕𝜂 ℎ
∫ 𝑈𝐵𝑑𝑧
𝜕𝑥 𝜂
− 𝐵𝜂
𝜕𝑡
= ∫𝜂 𝑞𝐵𝑑𝑧 (2-48)
or
𝜕𝜂 𝜕 ℎ ℎ
𝐵𝜂 = ∫ 𝑈𝐵𝑑𝑧 − ∫𝜂 𝑞𝐵𝑑𝑧 (2-49)
𝜕𝑡 𝜕𝑥 𝜂
where B is the width at the surface.
Equation of State
The density must be known for the solution of the momentum equations. The equation of state is an equa-
tion that relates density to temperature and concentration of dissolved substances. This equation is given
by:
 = f( T w ,TDS , ISS ) (2-50)
where f(Tw,TDS, ISS) is a density function dependent upon water temperature (Tw), concentration of total
dissolved solids or salinity (TDS), and concentration of inorganic suspended solids (ISS).
Summary of Governing Equations

Table 1 shows the governing equations after lateral averaging for a channel slope of zero and for an arbi-
trary channel slope. Parameters used in Table 1 are illustrated in Figure 12.

Table 1. Governing equations with and without channel slope.
Governing equation assuming no channel Governing equation assuming an arbitrary chan-
Equation slope and no momentum conservation at nel slope and conservation of momentum at
branch intersections branch intersections
 UB  UUB  WUB  UB  UUB  WUB
+ + = + + = gB sin 
t x z t x z
 gB z   g cos B 
z
x   x  x dz +
x- momentum gB − dz + + g cos B −
x  
1 B xx 1  B  x z 1 B xx 1  B  x z
+ + + qBU x
 x  z  x  z
1 P 1 P
z-momentum 0=g− 0 = g cos  −
 z  z
 UB  WB  UB  WB
continuity + = qB + = qB
x z x z
state  = f( T w ,TDS ,ss )  = f( T w ,TDS ,ss )
  h h
 
h h
t x   qBdz t x 
free surface B = UBdz − B = UBdz −  qBdz

U = horizontal velocity, m s-1 x = x-direction lateral average shear stress

W = vertical velocity, m s-1 y = y-direction lateral average shear stress
B = channel width  = density
P = pressure  = water surface
h gravity
z
channel slope = So = tan

datum
x
Figure 12. Definition sketch for channel slope.

Linkage of Mainstem Branches with Internal Head
Boundary Conditions
Since the CE-QUAL-W2 model can represent neighboring waterbodies with different vertical grids, the ver-
tical grids must be linked together to transfer volume, mass, and heat accurately.
In order for the volume, heat, and mass to be passed from one cell to another, the flow, temperature, and
concentration from the downstream segment [ID] of the upstream branch are transferred to the upstream
segment [IU] of the downstream branch. Spatial averaging to conserve flow, heat, and mass is illustrated in
Figure 13.
Branch 1
Branch 2
Figure 13. Transfer of mass and momentum between branches with unequal vertical grid spacing.
Linkage of Tributary or Side Branches

Tributary or side branches are not defined as “tributaries” in the model. “Tributaries” are inflows into a
branch, whereas side or tributary branches are modeled as part of the model domain. Version 2 of CE-
QUAL-W2 assumed all tributary branches came in at right angles to the main channel resulting in no longi-
tudinal momentum exchange between the branches. Version 3 and later versions have included momen-
tum transfer between the side branch and the main branch based on the segment orientation (Figure 14).

i-1
i x
I+
1
x U
Uy
Ux
trib
Figure 14. Linkage of tributary or side branch coming in at an angle to main branch.
The tributary branch inflow can create shear stress along both the longitudinal axis of the main stem branch
and along the y-axis of the segment. The cross-shear mixing has been added to the cross-shear wind stress
for the computation involving the vertical eddy viscosity and vertical diffusivity. This involves determining
the y and x velocity components of the entering branch (Figure 15).
Longitudinal Momentum
The vector component of velocity in the x-direction of the main channel, Ux, can be computed by analysis
of the channel orientations. This component in the x-direction would be:
Ux=Ucos (2-51)
where:
U = longitudinal velocity of the tributary at segment ID for the tributary branch
 = difference in the angle between the main stem and tributary segments (Figure 15).

trib =main-trib
main


U
Uy

Ux
Figure 15. Schematic of x and y velocity components where Θmain is the angle of the main branch seg-
ment and Θtrib is the angle of the tributary segment. Ux is the x-velocity component of the tributary
branch into the main segment. Uy is the y-velocity component of the tributary branch into the main seg-
ment.
The conservation of momentum about a control volume, the main stem segment, would result in an addi-
tional source of momentum. Lai (1986) shows that the correction to the x-momentum equation would be:
𝑞𝐵𝑈𝑥 (2-52)
where:
q = lateral inflow per unit length.
This arises from re-deriving the momentum equations and assuming that all the flow entering the segment
is moving at the velocity Ux. This correction to the x-momentum equation would be
 UB  UUB  WUB  g cos B 

z
t
+
x
+
z
x
−
  x dz
(2-53)
1 B xx 1  B  x z
+ + + qBU x
 x  z 
momentum from
side tributaries
Cross-shear of Tributary Inflow

Even though CE-QUAL-W2 assumes lateral averaging, the y-velocity coming into a reservoir may also con-
tribute to vertical mixing. The y component of a tributary inflow is Uy = Usin (Figure 15). Since there is no

y-momentum equation, this lateral shear contributes to the vertical shear stress. This is similar to how the
model also accounts for cross-shear stress from the wind given as  wy  C D  a Wh 2 sin(  1 −  2 )
where this cross-shear also contributes to the vertical eddy viscosity.
For lateral inflow, the additional shear stress was parameterized as an interfacial shear:
𝑓
𝜏𝑦𝑡𝑟𝑖𝑏 ≅ 𝜌 8 𝑈𝑦2 (2-54)
where:
f = interfacial friction factor
For two-layer flow systems, f has been found to be of order 0.01. Therefore, the vertical eddy viscosity
would be increased by a significant lateral tributary inflow.
Computation of Initial Water Surface Slope and Ve-

locity Field for River Simulation
The initial conditions for velocity are usually set to zero, and the initial water surface elevations are set in
the bathymetry input file. This is appropriate for a reservoir or lake or estuary system, but for a sloping
river section setting initial water surface elevations and velocities can be computed internally in the
model. The model user can turn this feature on for river systems. This feature allows for a much smoother
running of the river model initially and avoids having to compute the initial water surface elevation before
a model simulation. Initial water levels and horizontal velocities are computed using Manning’s normal
depth equation. The model bathymetry and flow files are used to determine the normal depth and hori-
zontal velocities at the beginning of a model simulation.
The flow through each model segment is estimated first by considering all inflows and outflows such as
tributaries, withdrawals, distributed tributaries, upstream branch inflows, flows associated with internal
head boundaries, and dam outflows. Once the flow through each segment has been estimated, the nor-
mal depth for each segment is calculated using Manning’s equation:
1 2 1
𝑄𝑖 = 𝐴𝑖 𝑅𝑖 ⁄3 𝑆 ⁄2
𝑛𝑖
where:
𝑄𝑖 : Flow through segment i
𝑛𝑖 : Manning’s friction coefficient for segment i
𝐴𝑖 : Cross-sectional area of segment i
𝑅𝑖 : Hydraulic Radius of segment i
𝑆: Branch slope
The cross-sectional area 𝐴𝑖 and hydraulic radius 𝑅𝑖 are a function of depth. The hydraulic radius is calcu-
lated using
𝐴𝑖
𝑅𝑖 =
𝑃𝑖
𝑃𝑖 is the wetted perimeter. To calculate the normal depth, the root to the following function is found:

1 2 1
𝑄𝑖 − 𝐴 𝑅 ⁄3 𝑆 ⁄2 = 0
𝑛𝑖 𝑖 𝑖
The method of bisection (Press et al., 1992) is used to find the root, or normal depth, because the func-
tion is discontinuous (has corners).
Once the normal depth for every segment has been calculated, the water surface of each sloping branch
is smoothed. If the predicted water level of a segment is less than the water level of a downstream seg-
ment, the downstream segment is considered controlling and the water level is increased to match that of
the downstream segment. The effect of spillways and gates is considered by calculating the head neces-
sary to convey the flow of the segment on the upstream side of the gate or spillway. If the normal depth
predicted water level of segments upstream of a gate or spillway are less than the necessary water level
at the gate or spillway, water levels of upstream segments are set to that of the segment immediately up-
stream of the gate or spillway.
Once water levels have been smoothed and gates or spillways have been accounted for, the average hori-
zontal velocity 𝑈𝑎𝑣𝑔 in each segment is estimated with
𝑄𝑖
𝑈𝑎𝑣𝑔 =
𝐴𝑖
If a branch is a “loop branch”, or a branch with upstream and downstream ends that are internal head
boundaries attached to segments of another, single branch, the initial velocity water level tool will set ini-
tial velocities of this branch equal to zero and estimate water levels by interpolating between the water
levels in upstream and downstream boundary condition segments.
Auxiliary Functions
Auxiliary functions are relationships that describe processes independent of basic hydrodynamic and
transport computational schemes in the model. Auxiliary functions include turbulent dispersion and wind
shear processes, heat exchange (including ice cover), evaporation, hydraulic structures, density, and selec-
tive withdrawal.
Surface Shear Stress

Referring to Figure 16, the shear stress at the water surface is defined as
 s = C D  a (Wh − u s )2  C D  a (Wh )2 (2-55)
where:
s = surface shear stress at water surface, kg/m/s2
us = surface velocity in water, m/s
Wh = wind velocity measured at a distance h above water surface in direction of shear, m/s
CD = drag coefficient, [-]
a = air density, kg/m3

W(z)
Wh
us
x
z
Figure 16. Shear stress at the air-water surface.
Note that this relationship leads to the “3% rule” for surface currents:
 s = C D  a (Wh − u s )2 = C D W u s2
 


air water
(2-56)
if C D ~ CD , then u s ~ 0.03Wh
air water 
3% rule
Usually the drag coefficient is a function of the measurement height, h, above the water surface. Most drag
coefficient formulae have been determined based on a 10 m wind speed measurement height. If wind
speeds are taken at other measurement heights for the shear stress calculation, these are corrected in the
model to a 10 m height.
The wind speed is a function of measurement height. To correct the measurement height to an elevation
z, the following approach is used assuming a logarithmic boundary layer:
z
ln( )
Wz z 0
= (2-57)
W z 1 ln( z 1 )
z0
where:
Wz = desired wind speed at elevation z
Wz1 = known wind speed at height z1
z0 = wind roughness height (some have used 0.003 ft or 9.14E-4 m for wind < 5 mph or 2.24 m/s and
0.015 ft or 0.0046 m for wind > 5 mph or 2.24 m/s; the range of values are between 0.0005 to
0.03 ft or 1.5E-4 and 0.009 m; this is a user-defined coefficient in the CE-QUAL-W2 model)

This term can then be used to compute the surface stress in the direction of the x-axis and the cross-shear
(the cross-shear term is used in the turbulent shear stress algorithm) as follows:
 wx  C D  a Wh 2 cos(  1 −  2 ) (2-58)
 wy  C D  a Wh 2 sin(  1 −  2 ) (2-59)
where:
wx = surface shear stress along x-axis due to wind
wy = surface shear stress along lateral direction due to wind
1 = wind orientation relative to North, radians
2 = segment orientation relative to North, radians
Segment and wind orientation are illustrated in Figure 17 and Figure 18, respectively. The angles are meas-
ured in radians clockwise from north. A wind from the east would have an angle of π/2 radians, π radians
from the south, 3π/2 radians from the west, and 0 or 2π radians from the north.
Figure 17. Segment orientation.

North
Wh
N to S,  = 0
E to W,  = 2
Figure 18. Wind orientation.
The CE-QUAL-W2 model Version 3.5 and earlier included a drag coefficient, CD, that was based on the
following formulae:
𝐶𝐷 = 0.0 𝑖𝑓 𝑊10 < 1 𝑚/𝑠
0.5
𝐶𝐷 = 0.005𝑊10 𝑖𝑓 1 𝑚/𝑠 ≤ 𝑊10 < 15 𝑚/𝑠
𝐶𝐷 = 0.0026 𝑖𝑓 𝑊10 ≥ 15 𝑚/𝑠
where W10 is the wind speed in m/s at 10 m.
This formulation shows that CD decreases as wind speed decreases. It has been observed though that
for wind speeds below 5 m/s, CD increases with lowering wind speed. According to Wuest and Lorke
(2003), the drag coefficient for a smooth surface (below 5 m/s) follows the following equation:
−1.15
𝐶𝐷 = 0.0044𝑊10
Above a wind speed of 5 m/s, CD can be calculated from the following implicit equation:
−2
10𝑔
𝐶𝐷 = {𝑘 −1 𝑙𝑛 [ 2 ] + 11.3}
𝐶𝐷𝑊10
where k is von Karman constant (=0.41), and the ‘10’ in the above equation has units of m. The above
equation for winds greater than 5 m/s are thought to be a lower limit to the drag coefficient since the
wind-wave effects at higher wind speeds probably increase the value of CD higher than that predicted
by the above implicit equation. The difference between these 2 equations is shown below in Figure 19.
Since the W2 approximation for wind speeds above 4 m/s is reasonable, for Version 3.6 and later ver-
sions the following drag coefficient formulation is used:

𝐶𝐷 = 0.01 𝑖𝑓 𝑊10 < 0.5 𝑚/𝑠
−1.15
𝐶𝐷 = 0.0044𝑊10 𝑖𝑓 𝑊10 < 4 𝑚/𝑠
0.5
𝐶𝐷 = 0.005𝑊10 𝑖𝑓 4 𝑚/𝑠 ≤ 𝑊10 < 15 𝑚/𝑠
𝐶𝐷 = 0.0026 𝑖𝑓 𝑊10 ≥ 15 𝑚/𝑠
This is shown as the dotted line in Figure 19.
This will have the effect of increasing the drag coefficient at low wind speeds. The impact of this on the
2
wind shear stress at the surface, 𝜏 = 𝜌𝑎 𝐶𝐷𝑊10 , is shown in Figure 20.
V3.5 and earlier W2 model

Wuest and Lorke (2003)
Revised W2 drag coefficient
0.01
Drag coefficient, CD [-]
0.001
0 4 8 12 16 20
Wind speed at 10 m, m/s
Figure 19. Comparison of current W2 model computation of CD and that recommended as a lower
limit by Wuest and Lorke (2003).

10
V3.5 and earlier W2 model
Revised W2 drag coefficient
1
Surface shear stress, kg/m/s2
0.1
0.01
0.001
0.0001
0 4 8 12 16 20
Wind speed at 10 m, m/s
Figure 20. Surface shear stress for CE-QUAL-W2 Version 3.6 and later compared to Version 3.5 and
earlier.
Wind Fetch Calculation

The wind fetch is the distance along a lake or reservoir where the wind blows unobstructed. This can con-
tribute to enhanced wind-wave generation as the fetch distance is increased. The vertical shear stress can
be enhanced as fetch increases.
The CE-QUAL-W2 model always calculates fetch distance for the vertical shear stress algorithm and for re-
aeration calculations. The calculation technique is a simplified one that is based on the prevailing wind
direction. The fetch at a segment is the sum of the lengths of the segments either upstream or downstream
based on the wind direction. For example, in Figure 21, the fetch distance of segment 5 (computed at the
∆𝑥
segment center) would be computed from the following: 𝐹𝑒𝑡𝑐ℎ = ∆𝑥2 + ∆𝑥3 + ∆𝑥4 + 5 . This is based on
2
the prevailing wind direction.
A more complex fetch correction has been reviewed by Henry (2008) and will be part of an updated algo-
rithm.
A fetch correction to the wind velocity can be used as determined by Fang and Stefan (1994). This correc-
tion is described under Dissolved Oxygen in the lake/reservoir reaeration section, but is not applicable to
rivers.

Figure 21. Illustration of fetch distance calculation.
Bottom Shear Stress

The shear stress is defined along the bottom of each cell (or for each cell in contact with side walls or chan-
nel bottom) as
𝜌𝑤 𝑔
𝜏𝑏 = 𝐶2
𝑈|𝑈| (2-60)
where:
C = Chezy friction coefficient
U = longitudinal velocity
w = density of water
Also, the model user can specify a Manning’s friction factor, n, where the Chezy coefficient, C, is related to
the Manning’s friction factor (in SI units) as
C = (1/n)R1/6 (2-61)
where:
n = Manning’s friction factor
R = hydraulic radius
In Version 2, the bottom shear stress was applied only to the bottom of each layer. In Version 3 and later,
the sidewall friction is accounted for because of its greater importance in river systems. The user can input
either the Chezy or Manning’s coefficient for each model segment.

Vertical Shear Stress
The vertical shear stress is defined as
𝜏𝑥𝑧 𝜕𝑈 𝜕𝑈
= 𝜈𝑡 = 𝐴𝑧 (2-62)
𝜌 𝜕𝑧 𝜕𝑧
where:
Az = vertical eddy viscosity, m2/s
t, = vertical eddy viscosity, m2/s
U = longitudinal velocity, m/s
 = density of water, kg/m3
xz = vertical shear stress, kg/m/s2
In Version 3 and later, the user must specify which formulation to use for Az or t, the vertical eddy viscosity.
The formulations are shown in Table 2.
Table 2. Vertical eddy viscosity formulations.

Formulation Formula Reference
𝜕𝑢
𝜈𝑡 = ℓ2𝑚 | | 𝑒 −𝐶𝑅𝑖
𝜕𝑧
Nickuradse (NICK) 𝑧 2 𝑧 4 Rodi (1993)
ℓ𝑚 = 𝐻 [0.14 − 0.08 (1 − ) − 0.06 (1 − ) ]
𝐻 𝐻
𝑧
Parabolic (PARAB) 𝜈𝑡 = 𝜅𝑢∗ 𝑧 (1 − ) 𝑒 −𝐶𝑅𝑖 Engelund (1978)
𝐻
2 1⁄2
W2 (used in Version 2)
2
𝑙𝑚 𝜕𝑈 2 𝜏𝑦,𝑤𝑖𝑛𝑑 𝑒 −2𝑘𝑧 + 𝜏𝑡𝑟𝑖𝑏𝑢𝑡𝑎𝑟𝑦 Cole and Buchak
𝜈𝑡 = 𝜅 [( ) + ( ) ] 𝑒 −𝐶𝑅𝑖 (1995)
2 𝜕𝑧 𝜌𝜈𝑡
ℓ𝑚 = 𝛥𝑧𝑚𝑎𝑥
2 1⁄2
W2 with mixing length
2
𝑙𝑚 𝜕𝑈 2 𝜏𝑦,𝑤𝑖𝑛𝑑 𝑒 −2𝑘𝑧 + 𝜏𝑡𝑟𝑖𝑏𝑢𝑡𝑎𝑟𝑦 Cole and Buchak
𝜈𝑡 = 𝜅 [( ) + ( ) ] 𝑒 −𝐶𝑅𝑖 (1995) and Rodi
of Nickuradse (W2N) 2 𝜕𝑧 𝜌𝜈𝑡 (1993)
𝑧 2 𝑧 4
ℓ𝑚 = 𝐻 [0.14 − 0.08 (1 − ) − 0.06 (1 − ) ]
𝐻 𝐻
RNG (re-normaliza-tion 3 1/3
zu 𝑧 3
𝜈𝑡 = 𝜈 [1 + 𝛹 (3𝜅 ( ∗) (1 − ) − 𝐶1 )] 𝑒 −𝐶𝑅𝑖 Simoes (1998)
group) 𝜈 𝐻
TKE (Turbulent kinetic 𝑘2 Wells (2001),
energy) 𝜈𝑡 = 𝐶𝜇 𝜀
where k and  are defined from Gould (2006)
𝜕𝑘𝐵 𝜕𝑘𝐵𝑈 𝜕𝑘𝐵𝑊 𝜕 𝜈𝑡 𝜕𝑘
+ + − (𝐵 )
𝜕𝑡 𝜕𝑥 𝜕𝑧 𝜕𝑧 𝜎𝑘 𝜕𝑧
𝜕 𝜈𝑡 𝜕𝑘
− (𝐵 ) = 𝐵(𝑃 + 𝐺 − 𝜀 + 𝑃𝑘 )
𝜕𝑥 𝜎𝑘 𝜕𝑥
𝜕𝜀𝐵 𝜕𝜀𝐵𝑈 𝜕𝜀𝐵𝑊 𝜕 𝜈𝑡 𝜕𝜀

+ + − (𝐵 )
𝜕𝑡 𝜕𝑥 𝜕𝑧 𝜕𝑧 𝜎𝜀 𝜕𝑧
𝜕 𝜈𝑡 𝜕𝜀 𝜀 𝜀2
− (𝐵 ) = 𝐵 (𝐶𝜀1 𝑃 + 𝐶𝜀2 + 𝑃𝜀 )
𝜕𝑥 𝜎𝜀 𝜕𝑥 𝑘 𝑘

Formulation Formula Reference
where:
m = mixing length C = constant (assumed 0.15) k = wave number (in W2 models)

z = vertical coordiante u* = shear velocity  = liquid density
H = depth  = von Karman constant (x) = max(0,x)
u = horizontal velocity wy = cross-shear from wind  = molecular viscosity
Ri = Richardson number zmax = maximum vertical grid spacing C1 = empirical constant, 100
B = width t = turbulent viscosity k = turbulent kinetic energy (in TKE model)
 = turbulent energy dissipation rate P = turbulent energy production from boundary friction
U = longitudinal velocity (laterally averaged) W = vertical velocity (laterally averaged)  = turbulent Prandlt number
𝜕𝑈 2 𝜈 𝑔 𝑑𝜌
Production term:𝑃 = 𝜈𝑡 [( ) ] Buoyancy term:𝐺 = − 𝑡 𝑁 2 Brunt–Vaisala frequency 𝑁 = √−
𝜕𝑧 𝜎 𝑡 𝜌 𝑑𝑧
C, Cµ= constants in the TKE model ytrib = cross-shear from lateral tributaries
The model user can also specify the maximum value of the vertical eddy viscosity [AZMAX]. This value is
specified because the time step for numerical stability is greatly reduced when solving the momentum
equations using an explicit numerical technique. In addition, the model user can choose whether to com-
pute the vertical momentum transfer with the longitudinal momentum equation using an implicit or an
explicit numerical technique. The explicit formulation was used in Version 2 with a fixed [AZMAX] of 1.0 x
10-5 m2 s-1. The implicit solution code was originally developed by Chapman and Cole and revised/updated.
Since the introduction of the k turbulence closure model, this formulation works well for all waterbody
types and is the suggested default model algorithm. Note that only the W2 and W2N include the effects of
cross-shear from wind explicitly and from tributary or branch inflows. The k turbulence model though ac-
counts for total turbulent kinetic energy production from wind, not just the x-component. The W2 or W2N
can be used for waterbodies with deep sections that could be stratified. The NICK, PARAB, or RNG could be
used for estuary or river systems where the maximum computed [AZMAX] could be as high as 1 to 5 m2 s-
1
. For the river model, the model user should use the implicit solution technique. To reproduce results from
Version 2 in a stratified reservoir, set [AZMAX] to 1.0 x 10-5 m2 s-1 and use the explicit solution.
How does one know which turbulent closure scheme to use for xz since, according to Hamblin and Salmon
(1975), "the vertical diffusion of momentum is probably the most important internal parameter" for pre-
dicting internal circulation patterns? Because of the disarray in the literature over which formulation is
best, Shanahan (1980) suggested that we "use theory and literature as a guide to develop alternative vis-
cosity functions and then test those functions in calibration runs against field data." In the absence of
expensive-to-obtain current velocity data, the use of temperature profiles is often used to test the adequacy
of the hydrodynamic regime against different formulations.
Typical variations of these formulations as predicted by the model are shown in Figure 22 for Manning's
fiction factor and in Figure 23 for a Chezy friction factor. Comparison of the various turbulence closure
theories to classical open channel flow theory for seven vertical layers is shown in Figure 24.

Manning's n=0.03 and S=0.0001
86
84
82
PARAB
Elevation, m
80
RNG
78 W2
W2N
76
NICK
74
72
70
0 0.05 0.1 0.15 0.2 0.25
Turbulent eddy viscosity, m2 s-1
Figure 22. Variation of turbulent vertical eddy viscosity for flow of 2574 m3 s-1 flow down a channel of
length 30 km and width of 100 m at x=15 km.
Chezy=50 and S=0.0001
86
84
82
PARAB
Elevation, m
80
RNG
78 W2
W2N
76
NICK
74
72
70
0 0.05 0.1 0.15 0.2 0.25
Turbulent eddy viscosity, m2s-1
Figure 23. Variation of turbulent vertical eddy viscosity for flow of 2574 m3 s-1 flow down a channel of
length 30 km and width of 100 m measured at x=15 km.

7 vertical layers, Manning's n=0.03 and S=0.0001
90
88
86
PARAB
84
Elevation, m
RNG
82
W2
80
W2N
78
NICK
76
Theory
74
72
70
0 0.5 1 1.5 2 2.5
Velocity, m s-1
Figure 24. Comparison of vertical velocity predictions of W2 model with various eddy viscosity models
compared to theory.
Formulation
In CE-QUAL-W2, the shear stress term includes the contribution to the shear stress from surface waves
induced by the wind. The wind can produce waves that produce decaying motions with depth (Figure 25).
wind stress
Figure 25. Conceptual diagram of wind induced motion.
The total longitudinal shear stress for a layer is defined as having contributions from interfacial
velocity shear, wind wave generated shear, and friction shear along boundaries:
𝜏𝑥𝑧 𝜕𝑈 𝜏𝑤𝑥 −2𝑘𝑧 𝜏𝑏
𝜌
= 𝐴𝑧 𝜕𝑧 + 𝜌
𝑒 + 𝜌
(2-63)
where:

wx = longitudinal wind shear at the surface
4𝜋2
k = wave number = 2
𝑔𝑇𝑤
Tw = wind wave period = 6.95𝐸 − 2 𝐹 0.233 |𝑊|0.534
F = wind fetch length, m
The turbulent eddy viscosity was conceptualized by Prandtl as:
𝑑𝑈
𝜈𝑡𝑢𝑟𝑏𝑢𝑙𝑒𝑛𝑡 = ℓ2 | | (2-64)
𝑑𝑧
where  is defined as the mixing length and can be interpreted as being proportional to the average size of
large eddies or the length scale of a turbulent eddy. This length is a function of distance from a boundary
or wall since the eddy sizes vary as a function of distance from a boundary. The goal in most turbulence
models is the determination of the mixing length as a function of position in the fluid. Because this concept
is not firmly grounded in theory, there have been many published formulations for determination of Az in
the literature (Shanahan and Harleman, 1982).
The mechanism for transporting the wind stress on the surface in the model is based on:
1/2
ℓ2 𝜕𝑈 2 𝜕𝑉 2
𝐴𝑧 = 𝜅 2
[( 𝜕𝑧 ) + ( 𝜕𝑧 ) ] 𝑒 −𝐶𝑅𝑖 (2-65)
where:
𝜕𝜌
𝑔
𝜕𝑧
Ri = Richardson number = 𝜕𝑈 2
𝜌( )
𝜕𝑧
 = von Karman constant, 0.4
C = constant, 1.5
 = vertical length scale chosen as cell thickness, z
The formulation is a typical mixing length formulation that is decreased or increased based on the Richard-
son number. The Richardson number accounts for the impact of density stratification on transfer of mo-
mentum between fluid parcels. In regions where there is no stratification, the Richardson number is zero
and the exponential term is one. For regions where there is strong stratification, the Richardson number
becomes large and the exponential term approaches zero.
In the longitudinal-vertical model, the lateral velocity, V, and its gradient, V/z, are due to the lateral com-
ponent of wind wave motion and are assumed to be zero when averaged laterally, but not necessarily the
square (V/z)2. It is assumed that cross wind shear, wy, generates lateral wave components and decays
exponentially with depth such that
𝜏𝑦𝑧 = 𝜏𝑤𝑦 𝑒 (−2𝑘𝑧) (2-66)
where:
wy = lateral wind shear at the surface
Then using:
𝜏𝑦𝑧 𝜕𝑉
= 𝐴𝑧 (2-67)
𝜌 𝜕𝑧

The lateral velocity gradient squared becomes:
2
𝜕𝑉 2 𝜏𝑤𝑦 𝑒 (−2𝑘𝑧)
( 𝜕𝑧 ) = [ 𝜌𝐴𝑧
] (2-68)
The final equation for the vertical eddy viscosity is then:
2 1⁄2
𝑙 2 𝜕𝑈 2 𝜏𝑦,𝑤𝑖𝑛𝑑 𝑒 −2𝑘𝑧
𝐴𝑧 = 𝜅 2 [( 𝜕𝑧 ) +( 𝜌𝐴𝑧
) ] 𝑒 −𝐶𝑅𝑖 (2-69)
The above equation is implicit. In the model, this equation is explicit since the value of Az in the lateral wind
shear term is used from the previous time step. Az is never less than the molecular kinematic viscosity for
water.
The above formulation of wind shear in horizontal momentum and evaluation of Az leads to wind driven
surface currents that are three to ten percent of the surface wind velocity with higher values appearing at
higher wind speeds. This is in accordance with the attempts to relate wind speed and surface current ve-
locity from field data appearing in the literature. With this formulation, the surface current does not reach
abnormal values as it does for the case of wind shear applied only to the surface and as the surface layer
thickness decreases. The depth of the wind driven surface layer increases with wind speed, and mass
transport due to wind appears to be insensitive to the finite difference layer thickness.
RNG Turbulent Eddy Viscosity Formulation
The RNG model was derived from the RNG model of Yakhot and Orszag (1986) by Simoes (1998). The tur-
bulent eddy viscosity is derived from Yakhot and Orzag (1986) as
1/3
𝜀ℓ4𝑚
𝜈𝑡 = 𝜈 [1 + 𝛹 (𝑎′ − 𝐶1 )] (2-70)
𝜈3
where:
Ψ(x) = max(0,x)
 = molecular viscosity
t = turbulent eddy viscosity
ℓ𝑚 = mixing length
 = turbulent energy dissipation rate
𝑎′  1
C1  100
Two additional equations are necessary in determining the mixing length and the turbulent energy dissipa-
tion. For the mixing length:
ℓ𝑚 𝑧 𝑧
𝐻
= 𝜅 𝐻 √1 − 𝐻 (2-71)
and for the turbulent eddy dissipation:
𝜀𝐻 3𝑧 𝑧 3/2
= (1 − ) (2-72)
𝑢∗3 𝐻 𝐻

where:
u* = shear velocity, m s-1
H = depth of the channel, m
Z = vertical coordinate measured from the bottom of the channel, m
 = von Karman’s constant, 0.41
Substituting these into Equation 2-70:
1/3
𝑧𝑢∗ 3 𝑧 3
𝜈𝑡 = 𝜈 [1 + 𝛹 (3𝜅 ( 𝜈
) (1 − 𝐻
) − 𝐶1 )] (2-73)
Simoes (1998) states that this model better represents experimental data than the more traditional para-
bolic eddy viscosity model of
𝑧
𝜈𝑡 = κz𝑢∗ (1 − 𝐻) (2-74)
A value of  was derived as a function of temperature based on values from Batchelor (1967) using a poly-
nomial curve fit between 0 and 30ºC.
This model was adjusted to account for stratified flow conditions by using the same Richardson number
criteria as used in the original W2 model (the approach of Mamayev as quoted in French, 1985),
𝜈𝑡 = 𝑚𝑎𝑥(𝜈, 𝜈𝑡𝑅𝑁𝐺 𝑒 −𝐶𝑅𝑖 ) (2-75)
where:
𝜕𝜌
𝑔
𝜕𝑧
Ri = Richardson number = 𝜕𝑈 2
𝜌( )
𝜕𝑧
C = empirical constant, 1.5 (French [1985] shows a value of 0.4)
The Richardson number accounts for the impact of density stratification on transfer of momentum between
fluid parcels. In regions where there is no stratification, Ri=0, and the exponential term is 1. For regions
𝑑𝜌
where there is strong stratification (or as → ∞), the Richardson number becomes large and the expo-
𝑑𝑧
nential term approaches 0.
Nikuradse Formulation
This model, as noted in Rodi (1993), is a mixing length model where the mixing length, m, and eddy viscos-
ity, t, are determined from
𝜕𝑢
𝜈𝑡 = ℓ2𝑚 | 𝜕𝑧 | (2-76)
𝑧 2 𝑧 4
ℓ𝑚 = 𝐻 [0.14 − 0.08 (1 − 𝐻) − 0.06 (1 − 𝐻) ] (2-77)
This results in a vertical distribution for the mixing length as shown in Figure 26.

Mixing Length Distribution
1
0.9
Fraction of depth
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0.0 0.5 1.0 1.5 2.0 2.5
Mixing length, m
Figure 26. Mixing length as a function of depth for the Nikuradse formulation.
The stability of the water column affects the mixing length. A Richardson number criteria has been applied
to correct the mixing length for stability effects such as
ℓ𝑚 = ℓ𝑚𝑜 (1 − 7𝑅𝑖) 𝑖𝑓 𝑅𝑖 ≥ 0 (2-78)
ℓ𝑚 = ℓ𝑚𝑜 (1 − 14𝑅𝑖)−0.25 𝑖𝑓 𝑅𝑖 < 0 (2-79)
This is different from the approach of Munk and Anderson (1948) where the Richardson number correction
was applied to the value of Az and not the mixing length directly.
In order to be compatible with the original formulation, the computed value of Az is corrected using the
Mamayev formulation:
𝜈𝑡 = 𝑚𝑎𝑥(𝜈, 𝜈𝑡𝑁𝐼𝐶𝐾 𝑒 −𝐶𝑅𝑖 ) (2-80)
Parabolic Formulation
Another formulation is a parabolic distribution for Az (Engelund, 1978):
𝑧
𝜈𝑡 = 𝜅𝑢∗ 𝑧 (1 − 𝐻) (2-81)
Figure 27 shows the spatial distribution of Az for the parabolic model.

Vertical Eddy Viscosity Distribution
1
Fraction of depth 0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0.00 0.05 0.10 0.15 0.20
Eddy viscosity, m2 s-1
Figure 27. Variation of Az with depth for the parabolic model of Engelund (1978).
In order to be compatible with the original formulation in the model, the computed value of A z is corrected
using the Mamayev formulation:
𝜈𝑡 = 𝑚𝑎𝑥(𝜈, 𝜈𝑡𝑃𝐴𝑅𝐴𝐵 𝑒 −𝐶𝑅𝑖 ) (2-82)
W2N Formulation
The W2N formulation is the same as the W2 model except that the mixing length is no longer the layer
thickness but is computed using Nickaradse's mixing length model. The equations for the W2N formulation
are:
 2    U  2   wy e- 2 kz  (- )
2
A z =      +  CR i
m
2    z    A z  e
  (2-83)
𝑧 2 𝑧 4
ℓ𝑚 = 𝐻 [0.14 − 0.08 (1 − 𝐻) − 0.06 (1 − 𝐻) ] (2-84)
TKE Formulation
The TKE formulation is a typical application of the k- turbulence model. This model includes both vertical
transport of kinetic energy and dissipation by advection and ‘diffusion’ but also horizontal transport by
advection. The model solves 2 partial differential equations for k and  in order to compute the turbulent
eddy viscosity:
𝑘2
𝜈𝑡 = 𝐶𝜇 𝜀
(2-85)
𝜕𝜀𝐵 𝜕𝜀𝐵𝑈 𝜕𝜀𝐵𝑊 𝜕 𝜈 𝜕𝜀 𝜕 𝜈 𝜕𝜀 𝜀 𝜀2

𝜕𝑡
+ 𝜕𝑥
+ 𝜕𝑧
− 𝜕𝑧 (𝐵 𝜎𝑡 𝜕𝑧) − 𝜕𝑥 (𝐵 𝜎𝑡 𝜕𝑥) = 𝐵 (𝐶𝜀1 𝑘 𝑃 + 𝐶𝜀2 𝑘
+ 𝑃𝜀 ) (2-86)
𝜀 𝜀

𝜕𝑘𝐵 𝜕𝑘𝐵𝑈 𝜕𝑘𝐵𝑊 𝜕 𝜈 𝜕𝑘 𝜕 𝜈 𝜕𝑘
+ + − (𝐵 𝑡 ) − (𝐵 𝑡 ) = 𝐵(𝑃 + 𝐺 − 𝜀 + 𝑃𝑘 ) (2-87)
𝜕𝑡 𝜕𝑥 𝜕𝑧 𝜕𝑧 𝜎𝑘 𝜕𝑧 𝜕𝑥 𝜎𝑘 𝜕𝑥
where C is an empirical constant

𝜕𝑈 2
𝑃 = 𝜈𝑡 [( ) ]
𝜕𝑧
𝜈𝑡 2
𝐺=− 𝑁
𝜎𝑡
𝑓 𝐶 𝑈3
𝑃𝑘 = (0.5𝐵) Production term from boundary friction
10𝐶𝑓1.25 𝑈 4
𝑃𝜀 = (0.5𝐵)2
Production term from boundary friction
Brunt–Vaisala frequency 𝑁 = √− 𝑔𝜌 𝑑𝜌
𝑑𝑧
B is the width, U and W are the advective velocities in x and z respectively, and other terms are empirical
constants.
.
Typical values of the above model constants that have been used in other studies are:
k=1.0
=1.3
C=0.09
C1=1.44
C2=1.92
Turbulent Prandtl/Schmidt number,t, =1.0.
The solution of the above equations for k and  used a split solution technique: explicit horizontal and
source-sink term solution followed by an implicit vertical solution. A description of the solution algorithm
below is based on Gould (2006).
Algorithm
The current algorithm is based on the laterally averaged k-ε turbulence model equations shown below:
𝑘2
𝜈𝑡 = 𝐶𝜇 𝜀
(2-88)
𝜕𝑘𝐵 𝜕(𝑘𝑈𝐵) 𝜕(𝑘𝑊𝐵) 𝜕 𝜈 𝜕𝑘
⏟ + ⏟𝜕𝑥 + 𝜕𝑧 − 𝜕𝑧 (𝐵 𝜎𝑡 𝜕𝑧 ) = ⏟ 𝐵(𝑃 + 𝐺 − 𝜀 + 𝑃𝑘 ) (2-89)
𝜕𝑡 ⏟ 𝑘
unsteady convection source/sink
diffusion
𝜕𝜀𝐵 𝜕(𝜀𝑈𝐵) 𝜕(𝜀𝑊𝐵) 𝜕 𝜈𝑡 𝜕𝜀 𝜀 𝜀2
⏟ + ⏟𝜕𝑥 + 𝜕𝑧 − 𝜕𝑧 (𝐵 𝜎 𝜕𝑧) = 𝐵 ⏟ (𝐶𝜀1 𝑘 𝑃 − 𝐶𝜀2 𝑘 + 𝑃𝑒 ) (2-90)
𝜕𝑡 ⏟ 𝜀
unsteady convection diffusion source/sink
where νt, k, and ε are the eddy viscosity, turbulent kinetic energy and turbulent dissipation. P is the pro-
duction term and is calculated using the following equation:
𝜕𝑈 2
𝑃 = 𝜈𝑡 ( 𝜕𝑧 ) (2-91)
and G is the buoyancy term and is represented using
𝜈
𝐺 = − 𝜎𝑡 𝑁 2 (2-92)
𝑡
The production of turbulent kinetic energy and turbulent dissipation from boundary friction are repre-
sented by the terms Pk and Pε. These terms are calculated using the following equations:

𝐶𝑓 |𝑈 3 |
𝑃𝑘 = (0.5𝐵)
(2-93)
10𝐶𝑓1.25 𝑈 4
𝑃𝜀 = (0.5𝐵)2
(2-94)
The friction coefficient Cf is defined as

𝑔 𝑔𝑛2
𝐶𝑓 = 𝐶 2 = 1⁄ (2-95)
𝑅ℎ 3
Where g is gravitational acceleration, C is a Chezy friction factor, n is a Manning’s friction factor, and Rh is
the hydraulic radius. The remaining terms Cμ, Cε, Cε, σk, and σε are empirical constants and the values
used in the current model are shown in Table 3 below (Rodi, 1993).
Table 3. Constants in k-ε model (Rodi, 1993)

𝐶𝜇 𝐶𝜀1 𝐶𝜀2 𝜎𝑘 𝜎𝜀
0.09 1.44 1.92 1.0 1.3
The equations above are solved using a split solution technique in a similar fashion to the horizontal mo-
mentum equation in CE-QUAL-W2. Two different methods for solving the equations were implemented.
In the first method the vertical transport term was solved using an explicit finite difference along with the
horizontal transport term. The second method involved integrating the vertical transport term into a fully
implicit finite difference. Both methods are based on the solutions developed by Wells (2001). The com-
putational grid used in these calculations is shown below.
CE-QUAL-W2 coordinate system ,,P,B

U,Ax,Dx, xx
Segment
W,D z Az
Layer i-1 i i+1
x
z=0 
kt k-1
z
kt+1 k
kb k+1 Hk=zk
xi
z=h at bottom
Figure 28. CE-QUAL-W2 computational grid. Width, density, pressure and water quality state variables
are defined at cell centers. Horizontal velocity, longitudinal eddy viscosity and diffusivity, and longitudi-
nal shear stress are defined at the right-hand side of the cell. Vertical velocity and vertical diffusivity are
defined at the bottom of the cell, and the vertical eddy viscosity is defined at the lower right corner of
the cell.

Explicit Vertical Convection
The first method developed for implementing the convective terms involved using an explicit finite differ-
ence for the vertical convection term. The first step in this method is to split the turbulent kinetic energy
transport equation, equation 2-89, into the two equations shown below
𝜕𝑘𝐵 𝜕(𝑘𝑈𝐵) 𝜕(𝑘𝑊𝐵)
𝜕𝑡
+ 𝜕𝑥
+ 𝜕𝑧 = 𝐵(𝑃 + 𝐺 − 𝜀 + 𝑃𝑘 ) (2-96)
𝜕𝑘𝐵 𝜕 𝜈 𝜕𝑘
𝜕𝑡
= 𝜕𝑧 (𝐵 𝜎𝑡 𝜕𝑧 ) (2-97)
𝑘
Next each equation is averaged over a layer:
1 𝜕 1 𝜕 1 𝜕 1
⏟ (𝑘𝐵𝛥𝑧) + ⏟ (𝑘𝑈𝐵𝛥𝑧) + ⏟ (𝑘𝑊𝐵𝛥𝑧) = ⏟ 𝐵𝛥𝑧(𝑃 + 𝐺 − 𝜀 + 𝑃𝑘 ) (2-98)
𝛥𝑧 𝜕𝑡 𝛥𝑧 𝜕𝑥 𝛥𝑧 𝜕𝑧 𝛥𝑧
unsteady convection convection source/sink
1 𝜕 1 𝜈𝑡 𝜕𝑘 𝑧+𝛥𝑧 𝜈𝑡 𝜕𝑘 𝑧
⏟ (𝑘𝐵𝛥𝑧) = [𝐵 | − 𝐵 | ] (2-99)
𝛥𝑧 𝜕𝑡 𝛥𝑧
⏟ 𝜎𝑘 𝜕𝑧 𝜎𝑘 𝜕𝑧
unsteady diffusion
The partial derivatives in equation 2-98 can be replaced by the finite difference schemes below. The un-
steady term is represented as an explicit finite difference, where k* is the kinetic turbulent energy at the
next time step before the application of equation 2-99.
∗ 𝑛+1 𝑛+1 𝑛 𝑛 𝑛
𝜕 𝑘𝑖,𝑘 𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘 −𝑘𝑖,𝑘 𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘
(𝑘𝐵𝛥𝑧)| = (2-100)
𝜕𝑡 𝑖,𝑘 𝛥𝑡
The horizontal convective term is represented by an upwind difference scheme. The order of differencing
depends on the direction of the horizontal component of the velocity. The difference shown below is for
U>0.
𝑛 𝑛 𝑛 𝑛 𝑛 𝑛 𝑛 𝑛
𝜕 𝑘𝑖,𝑘 𝑈𝑖,𝑘 𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘 −𝑘𝑖−1,𝑘 𝑈𝑖−1,𝑘 𝐵𝑖−1,𝑘 𝛥𝑧𝑖−1,𝑘
(𝑘𝑈𝐵𝛥𝑧)| = 𝑛 (2-101)
𝜕𝑥 𝑖,𝑘 𝛥𝑥𝑖,𝑘
The vertical convective term is also represented by an upwind difference scheme. The order of differenc-
ing depends on the direction of the vertical component of the velocity.
The difference shown below is for W>0.
𝑛 𝑛 𝑛 𝑛 𝑛 𝑛 𝑛 𝑛
𝜕 𝑘𝑖,𝑘 𝑊𝑖,𝑘 𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘 −𝑘𝑖−1,𝑘 𝑊𝑖−1,𝑘 𝐵𝑖−1,𝑘 𝛥𝑧𝑖−1,𝑘
(𝑘𝑊𝐵𝛥𝑧)| = 𝑛 (2-102)
𝜕𝑧 𝑖,𝑘 𝛥𝑧𝑖,𝑘
These differences can be inserted into equation 2-96 and rearranged to give the following explicit solution
for k*.
𝑛 𝑛 𝑛
𝑘𝑖,𝑘 𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘 𝛥𝑡 𝑛 𝑛 𝑛 𝑛 𝑛 𝑛 𝑛 𝑛
𝑘 ∗= 𝑛+1 𝑛+1 − 𝑛+1 𝑛+1 (𝑘𝑖,𝑘 𝑈𝑖,𝑘 𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘 − 𝑘𝑖−1,𝑘 𝑈𝑖−1,𝑘 𝐵𝑖−1,𝑘 𝛥𝑧𝑖−1,𝑘 )
𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘 𝛥𝑥𝑖,𝑘 𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘
𝛥𝑡 𝑛 𝑛 𝑛 𝑛 𝑛 𝑛 𝑛 𝑛
− 𝑛+1 𝑛+1 (𝑘𝑖,𝑘 𝑊𝑖,𝑘 𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘 − 𝑘𝑖−1,𝑘 𝑊𝑖−1,𝑘 𝐵𝑖−1,𝑘 𝛥𝑧𝑖−1,𝑘 )
𝛥𝑥𝑖,𝑘 𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘
𝛥𝑡𝐵𝑛 𝛥𝑧 𝑛
+ 𝐵𝑛+1𝑖,𝑘𝛥𝑧𝑛+1
𝑖,𝑘 𝑛
(𝑃𝑖,𝑘 𝑛
+ 𝐺𝑖,𝑘 𝑛
− 𝜀𝑖,𝑘 + 𝑃𝑘 𝑛𝑖,𝑘 ) (2-103)
𝑖,𝑘 𝑖,𝑘
Where the terms P and G are calculated using the explicit differences shown below.
𝑛 𝑛 𝑛 𝑛 2
(0.5⋅(𝑈𝑘,𝑖 +𝑈𝑘,𝑖−1 )−0.5⋅(𝑈𝑘+1,𝑖 +𝑈𝑘+1,𝑖−1 ))
𝑛
𝑃𝑖,𝑘 = 𝜈𝑡 ( 𝑛
𝛥𝑧𝑘+1/2,𝑖
) (2-104)

𝑛 𝑛
𝑛 𝜌𝑘+1,𝑖 −𝜌𝑘,𝑖
𝐺𝑖,𝑘 = 𝜈𝑡 𝑔 ( 𝑛
𝛥𝑧𝑘,𝑖 𝜌𝑤
) (2-105)
Equation 2-102 is calculated for all of the layers in a segment except for the top and bottom layer.
Next the partial derivatives in equation 2-99 are replaced with the following differences.
𝑛+1 𝑛+1 𝑛+1 ∗ 𝑛+1 𝑛+1

𝜕 𝑘𝑖,𝑘 𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘 −𝑘𝑖,𝑘 𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘
(𝑘𝐵𝛥𝑧)| = (2-106)
𝜕𝑡 𝑖,𝑘 𝛥𝑡
𝜈𝑡 𝜕𝑘 𝑧+𝛥𝑧 𝜈𝑡 𝜕𝑘 𝑧
[𝐵 | −𝐵 | ]| =
𝜎𝑘 𝜕𝑧 𝜎𝑘 𝜕𝑧
𝑖,𝑘
𝑛 𝜈𝑡 𝑛 𝑛+1 𝑛+1
𝑖,𝑘+1/2 𝑘𝑖,𝑘+1 −𝑘𝑖,𝑘 𝑛 𝜈𝑡 𝑛 𝑛+1 𝑛+1
𝑖,𝑘−1/2 𝑘𝑖,𝑘 −𝑘𝑖,𝑘−1
𝐵𝑖,𝑘+1/2 ( 𝑛+1 ) + 𝐵𝑖,𝑘−1/2 ( 𝑛+1 ) (2-107)
𝜎𝑘 𝛥𝑧𝑖,𝑘+1/2 𝜎𝑘 𝛥𝑧𝑖,𝑘−1/2
The resulting equation can then be simplified to a tridiagonal matrix solution format,
𝑛+1
𝐴𝑇 ⋅ 𝑘𝑘−1 + 𝑉𝑇 ⋅ 𝑘𝑘𝑛+1 + 𝐶𝑇 ⋅ 𝑘𝑘+1
𝑛+1
= 𝐷𝑇 (2-108)
where
𝑛
𝛥𝑡 𝐵𝑖,𝑘−1/2 𝜐𝑡 𝑛𝑖,𝑘−1/2
𝐴𝑇 = −
𝑛+1 𝑛+1 ( 𝑛+1 )
𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘 𝛥𝑧𝑖,𝑘−1/2 𝜎𝑘
𝑛 𝑛
𝛥𝑡 𝐵𝑖,𝑘+1/2 𝜈𝑡 𝑛𝑖,𝑘+1/2 𝐵𝑖,𝑘−1/2 𝜈𝑡 𝑛𝑖,𝑘−1/2
𝑉𝑇 = 𝑛+1 𝑛+1 (1 + 𝑛+1 + 𝑛+1 )
𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘 𝛥𝑧𝑖,𝑘+1/2 𝛥𝑧𝑖,𝑘−1/2
𝑛
𝛥𝑡 𝐵𝑖,𝑘+1/2 𝜈𝑡 𝑛𝑖,𝑘+1/2
𝐶𝑇 = − 𝑛+1 𝑛+1 ( 𝑛+1 )
𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘 𝛥𝑧𝑖,𝑘+1/2 𝜎𝑘
𝑛 𝑛
∗
𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘
𝐷𝑇 = 𝑘𝑖,𝑘 𝑛+1 𝑛+1
𝑛 𝑛
and 𝑛+1 𝑛+1
= 1 except at the top layer. The boundary conditions at the top and bottom layers are
implemented using the method shown in the following section. Then kn+1 can be found by solving the equa-
tions using the Thomas algorithm.
The turbulent dissipation is found using the same procedure. Equation 2-90 is split as shown below:
𝜕𝜀𝐵 𝜕(𝜀𝑈𝐵) 𝜕(𝜀𝑊𝐵) 𝜀 (𝜀)2
𝜕𝑡
+ 𝜕𝑥
+ 𝜕𝑧 = 𝐵 (𝐶𝜀1 𝑘 𝑃 − 𝐶𝜀2 𝑘
+ 𝑃𝑒 ) (2-109)
𝜕𝜀𝐵 𝜕 𝜈 𝜕𝜀
𝜕𝑡
= 𝜕𝑧 (𝐵 𝜎𝑡 𝜕𝑧) (2-110)
𝜀
These equations are then averaged over a layer.
1 𝜕 1 𝜕 1 𝜕
(𝜀𝐵𝛥𝑧) + (𝜀𝑈𝐵𝛥𝑧) + (𝜀𝑊𝐵𝛥𝑧) =
⏟
𝛥𝑧 𝜕𝑡 ⏟
𝛥𝑧 𝜕𝑥 ⏟
𝛥𝑧 𝜕𝑧
unsteady convection convection
1 𝜀 (𝜀)2
⏟
𝛥𝑧
𝐵𝛥𝑧 (𝐶 𝜀1 𝑘
𝑃 − 𝐶 𝜀2 𝑘
+ 𝑃𝑒 ) (2-111)
source/sink
1 𝜕 1 𝜈 𝜕𝜀 𝑧+𝛥𝑧 𝜈𝑡 𝜕𝜀 𝑧
⏟ (𝜀𝐵𝛥𝑧) = [𝐵 𝑡 | − 𝐵 | ] (2-112)
𝛥𝑧 𝜕𝑡 𝛥𝑧
⏟ 𝜎𝑘 𝜕𝑧 𝜎𝑘 𝜕𝑧
unsteady diffusion

The partial derivatives in equation 2-111 are replaced with finite differences using similar schemes to
those used in the turbulent kinetic energy solution. This results in the explicit solution for ε* shown be-
low.
𝑛 𝑛 𝑛
𝜀𝑖,𝑘 𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘 𝛥𝑡 𝑛 𝑛 𝑛 𝑛 𝑛 𝑛 𝑛 𝑛
𝜀 ∗= 𝑛+1 𝑛+1 − 𝑛+1 𝑛+1 (𝜀𝑖,𝑘 𝑈𝑖,𝑘 𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘 − 𝜀𝑖−1,𝑘 𝑈𝑖−1,𝑘 𝐵𝑖−1,𝑘 𝛥𝑧𝑖−1,𝑘 )
𝛥𝑡 𝑛 𝑛 𝑛 𝑛 𝑛 𝑛 𝑛 𝑛
− 𝑛+1 𝑛+1 (𝜀𝑖,𝑘 𝑊𝑖,𝑘 𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘 − 𝜀𝑖−1,𝑘 𝑊𝑖−1,𝑘 𝐵𝑖−1,𝑘 𝛥𝑧𝑖−1,𝑘 )
𝛥𝑥𝑖,𝑘 𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘
𝛥𝑡𝐵𝑛 𝛥𝑧 𝑛 𝜀𝑛 𝑛 2
(𝜀𝑖,𝑘 )
+ 𝐵𝑛+1𝑖,𝑘𝛥𝑧𝑛+1
𝑖,𝑘
(𝐶𝜀1 𝑘𝑖,𝑘 𝑛
𝑛 𝑃𝑖,𝑘 + 𝐶𝜀2 𝑛
𝑘𝑖,𝑘
+ 𝑃𝑒 𝑛𝑖,𝑘 ) (2-113)
𝑖,𝑘 𝑖,𝑘 𝑖,𝑘
where ε* is the turbulent dissipation at the next time step before the application of equation 2-112. Equa-
tion 2-113 is calculated for all of the layers in a segment except for the top and bottom layer.
Next the partial derivatives in equation 2-112 are replaced with finite differences using similar schemes to
those used in the turbulent kinetic energy solution. This again results in a set of equations in a tridiagonal
format.
𝑛+1
𝐴𝑇 ⋅ 𝜀𝑘−1 + 𝑉𝑇 ⋅ 𝜀𝑘𝑛+1 + 𝐶𝑇 ⋅ 𝜀𝑘+1
𝑛+1
= 𝐷𝑇 (2-114)
where
𝑛
𝛥𝑡 𝐵𝑖,𝑘−1/2 𝜐𝑡 𝑛𝑖,𝑘−1/2
𝐴𝑇 = − 𝑛+1 𝑛+1 ( 𝑛+1 )
𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘 𝛥𝑧𝑖,𝑘−1/2 𝜎𝑘
𝑛 𝑛
𝛥𝑡 𝐵𝑖,𝑘+1/2 𝜈𝑡 𝑛𝑖,𝑘+1/2 𝐵𝑖,𝑘−1/2 𝜈𝑡 𝑛𝑖,𝑘−1/2
𝑉𝑇 = 𝑛+1 𝑛+1 (1 + 𝑛+1 + 𝑛+1 )
𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘 𝛥𝑧𝑖,𝑘+1/2 𝛥𝑧𝑖,𝑘−1/2
𝑛
𝛥𝑡 𝐵𝑖,𝑘+1/2 𝜈𝑡 𝑛𝑖,𝑘+1/2
𝐶𝑇 = − 𝑛+1 𝑛+1 ( 𝑛+1 )
𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘 𝛥𝑧𝑖,𝑘+1/2 𝜎𝑘
𝑛 𝑛
∗
𝐷𝑇 = 𝜀𝑖,𝑘 𝑛+1 𝑛+1
𝐵𝑛 𝛥𝑧 𝑛
𝑖,𝑘 𝑖,𝑘
and 𝐵𝑛+1 𝛥𝑧 𝑛+1
= 1 except at the top layer. These equations are solved for εn+1 using the Thomas algo-
𝑖,𝑘 𝑖,𝑘
rithm.
Implicit Vertical Convection
The second method is very similar to the technique developed above. The only difference is that the ver-
tical convection term is integrated into the implicit part of the solution. First equation 2-99 is split into
horizontal and vertical components as shown below:
𝜕𝑘𝐵 𝜕(𝑘𝑈𝐵)
𝜕𝑡
+ 𝜕𝑥
= 𝐵(𝑃 + 𝐺 − 𝜀 + 𝑃𝑘 ) (2-115)
𝜕𝑘𝐵 𝜕(𝑘𝑊𝐵) 𝜕 𝜈 𝜕𝑘
𝜕𝑡
+ 𝜕𝑧
= 𝜕𝑧 (𝐵 𝜎𝑡 𝜕𝑧 ) (2-116)
𝑘
Next each equation is averaged over a layer.

1 𝜕 1 𝜕 1
⏟ (𝑘𝐵𝛥𝑧) + ⏟ (𝑘𝑈𝐵𝛥𝑧) = ⏟ 𝐵𝛥𝑧(𝑃 + 𝐺 − 𝜀 + 𝑃𝑘 ) (2-117)
𝛥𝑧 𝜕𝑡 𝛥𝑧 𝜕𝑥 𝛥𝑧

1 𝜕 1 𝜕 1 𝜈𝑡 𝜕𝑘 𝑧+𝛥𝑧
⏟ (𝑘𝐵𝛥𝑧) + ⏟ (𝑘𝑊𝐵𝛥𝑧) = 𝐵 | (2-118)
𝛥𝑧 𝜕𝑡 𝛥𝑧 𝜕𝑧 𝛥𝑧 𝜎𝑘 𝜕𝑧 𝑧
⏟
unsteady convection diffusion
Then the differences developed in the previous section, equations 2-100 and 2-101, are substituted into
equation 2-117. The resulting equation can then be rearranged and simplified in order to explicitly solve
for k* as shown in the equation below.
𝑛 𝑛 𝑛
𝑘𝑖,𝑘 𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘 𝛥𝑡 𝑛 𝑛 𝑛 𝑛 𝑛 𝑛 𝑛 𝑛
𝑘 ∗= 𝑛+1 𝑛+1 − 𝑛+1 𝑛+1 (𝑘𝑖,𝑘 𝑈𝑖,𝑘 𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘 − 𝑘𝑖−1,𝑘 𝑈𝑖−1,𝑘 𝐵𝑖−1,𝑘 𝛥𝑧𝑖−1,𝑘 )
𝛥𝑡𝐵𝑛 𝛥𝑧𝑖,𝑘
𝑛
+ 𝑛+1𝑖,𝑘 𝑛+1 𝑛
(𝑃𝑖,𝑘 𝑛
+ 𝐺𝑖,𝑘 𝑛
− 𝜀𝑖,𝑘 + 𝑃𝑘 𝑛𝑖,𝑘 ) (2-119)
𝐵 𝛥𝑧
𝑖,𝑘 𝑖,𝑘
where k* is the kinetic turbulent energy at the next time step before the application of equation 2-118.
Next the finite differences developed in the previous section, equations 2-106 and 2-107, along with the
finite difference for the vertical convective term,
𝑛+1 𝑛 𝑛 𝑛+1 𝑛+1 𝑛 𝑛 𝑛+1

𝜕 𝑘𝑖,𝑘+1/2 𝑊𝑖,𝑘 𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘 −𝑘𝑖,𝑘−1/2 𝑊𝑖,𝑘 𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘
(𝑘𝑊𝐵𝛥𝑧) = (2-120)
𝜕𝑧 𝛥𝑧𝑘𝑛+1
𝑛+1 𝑛+1
𝑛+1 𝑘𝑖,𝑘+1 +𝑘𝑖,𝑘
𝑘𝑖,𝑘+1/2 = 2
(2-121)
𝑛+1 𝑛+1
𝑛+1 𝑘𝑖,𝑘 +𝑘𝑖,𝑘−1
𝑘𝑖,𝑘−1/2 = (2-122)
2
are substituted into equation 2-118.
The resulting equation can then be simplified to a tridiagonal matrix solution format,
𝑛+1
𝐴𝑇 ⋅ 𝑘𝑘−1 + 𝑉𝑇 ⋅ 𝑘𝑘𝑛+1 + 𝐶𝑇 ⋅ 𝑘𝑘+1
𝑛+1
= 𝐷𝑇 (2-123)
where
𝑛
𝛥𝑡 𝑛
𝑊𝑖,𝑘−1/2 𝑛
𝐵𝑖,𝑘−1/2 𝐵𝑖,𝑘−1/2 𝜐𝑡 𝑛𝑖,𝑘−1/2
𝐴𝑇 = − 𝑛+1 𝑛+1 ( + 𝑛+1 )
𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘 2 𝛥𝑧𝑖,𝑘−1/2 𝜎𝑘
𝑛
𝛥𝑡 𝑛+1
𝐵𝑖,𝑘 𝑛+1
𝛥𝑧𝑖,𝑘 𝑛
𝑊𝑖,𝑘+1/2 𝑛
𝐵𝑖,𝑘+1/2 𝑛
𝐵𝑖,𝑘+1/2 𝐵𝑖,𝑘+1/2 𝜈𝑡 𝑛𝑖,𝑘+1/2
𝑉𝑇 = 𝑛+1 𝑛+1 (1 + + − + 𝑛+1
𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘 𝛥𝑡 2 2 𝛥𝑧𝑖,𝑘+1/2
𝑛
𝐵𝑖,𝑘−1/2 𝜈𝑡 𝑛𝑖,𝑘−1/2
+ 𝑛+1 )
𝛥𝑧𝑖,𝑘−1/2
𝑛
𝛥𝑡 𝑛
𝐵𝑖,𝑘+1/2 𝑛+1
𝛥𝑧𝑖,𝑘 𝐵𝑖,𝑘+1/2 𝜈𝑡 𝑛𝑖,𝑘+1/2
𝐶𝑇 = − 𝑛+1 𝑛+1 ( 𝑛+1 − 𝑛+1 )
𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘 2𝛥𝑧𝑖,𝑘 𝛥𝑧𝑖,𝑘+1/2 𝜎𝑘
∗ 𝐵𝑛 𝛥𝑧 𝑛
𝑖,𝑘 𝑖,𝑘
𝐷𝑇 = 𝑘𝑖,𝑘 𝐵𝑛+1 𝛥𝑧 𝑛+1
𝑖,𝑘 𝑖,𝑘
𝑛 𝑛
and 𝑛+1
𝛥𝑧𝑖,𝑘
= 1 except at the top layer. The boundary conditions at the top and bottom layers are
implemented using the method shown in the following section. Then kn+1 is found by solving the equa-
tions using the Thomas algorithm.
The turbulent dissipation is found using the same procedure. Equation 2-99 is split as shown below:

𝜕𝜀𝐵 𝜕(𝜀𝑈𝐵) 𝜀 (𝜀)2
𝜕𝑡
+ 𝜕𝑥
= 𝐵 (𝐶𝜀1 𝑘 𝑃 − 𝐶𝜀2 𝑘
+ 𝑃𝑒 ) (2-124)
𝜕𝜀𝐵 𝜕(𝜀𝑊𝐵) 𝜕 𝜈 𝜕𝜀
+ = (𝐵 𝑡 ) (2-125)
𝜕𝑡 𝜕𝑧 𝜕𝑧 𝜎𝜀 𝜕𝑧
These equations are then averaged over a layer.
1 𝜕 1 𝜕 1 𝜀 (𝜀)2
⏟ 𝜕𝑡 (𝜀𝐵𝛥𝑧) + ⏟ (𝜀𝑈𝐵𝛥𝑧) = ⏟ 𝐵𝛥𝑧 (𝐶 𝜀1 𝑃 − 𝐶𝜀2 + 𝑃𝑒 ) (2-126)
𝛥𝑧 𝛥𝑧 𝜕𝑥 𝛥𝑧 𝑘 𝑘
1 𝜕 1 𝜕 1 𝜈𝑡 𝜕𝜀 𝑧+𝛥𝑧
⏟ (𝜀𝐵𝛥𝑧) + ⏟ (𝜀𝑊𝐵𝛥𝑧) = 𝐵 | (2-127)
𝛥𝑧 𝜕𝑡 𝛥𝑧 𝜕𝑧 𝛥𝑧
⏟ 𝜎𝜀 𝜕𝑧 𝑧
unsteady unsteady diffusion
The partial derivatives in equation 2-111 are replaced with finite differences using similar schemes to
those used in the turbulent kinetic energy solution. This results in the explicit solution for ε* shown be-
low.
𝑛 𝑛 𝑛
𝜀𝑖,𝑘 𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘 𝛥𝑡 𝑛 𝑛 𝑛 𝑛 𝑛 𝑛 𝑛 𝑛
𝜀 ∗= 𝑛+1 𝑛+1 − 𝑛+1 𝑛+1 (𝜀𝑖,𝑘 𝑈𝑖,𝑘 𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘 − 𝜀𝑖−1,𝑘 𝑈𝑖−1,𝑘 𝐵𝑖−1,𝑘 𝛥𝑧𝑖−1,𝑘 )
𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘 𝛥𝑥𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘
𝛥𝑡𝐵𝑛 𝛥𝑧𝑖,𝑘
𝑛
𝜀𝑛 𝑛
𝑛 2
(𝜀𝑖,𝑘 ) 𝑛
+ 𝐵𝑛+1𝑖,𝑘𝛥𝑧𝑛+1 (𝐶𝜀1 𝑘𝑖,𝑘
𝑛 𝑃 𝑘,𝑖 + 𝐶 𝜀2 𝑘 𝑛 + 𝑃𝑒 𝑖,𝑘 ) (2-128)
𝑖,𝑘 𝑖,𝑘 𝑖,𝑘 𝑖,𝑘
Where ε* is the turbulent dissipation at the next time step before the application of equation 2-126.
Next the partial derivatives in equation 2-127 are replaced with finite differences using similar schemes to
those used in the turbulent kinetic energy solution. This again results in a set of equations in a tridiagonal
format.
𝑛+1
𝐴𝑇 ⋅ 𝜀𝑘−1 + 𝑉𝑇 ⋅ 𝜀𝑘𝑛+1 + 𝐶𝑇 ⋅ 𝜀𝑘+1
𝑛+1
= 𝐷𝑇 (2-129)
where
𝑛
𝛥𝑡 𝑛
𝐵𝑖,𝑘−1/2 𝐵𝑖,𝑘−1/2 𝜈𝑡 𝑛𝑖,𝑘−1/2
𝐴𝑇 = − 𝑛+1 𝑛+1 ( + 𝑛+1 )
𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘 2 𝛥𝑧𝑖,𝑘−1/2 𝜎𝑘
𝑛
𝛥𝑡 𝑛+1
𝛥𝑧𝑖,𝑘 𝑛
𝐵𝑖,𝑘+1/2 𝑛
𝐵𝑖,𝑘+1/2 𝐵𝑖,𝑘+1/2 𝜈𝑡 𝑛𝑖,𝑘+1/2
𝑉𝑇 = 𝑛+1 𝑛+1 (1 + + − + 𝑛+1
𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘 𝛥𝑡 2 2 𝛥𝑧𝑖,𝑘+1/2
𝑛
𝐵𝑖,𝑘−1/2 𝜈𝑡 𝑛𝑖,𝑘−1/2
+ 𝑛+1 )
𝛥𝑧𝑖,𝑘−1/2
𝑛
𝛥𝑡
𝑛
𝐵𝑖,𝑘+1/2 𝐵𝑖,𝑘+1/2 𝜈𝑡 𝑛𝑖,𝑘+1/2
𝐶𝑇 = − 𝑛+1 𝑛+1 ( − 𝑛+1 )
𝐵𝑖,𝑘 𝛥𝑧𝑖,𝑘 2 𝛥𝑧𝑖,𝑘+1/2 𝜎𝑘
𝑛 𝑛
∗
𝐷𝑇 = 𝜀𝑖,𝑘 𝑛+1 𝑛+1

𝐵𝑛 𝛥𝑧 𝑛
𝑖,𝑘 𝑖,𝑘
and 𝐵𝑛+1 𝛥𝑧 𝑛+1
= 1except at the top layer. These equations are then solved for εn+1 using the
𝑖,𝑘 𝑖,𝑘
Thomas algorithm.
Boundary and initial conditions for k- model
In order to be able to apply the algorithm developed above to the typical hydraulic problems investigated
using CE-QUAL-W2 the appropriate boundary conditions for the bed, free surface, inlet and outlet as well
as initial conditions must be specified. This section outlines the boundary conditions and initial conditions
implemented in the current version of the k-ε algorithm.
At the bed steep gradients of turbulent kinetic energy and turbulent dissipation are present in the viscous
sublayer. To accurately calculate these steep gradients a large number of grid points are needed in the
viscous sublayer and a low Reynolds number version of the k-ε turbulence model must be used. To avoid
this problem an empirical law can be used to construct a simple wall function.
To construct a wall function several assumptions must be made about flow in the inner region near the
wall. The largest assumption is that the flow parallel to the wall obeys the law of the wall shown below.
𝑈 1 𝑦𝑈∗
= 𝑙𝑛 (𝐸 ) (2-130)
𝑈∗ 𝜅 𝜈
Where U* is the bed friction velocity, κ is the von Karman constant, y is the distance of the grid point from
the bed, E is the roughness coefficient and υ is the kinematic viscosity. In addition to the assumption that
the flow has a logarithmic velocity profile it is necessary to assume that the total shear stress remains
constant in the inner region and is equal to the bed stress. These assumptions can normally be made if
the first grid point is in the range of 30 < 𝑦 + < 100 from the wall (Rodi, 1993).
𝑦𝑈∗
𝑦+ = (2-131)
𝜈
In the range of y+ specified above the Reynolds stresses are relatively constant. In this region local equilib-
rium prevails, so if buoyancy effects are negligible then the production equals dissipation which implies
that P = ε. It is also assumed that the total shear stress remains constant and is approximately equal to
the wall shear stress in this region. Using these assumptions and equation 1 it is possible to derive the
equation for turbulent kinetic energy at the first grid point shown below (Rodi, 1993).
𝑈∗2
𝑘= (2-132)
√𝑐𝜇
𝜕𝑈 𝜕𝑈
The boundary condition for turbulent dissipation can be determined using 𝜀 = 𝑃 = 𝑈∗2 where is de-
𝜕𝑦 𝜕𝑦
termined from equation 2-130 and is given below.
𝑈∗3
𝜀= (2-133)
𝜅𝑦
Since y, U, E, 𝜅, ν and E are known, a very simple wall function can be implemented by solving equation 2-
130 for U*. In the current version of the k-ε model U* was found using a root solver based on the bisec-
tion method from Press et al. (1996). Then equations 2-132 and 2-133 are used to determine the turbu-
lent kinetic energy and the turbulent dissipation at the first grid point. This simple approach has some
limitations, such as not being able to handle separated flow, but seemed like a reasonable first approxi-
mation.

The wall function outlined above is valid for both smooth and rough boundaries. For hydrau-
lically smooth beds the roughness coefficient E is set to 9.535 (Ferziger and Peric, 2002). For rough
boundaries a smaller value of E is used. Two options are available in the current algorithm for specifying
the roughness coefficient. A uniform constant roughness coefficient value can be applied to all segments
or the roughness coefficient can be calculated based on the Manning’s coefficients specified in the ba-
thymetry file.
In the first option the user specifies the roughness coefficient and this value is used to calculate
the bed friction velocity using equation 2-130. After the bed friction velocity has been determined a new
Manning’s coefficient is calculated using equation 2-134 below. This method was used primarily for ex-
perimental channel test cases where the bed was hydraulically smooth and had a roughness coefficient
equal to 9.535.
1/6
𝑅ℎ
𝑛= 𝑔
(2-134)
√( 𝑈∗ ) 2
𝑈𝑚
In the second option the roughness coefficient is calculated using a Strickler relationship and an approxi-
mation of the Nikuradse’s data shown in equations 2-135 to 2-137 below (Krishnappan and Lau, 1986).
(𝜅𝐵𝑠 )
𝐸 = 𝑒𝑥𝑝 ( 𝑈 𝑘 ) (2-135)
( ∗ 𝑠)
𝜈
𝑈∗ 𝑘𝑠 𝑈∗ 𝑘𝑠 2
𝐵𝑠 = [5.50 + 2.5 𝑙𝑛 ( )] 𝑒𝑥𝑝 {−0.217 [𝑙𝑛 ( )] }
𝜈 𝜈
𝑈∗ 𝑘𝑠 2
+8.5 (1 − 𝑒𝑥𝑝 {−0.217 [𝑙𝑛 ( )] }) (2-136)
𝜈
𝑘𝑠 = (𝑛 ⋅ 24.04)6 (ks in m) (2-137)
The appropriate boundary condition for the free surface is harder to determine. Unfortunately, there is
very little experimental data available for turbulent statistics at the free surface (Rodi, 1993). It is difficult
to make such measurements because hot wire anemometers tend to disturb the surface and laser Dop-
pler anemometry systems, LDA, have issues with reflections and refraction caused by the free surface
(Swean et al., 1991). In many cases researchers have turned to Direct Navier Stokes simulations for more
information on the relationship between turbulent kinetic energy and turbulent dissipation and the free
surface (Swean et al., 1991, Cotton et al., 2005).
A first approximation for the boundary conditions at the free surface in absence of a wind induced shear
is to use a symmetry plane where both turbulent kinetic energy and turbulent dissipation are found using
a zero-gradient condition (Rodi, 1993). However, it is generally thought that the presence of the free sur-
face should reduce the length scale of turbulence and reduce the turbulent dissipation. An attempt to
take this into account was made in the empirically determined boundary conditions shown below (Celik
and Rodi, 1984).
𝜕𝑘
𝜕𝑦
=0 (2-138)
𝑘 3/2
𝜀= 𝑎𝐻
(2-139)
Where a is an empirical constant and H is the depth of the shear layer, in this case the depth of the chan-
nel. The coefficient a was originally set to a value of 0.18 by Celik and Rodi (1984) and then later changed
to a value of 0.43 by Celik and Rodi (1988).

For a free surface with wind induced surface shear, the boundary conditions similar to those used for wall
boundaries are appropriate, equations 2-130, 2-132, and 2-133. The free surface can be seen as a moving
wall. The bed friction velocity in equations 2-132 and 2-133 can be replaced with the wind induced sur-
face friction velocity (Rodi, 1993).
Rodi presented a set of boundary conditions that could be used for free surfaces with or without wind in-
duced surface shear. When the wind induced surface shear is large then the following boundary condition
is used for turbulent kinetic energy.
𝑈∗2𝑆
𝑘= (2-140)
√𝐶𝜇
When the surface shear is small, 𝑘𝑠 √𝑐𝜇 > 𝑈∗2𝑆 ,
then the symmetry condition for turbulent kinetic energy
shown in equation 2-138 is used. Where U*S is the wind induced surface friction velocity. In a similar
fashion the turbulent dissipation boundary condition for a free surface with and without wind induced
surface shear was developed and is given as
3/2
(𝑘𝑠 √𝐶𝜇 )
𝜀= 𝑈2
(2-141)
𝜅[𝑦+𝑎𝐻(1− ∗𝑠 )]
𝑘𝑠√𝐶𝜇
The coefficient a was set to a value of 0.07 by Rodi in 1983. When surface shear is present, equation 2-
3
𝑈∗𝑠 𝑘 3/2
140 reduces to 𝜀 = . When surface shear is not present, equation 2-141 becomes 𝜀 = which is
𝜅⋅𝑦 𝜅[𝑦+𝑎𝐻]
similar to equation 2-139 above.
The original CE-QUALW2 free surface boundary conditions were

2 2
(𝑈∗𝑠 + 𝑈∗𝑏𝑜𝑢𝑛𝑑𝑎𝑟𝑦 )
𝑘=
√𝐶𝜇
3 3
(𝑈∗𝑠 + 𝑈∗𝑏𝑜𝑢𝑛𝑑𝑎𝑟𝑦 )
𝜀=
𝜅⋅𝑦
These are similar to the boundary conditions presented in equations 2-140 and 2-141 when wind induced
shear is present. The only difference is that both the turbulent kinetic energy and turbulent dissipation
boundary conditions include an additional term for the friction velocity of the lateral boundaries.
The initial conditions and inlet boundary conditions are specified by setting the eddy viscosity, turbulent
kinetic energy, and turbulent dissipation terms to the following values.
υt = 1.4∙10-6
k = 1.25∙10-7
ε = 1.0∙10-9
The outlet values of turbulent kinetic energy and turbulent dissipation are determined using a zero-gradi-
ent approach as shown below
𝜕𝑘
=0
𝜕𝑥
𝜕𝜀
=0
𝜕𝑥
This is done by setting the value of turbulent kinetic energy and turbulent dissipation at the outlet cell
equal to the value of the cell in the upstream segment.
Computations of k and  compared to experimental data are shown in Gould (2006). The experimental
data were taken from a uniform open channel experiment conducted by Nakagawa, Neuzu, and Ueda
(1975). Figure 29 and Figure 30 show CE-QUAL-W2 model comparisons to the experimental data varying

Manning’s friction factor. As shown in Gould (2006), the number of vertical layers used in the model appli-
cation was very important in being able to match experimental laboratory data.
Turbulent Dissipation
1.0
0.9
0.8
0.7
0.6 n = 0.009
n = 0.010
y/h
0.5
n = 0.012
0.4 Exp
0.3
0.2
0.1
0.0
0.0E+00 1.0E-04 2.0E-04 3.0E-04 4.0E-04 5.0E-04 6.0E-04
 [m /s ]
2 3
Figure 29. Comparison of CE-QUAL-W2 model predictions of turbulent dissipation with field data from
Nakagawa, Neuzu, and Ueda (1975).
Turbulent Kinetic Energy

1.0
0.9
0.8
0.7
0.6 n = 0.009
n = 0.010
y/h
0.5
n = 0.012
0.4 Exp
0.3
0.2
0.1
0.0
0.0E+00 1.0E-04 2.0E-04 3.0E-04 4.0E-04
2 2
k [m /s ]
Figure 30. Comparison of CE-QUAL-W2 model predictions of turbulent kinetic energy with field data
from Nakagawa, Neuzu, and Ueda (1975).

Effect of Vertical Layer Numbers on Vertical Turbulence
In contrast to other riverine models that assume vertically well-mixed systems, CE-QUAL-W2 accounts for
the vertical variation of velocity in a riverine reach. Even though there is an added computational burden
of computing the 2D velocity profile, the advantage of making this computation is that the friction factor
(Manning's or Chezy) for a segment can be flow or stage invariant depending on the number of vertical
layers schematized.
Many 1D hydraulic flow models, such as CE-QUAL-RIV1 and UNET (Barkau, 1997), allow the model user to
specify how Manning’s friction factor changes with depth. The Mannning’s friction factor, n, has been
thought to vary as a function of depth, Reynolds number, and roughness factor or scale of bed grain size
(Ugarte and Madrid, 1994; Soong, et. el., 1995). Some of these formulations for variation of Manning's
friction factor with hydraulic radius, R, are shown in Figure 31:
7 Jarrett (1984) :
6 n = 0.39S 0.38 R −0.16
Hydraulic radius, ft
5 Limerinos (1970 ) :
0.0926 R1/ 6
4
n=
 R 
3 1.16 + 2 log  
 d 84 
2
0
0.02 0.025 0.03 0.035 0.04 0.045 0.05 0.055 0.06 0.065 Limerinos (1970)
Manning's friction factor Jarrett (1984)
Figure 31. Variation of Manning's friction factor using formulae from Limerinos (1970) and Jarrett
(1984, 1985) for a channel slope, S, of 0.0005 and 84th pecentile diameter of the bed material, d84, of 50.
Researchers understand that the friction factor, when representing a hydraulic element with uniform
roughness, should be flow invariant with depth (Henderson 1966). However, many assert that the friction
factor changes with depth because the friction coefficient is variable with the wetted perimeter. Some
investigators reason that it is to be expected that at shallow depths the larger size of the bed material
produces a higher overall friction factor than a deeper flow where the sidewalls may have a smaller friction.
Since most researchers used 1D, cross-sectionally averaged flow equations such as Manning’s Equation or
1D dynamic hydraulic models, this parameterization itself has been responsible for the seeming variation
of Manning’s friction factor with depth. For example, all 1D hydraulic models implicitly assume that the
rate of transfer of momentum from the bottom of the channel to the top is “infinite” in that regardless of
the depth, the shear stress of the bottom is felt even at the surface. For these hydraulic models, even as
the depth of the channel increases, they still assume the surface feels the shear stress at the channel bot-
tom and this effect is transferred over the depth of the channel. Therefore, as the water depth increases,
the apparent friction factor must be reduced because of the assumption of momentum transfer between
the bottom and the surface.

However, in a 2-D, longitudinal-vertical, river model, Manning's friction factor does not have to vary with
stage in order to produce the effect that as the river stage increases, the apparent friction decreases. The
water surface set-up changes significantly as the number of layers increase. In general, the water surface
slope increases as the number of computational layers decreases. This is because the average eddy viscos-
ity in the water column increases as the number of layers decrease until at the limit of a one-layer system,
the average vertical eddy viscosity is infinite. The fact that the Manning’s friction factor seems to decrease
with depth in 1D models is accounted for in modeling the river channel as a 2D, longitudinal-vertical system.
The model allows choosing between six different vertical eddy viscosity formulations. These formulations
are shown in Table 2. Typical variation of these formulations is shown in Figure 24 for Manning's friction
factor for an open-channel, non-stratified flow regime as compared to theory of steady, uniform channel
flow.
The number of vertical layers significantly affects model predictions. For example, Figure 32 shows a com-
parison of vertical velocity profiles from a model with one, three, and seven vertical layers using the para-
bolic eddy viscosity model.
S=0.001, Q=2574 m3 s-1, n=0.03, PARAB
86
84
82
Elevation, m
80
3-layer
78 7-layer
1-layer
76
74
72
70
0 0.5 1 1.5 2 2.5
Velocity, m/s
Figure 32. Comparison of vertical velocity predictions with one, three, and seven vertical layers
Figure 33 shows how the change in the number of vertical layers affects the water surface slope over the
domain length for a steady-state flow. In order to model the water surface slope of the 1-layer model with
the 7-layer model, the apparent value of Manning's friction factor would have to be reduced. Hence, the
apparent friction decreases as the number of layers increase.

Q=2574 m3 s-1, S=0.0001, n=0.03
2.94
Elevation drop over 30 km distance, m
2.93
2.92
2.91
2.9
2.89
2.88
0 1 2 3 4 5 6 7 8
# of vertical layers
Figure 33. Comparison of elevation drop of W2 model with one, three, and seven vertical lay-
ers with same Manning's friction factor.
CE-QUAL-W2 has also been compared to the 1D models DYNHYD (Ambrose et al., 1988) and CE-QUAL-RIV1
(Environmental Laboratory, 1995) by running W2 with only a single vertical layer. The average velocities
between the three models agreed well with theory but the water surface slopes are different. W2 predicts
an elevation difference of 2.93 m, compared to 2.07 m for DYNHYD and 2.05 m for RIV1 over 30 km for a
flow of 2574 m3 s-1, a Manning’s n of 0.03, a slope of 0.001, and a channel width of 100 m. Based on steady-
state theory, the actual difference should be 2.9 m. Both the DYNHYD and RIV1 models require friction
factors greater than expected to correspond to classical theory. This may be a result of these models not
incorporating sidewall friction that was important during these test runs where the depth was 15 m and
the width was 100 m.

Longitudinal Shear Stress
The longitudinal turbulent shear stress is defined as
𝜏𝑥𝑥 𝜕𝑈 𝜕𝑈
𝜌
= 𝜈𝑡 𝜕𝑥 = 𝐴𝑥 𝜕𝑥 (2-142)
where:
Ax = t = longitudinal eddy viscosity
Ax is a user-defined constant or it can be set to scale with the average longitudinal velocity. This turbulence
closure approximation is termed a zero-order closure model since no further equations are necessary to
solve for the transmission of shear stress within the fluid.
This term is usually of very low magnitude except in areas near boundaries such as at a dam face where the
longitudinal velocity goes to zero.
The equation for computing Ax is

𝐴𝑥 =∝ 𝑈 ̿∆𝑧
Where α is a user input value, U with double bars are the time-averaged, lateral averaged longitudinal ve-
locity at a segment and layer.
Hydraulic Structures
Pipes
The model user can specify a pipe or culvert between model segments (Berger and Wells, 1999) using a 1D,
unsteady hydraulic model that computes the flow between the two linked segments. The model computes
the selective withdrawal outflow from the upstream segment with the model user specifying whether the
inflow to the downstream segment is treated as mixed over the depth, inflow depth is determined from
inflow density, or inflow depth is specified between an upper and lower elevation. The flow between an
upstream segment and a downstream segment is shown in Figure 34.
upstream downstream
branch branch
computational
cells
Figure 34. Schematic of linkage of model segments with a culvert.
This model is only appropriate for simple piping systems that are not suddenly under a large hydraulic head.
The governing equations for computing the flow and the numerical solution technique are shown below.

HYDRAULIC STRUCTURES AUXILIARY FUNCTIONS
The governing equations used to predict flow through culverts are the 1D, time-dependent conservation of
momentum and continuity equations (Yen, 1973).
𝜕𝑢 𝜕𝑢 𝜕ℎ
+𝑢 + 𝑔 𝑐𝑜𝑠 𝜑 − 𝑔(𝑆𝑜 − 𝑆𝑓 − 𝑆𝑚 ) = 0 (2-143)
𝜕𝑡 𝜕𝑥 𝜕𝑥
𝜕ℎ 𝜕ℎ 𝐴 𝜕𝑢
𝜕𝑡
+ 𝑢 𝜕𝑥 + 𝑇 𝜕𝑥 = 0 (2-144)
where:
u = velocity, m s-1
t = time, s
h = piezometric head, m
g = gravitational acceleration, m2 s-1
x = distance along axis of culvert, m
A = cross-sectional area of culvert filled with water, m2
T = width of water level surface, m
 = angle between culvert axis and horizontal
So = culvert slope
Sf = friction slope.
Sm = minor loss slope
The friction slope Sf is estimated using the Manning formula:
𝑛2
𝑆𝑓 = 4 𝑢|𝑢| (2-145)
𝑅 ⁄3
where:
𝑛 = Mannings roughness factor
R = hydraulic radius.
Minor losses due to entrance configuration, gates, valves, and corners are accounted for in the minor loss
term Sm:
𝑢|𝑢| 1
𝑆𝑚 = 𝑘 (2-146)
2𝑔 𝐿
where:
k = sum of minor loss coefficients
L = length
Pressurized or full culvert flow is modeled assuming a fictitious water surface width called a Preissmann
slot (Yen, 1986). If the culvert is full, the surface width T is zero and the governing equations become sin-
gular. Using a Preissmann slot avoids having to switch between the open channel and pressurized flow
equations. The slot must be narrow enough to minimize error in the mass and momentum balance but
large enough to maintain numerical stability when solving the open channel St. Venant equations. A top
width of 0.5% of the diameter is assumed for culverts flowing full.
The advantages of using a Preissmann slot are (Yen, 1986):

AUXILIARY FUNCTIONS HYDRAULIC STRUCTURES
1. uses only Saint-Venant equations and avoids switching between the surcharge equation
and open-channel flow equations and avoids the associated separate treatment of the
boundary conditions
2. no need to define surcharge criteria
3. not necessary to keep inventory of the pipes that are surcharged at different times
4. permits the flow transition to progress computationally reach by reach in a sewer, as in the
open-channel case, and hence it can account for the situation when only part of the length
of the pipe is full
5. requires few additional assumptions than the standard approach to achieve numerical sta-
bility
6. simpler to program
The disadvantages are:
1. introduces a potential accuracy problem in the mass and momentum balance of the flow if
the slot is too wide, and stability problems if it is too narrow
2. requires computation of two equations (continuity and momentum) for each of the reaches
of the sewer when the sewer is full surcharged, whereas in the standard surcharge compu-
tation only one equation is applied to the entire length of the sewer
3. hypothetical rather than real
The Preissmann slot concept has been applied to other models for surcharged flow including the model
described by Abbot (1982) and SWMM EXTRAN (Roesner et al. 1988).
The boundary condition used for solving the governing equations is the head or water level at each end of
the culvert. However, if the water level at the downstream end of the culvert is less than the critical depth,
the critical depth is used. Momentum is not transferred between model segments and the culverts. Initial
conditions are the calculated velocities and heads at the previous time step.
The governing equations cannot be solved analytically and an implicit finite difference scheme is used to
approximate the solution. The solution method employs the “leap-frog scheme” which calculates the head
and velocity at alternating computational nodes (Anderson, et. al., 1984). The finite difference forms of the
continuity and momentum equations are:
ℎ𝑖𝑛+1 −ℎ𝑖𝑛 𝑛+1

ℎ𝑖+2 𝑛+1
−ℎ𝑖−2 𝑛
ℎ𝑖+2 𝑛
−ℎ𝑖−2 𝐴𝑛 𝑛+1 𝑛+1
𝑖 𝑢𝑖+1 −𝑢𝑖−1
0= + 𝜃𝑢𝑖𝑛 + (1 − 𝜃)𝑢𝑖𝑛 +𝜃
𝛥𝑡 2𝛥𝑥 2𝛥𝑥 𝑇𝑖𝑛 𝛥𝑥
𝐴𝑛𝑖
𝑛
𝑢𝑖+1 𝑛
−𝑢𝑖−1
+(1 − 𝜃) 𝑇𝑖𝑛 𝛥𝑥
(2-147)
𝑢𝑖𝑛+1 −𝑢𝑖𝑛
𝑛 𝑖+1 𝑖−1 𝑢𝑛+1 −𝑢𝑛+1
𝑛 𝑖+1 𝑖−1 𝑢𝑛 −𝑢𝑛 ℎ 𝑛+1 −ℎ𝑖𝑛+1
𝛥𝑡
0=
+ 𝜃𝑢𝑖+1 𝛥𝑥
+ (1 − 𝜃)𝑢𝑖+1 𝛥𝑥
+ 𝜃𝑔 𝑖+2 𝛥𝑥
𝑛
ℎ𝑖+2 − ℎ𝑖𝑛 𝑛2 𝑛+1 𝑛 𝑛2 𝑛 𝑛 |
+(1 − 𝜃)𝑔 + 𝜃𝑔 4 𝑢𝑖+1 𝑖+1 + (1 − 𝜃)𝑔 4 𝑢𝑖+1 |𝑢𝑖+1
|𝑢 |
𝛥𝑥 𝑅 ⁄3
𝑅 ⁄3
𝑘 𝑛+1 𝑛 𝑘 𝑛 𝑛 |
|𝑢 | |𝑢
+𝜃 𝑢𝑖+1 𝑖+1 + (1 − 𝜃) 𝑢𝑖+1 𝑖+1 (2-148)
𝐿 𝐿
where n refers to the time level and i references the spatial node (Figure 35).

ui-1 ui ui+1
hi-2 hi hi+2
Figure 35. Linkage schematic of model segments with a culvert.
Figure 36 compares flow predictions using the dynamic culvert model with flow data taken within a culvert
at NE 47th bridge in the Upper Columbia Slough, Portland, Oregon. Data was recorded using a flow meter
placed directly in a culvert. The cyclical flows are the result of turning pumps on and off at a downstream
pump station. The culvert was calibrated by adjusting the minor loss parameter.
data
dynamic flow algorithm
Flow Rate at NE 47th
29-Sep 1-Oct 3-Oct 5-Oct 7-Oct 9-Oct 11-Oct 13-Oct 15-Oct
30
0.8
25 0.7
0.6
20
Flow rate, ft3 s-1
Flow rate, m3 s-1

0.5
15
0.4
10 0.3
0.2
5
0.1
0 0.0
272 274 276 278 280 282 284 286 288
Julian Day
Figure 36. Computed versus observed flow using dynamic culvert model.
Internal Weirs
The model can be used to set internal weirs at specified cell locations. The user specifies the location of the
internal weir by providing a segment and layer number. The “weir” effectively acts as a barrier to flow and
diffusion of mass/heat across the width of the waterbody as shown in Figure 37. This can be used to simu-
late submerged and curtain weirs within a waterbody.

Layer # Segment # Dam
Skimmer weir
Outflow
Submerged weir
Figure 37. Schematic representation of internal weirs.
Water Level Control or Pumps
Many times, outflows in reservoirs are controlled by water levels. In order to facilitate management of the
water body, a water level control or pump algorithm was added to the code. Essentially, this is a pump
based on a float controller.
The algorithm allows the user to specify the upstream and downstream segment for water to be transferred
at a given flow rate based on the water level at the upstream segment. Reverse flow is not allowed. The
withdrawal is treated as a lateral selective withdrawal and the segment that receives the inflow is treated
as a tributary.
Outlet Structures
Outflows through hydraulic structures (Figure 38) can either be specified or computed by the model based
on user-supplied rating curves.

Tainter gate Spillway/weir
Figure 38. Radial gates and spillway flow.
The UNET model (HEC, 1997a), a one-dimensional unsteady hydraulic model, formally accounts for spillway
flow from weirs and spillways. For free or submerged flow from a spillway with a radial gate, UNET uses a
general equation of the form:
𝑄𝑠𝑝 = 𝐶𝑊𝐴𝛼 𝐵𝛽 𝐻 𝜂 (2-149)
where:
 = empirical coefficient
 = empirical coefficient
 = empirical coefficient
Qsp = flow rate, m3 s-1
A = trunnion height, m
B = gate opening, m
C = empirical coefficient
W = gate width, m
and
𝐻 = 𝑍𝑢 − 𝐾𝑍𝑑 − (1 − 𝐾)𝑍𝑠𝑝 (2-150)
where:
Zu = headwater elevation
Zd = tailwater elevation
K = 1 for submerged flow and 0 for free flow
Zsp = spillway elevation
This equation was developed based on rating curves for hydraulic control structures in Arizona. Submerg-
ence is defined as:

𝑍𝑑 −𝑍𝑠𝑝 2
𝑍𝑢 −𝑍𝑠𝑝
>3 (2-151)
Note that weir flow is assumed to occur whenever B = 0.8H and is computed as:
𝑄𝑤𝑒𝑖𝑟 = 𝐶𝑤 𝐹𝑊((1 − 𝐾)𝑍𝑢 + 𝐾𝑍𝑑 − 𝑍𝑠𝑝 )𝐻1/2 (2-152)
where:
Cw = weir coefficient
[𝑍𝑑 −𝑍𝑠𝑝 ]
𝐹 = 3 (1 − ) when K = 1
[𝑍𝑢−𝑍𝑠𝑝 ]
F = 1 when K = 0
For a concrete spillway, HEC (1997a) suggests using a weir coefficient value of 4. Note that the above two
equations are considered equivalent whenever B = 0.8H.
HEC-RAS, a one-dimensional, steady-state hydraulic model, (HEC, 1997b), includes the ability to model flow
over spillways including tainter and sluice gates, broad-crested weirs, and an ogee crest. Ineffective flow
area, that area below the weir or gate opening, is used to block a part of the channel until it reaches the
level of a spillway or weir.
A summary of the equations used by HEC-RAS (HEC, 1997b) as well as explanations are shown in Table 4.
Table 4. HEC-RAS flow rates through weirs and sluice gates.

Condition Equation Description
When the upstream water surface is 1.25 times the gate open-
ing height (above the spillway crest),
Q =flow, cfs
C =discharge coefficient (between 0.6 and 0.8)
W =gated spillway width, ft
T =trunnion height (from spillway crest to trunnion pivot
Radial flow
point), ft
gate, flowing 𝑄 = 𝐶√2𝑔𝑊𝑇 𝑇𝐸 𝐵 𝐵𝐸 𝐻𝐻𝐸
B = gate opening height, ft
freely
H = upstream energy head above spillway crest, Zu-Zsp, ft
Zu = upstream energy grade line elevation, ft
Zd = downstream water surface elevation, ft
Zsp = spillway crest elevation, ft
TE = empirical trunnion height exponent, 0.16
BE = gate opening coefficient, 0.72
HE = head exponent, 0.62
radial gate ing height (above the spillway crest), whenever the tailwater
flowing under depth divided by the energy depth above the spillway is greater
𝑄 = 3𝐶 √2𝑔𝑊𝑇 𝑇𝐸 𝐵 𝐵𝐸 𝐻𝐻𝐸
submerged than 0.67
conditions
H = Zu-Zd
ing height above the spillway crest
freely flowing
𝑄 = 𝐶 √2𝑔𝐻𝑊𝐵
sluice gate
H = upstream energy head above the spillway, Zu-Zsp
C = discharge coefficient, 0.5 to 0.7
ing height above the spillway crest, whenever the tailwater
submerged
𝑄 = 3𝐶√2𝑔𝐻𝑊𝐵 depth divided by the energy depth above the spillway is greater
sluice gate
than 0.67

Condition Equation Description

H = Zu-Zd
When upstream water level is equal to or less than the top of
the gate opening, weir flow equation is used
C = weir coefficient, 2.6-4.0 depending on broad crested

Low flow
3 or Ogee spillway and length of spillway crest
through gated 𝑄= 𝐶𝐿𝐻 2 H = upstream energy head above spillway crest, for an
structure
Ogee spillway the value of C is adjusted according to a
1977 Bureau of Reclamation study on variability of C
for Ogee spillways, suggested values of C are 2.6 for
bridge decks and 3.0 for flow over elevated roadways
Spillways/Weirs
Analysis of flow over weirs has been studied extensively. Martin and McCutcheon (1999) show that a typical
relationship between the pool depth and flow over a weir is:
𝜂
𝑄 = 𝐶𝑒 𝑊𝑐 ℎ𝑤 (2-153)
where Ce and  are empirical coefficients, Wc is the length of the weir crest, and hw is the height of the pool
above the weir crest. Theoretical calculations of steady-state flow over a weir can be complex depending
on whether the weirs are sharp-crested, broad-crested, V-notched, rectangular, Cipolletti, parabolic, or
some other type. Table 5 shows some examples from French (1985) and USBR (2001) on typical equations
used for the different weir types. For many regular weir types, formulae exist for accurate estimation of
the flow. However, in most cases a rating curve for a given installation is necessary because of the uncer-
tainty of end effects, flow alignments, shallowness in the upstream pool, and other unique features of the
installation (Martin and McCutcheon, 1999).

Table 5. List of weir types (French, 1985; USBR, 2001)

Weir type Weir Equation Description
Valid when 0.08 <H/L<0.5
Q = flow rate
CD = discharge coefficient (0.84 to 1.06)
Cv = velocity coefficient accounting for neglecting the
Rectangular
2 2 3/2 velocity head in the approach channel (between
broad crested 𝑄 = 𝐶𝐷 𝐶𝑣 √ 𝑔𝑊𝐻
3 3 1.0 and 1.2)
weir
W = width at surface
H = upstream head above spillway crest
Zu= upstream energy grade line elevation, ft
g = gravity acceleration
Where B = width at bottom of weir crest
Olson and Wright (1990) show that Ce depends on the ap-
proach velocity head, V2/2g, and the contraction of stream-
lines just beyond the weir crest and show that
Ce = 0.611+0.075*(H/Z)
H = weir head, ft
Rectangular,
2 3/2 Z = weir crest head measured from the channel bot-
sharp crested 𝑄 = 𝐶𝑒 √2𝑔𝐵𝐻
3 tom, ft
weir
Clay (1995) suggests a simple equation of Q=3.33BH3/2 of
this form when approach velocities are less than 1 fps or
Q=3.33B[(H+h)3/2-hv3/2]
hv = V2/2g
V = approach velocity, ft s-1
Parabolic, broad- 3 f = distance from the bottom point of the weir to the
crested 𝑄 = 𝐶𝐷 𝐶𝑣 √ 𝑓𝑔𝐻 2 weir focal point, ft
4
Parabolic, sharp- 1
crested 𝑄 = 𝐶𝑒 𝜋√𝑓𝑔𝐻2 Ce = effective discharge coefficient
2
Triangular, 16 2
𝑄 = 𝐶𝐷 𝐶𝑣 √ 𝑔 𝑡𝑎𝑛( 0.5𝛩)𝐻 5/2  = half angle of the triangular notch
broad-crested 25 5
Triangular, sharp- 8 5/2 Ce = function of notch angle and varies from 0.59 to 0.57
crested 𝑄 = 𝐶𝑒 √2𝑔 𝑡𝑎𝑛( 0.5𝛩)𝐻 for angles between 20 and 100 degrees
25
W = top width of trapezoidal weir
Trapezoidal, m = slope of trapezoidal weir
𝑄 = 𝐶𝐷 (𝑊𝑦𝑐 + 𝑚𝑦𝑐2 )[2𝑔(𝐻 − 𝑦𝑐 )]1/2
broad-crested yc = depth of water at the weir
H = energy head upstream of weir
B= bottom width of trapezoidal weir
Trapezoidal, 2 4
𝑄 = 𝐶𝑒 √2𝑔(𝐵 + 𝐻 𝑡𝑎𝑛 0 . 5𝛩)𝐻 1/2
= angle of the trapezoid at a convergence point of the 2
sharp-crested 3 5 sides
Use when H > 1.25Hb, otherwise use equation for broad
crested triangular weir
Truncated trian-
2 2
gular, broad- 𝑄 = 𝐶𝐷 𝐶𝑣 √ 𝑔𝑊(𝐻 − 0.5𝐻𝑏 )3/2
3 3 Hb = depth from the bottom of the truncated triangular
crested
weir to the top of the triangle and the beginning of
the rectangular section
Truncated trian- 4 𝑊 Use when H > Hb, otherwise use equation for sharp crested
gular, sharp- 𝑄 = 𝐶𝑒 √2𝑔 (𝐻 2.5 − (𝐻 − 𝐻𝑏 )2.5 )
15 𝐻𝑏 triangular weir
crested
A modification of the contracted, rectangular, sharp-crested
weir with a trapezoidal control section and sides sloping
outward with slopes of 4:1
2 3/2 W = top width of weir
Cipoletti 𝑄 = 𝐶𝐷 𝐶𝑣 √2𝑔𝑊𝐻
3 CD0.63
Cv varies from 1 to 1.2 and is a function of CD and the ratio
of area upstream of the control section and at the control
section

Weir type Weir Equation Description
B = bottom width of weir

Proportional or 1 a = distance from weir bottom to top of the rectangular
𝑄 = 𝐶𝐷 𝐵√2𝑔𝑎 (𝐻 − 𝑎)
sutro weir 3 weir section
CD = drag coefficient, 0.597-0.619 for symmetrical Sutro
weir, 0.625- 0.603 for unsymmetrical Sutro weir
Since all weirs in practice are calibrated and a head discharge relationship is usually determined, the flow
versus head relationship is used rather than an equation from Table 5. The user must then analyze the weir
or spillway and input a relationship based on the weir or spillway geometry. The model accepts equations
in the form of a power function for freely flowing conditions:
𝑄 = 𝛼1 𝛥ℎ𝛽1 (A-2-154)
where:
1 = empirical parameter
1 = empirical parameter
h = Zu-Zsp
Zu = upstream head
Zsp = spillway crest elevation
and for submerged conditions
𝑄 = 𝛼2 𝛥ℎ𝛽2 (A-2-155)
where:
2 = empirical parameter
2 = empirical parameter
h = Zu-Zd
Zu = upstream head
Zd = downstream head
Submerged conditions are defined when the tailwater depth over the upstream energy head (static head
and velocity head) is greater than 0.67 (HEC, 1997b). Even though negative flow rates are possible using
the second equation whenever Zd > Zu, these results should be used with caution since rarely are rating
curves done for reverse flow over a spillway. The user needs to ensure there is a smooth transition between
submerged flow conditions and free flowing conditions by proper choice of model coefficients. The follow-
ing discussion shows how to generate a smooth flow transition from free flowing to submerged flow con-
ditions (Figure 39).

Free flow condition

Head difference from upstream to tailwater, m
Submerged flow condition
Reverse flow when tailwater elevation

greater than upstream elevation
-2
-3 -2 -1 0 1 2 3 4 5 6 7
. Flow rate, m3 s-1
Figure 39. Flow rate over a spillway or weir for submerged and free flowing conditions.
Consider the following weir flow condition in Figure 40.
H1
H2
Hweir
Figure 40. Flow at a submerged weir.
In order to have a smooth transition, the two flows must be equal at the transition point. Using
𝑢2
𝐻2 − 𝐻𝑤𝑒𝑖𝑟 = 0.67(𝛥ℎ1 + 2𝑔) ≈ 0.67(𝐻1 − 𝐻𝑤𝑒𝑖𝑟 ) (2-156)
and substituting the rating curves for free flow and submerged conditions and solving for 2 results in:

𝛼1 (𝐻1 −𝐻𝑤𝑒𝑖𝑟 )𝛽1−𝛽2

𝛼2 = (0.33)𝛽2
(2-157)
In many cases a weir can be set as the downstream boundary condition of a river. In CE-QUAL-W2, the user
can specify the weir crest as the channel bottom elevation, such that the weir equation is of the form:
𝑄 = 𝛼1 𝐻𝛽1 (2-158)
where H is the depth of the water at the weir. Setting this in the form of a stage-discharge relationship:
1
𝑄 𝛽1
𝐻 = 𝑎𝑄 𝑏 = (𝛼 ) (2-159)
1
where a and b are empirical coefficients. Writing this equation in a form compatible with the stage discharge
relationship using the a and b coefficients,
1 1
1 𝛽
𝐻= ( ) 1 𝑄 𝛽1 (2-160)
𝛼1
Then the a and b coefficients become

1
1 𝛽
𝑎= ( ) 1 (2-161)
𝛼1
1
𝑏=𝛽 (2-162)
1
or vice versa:
𝛼1 = 𝑎−𝛽1 (2-163)
1
𝛽1 = 𝑏 (2-164)
allows the user to set a weir-rating curve that reproduces the stage discharge relationship at a downstream
boundary.
A dynamic weir or spillway crest elevation can also be used to simulate dynamic raising of flashboards.
These are designated in the model as gates.
Gates
For a gated structure or sluice gate, a more complex rating curve is required based on the opening and the
head difference between the upstream and downstream condition (the spillway crest if free flow and the
tailwater elevation if submerged flow).
The following equation is used for freely flowing conditions:
𝑄 = 𝛼1 𝛥ℎ𝛽1 𝐵𝛾1 (2-165)

where:
1 = empirical coefficient
1 = empirical coefficient
1 = empirical coefficient
h = Zu - Zsp
Zu = upstream head
Zsp = spillway crest elevation
and the following equation is used for submerged flow:
𝑄 = 𝛼2 𝛥ℎ𝛽2 𝐵𝛾2 (2-166)
where:
2 = empirical coefficient
2 = empirical coefficient
2 = empirical coefficient
h = Zu - Zd
Zd = downstream head
B = gate opening, m
In defining these parameters, the user also has to generate a time series file with the gate opening in m
where a gate opening of 0 m is closed. Whenever the gate opening is equal to or greater than 0.8 h, a
weir equation is used with no functional dependency on the gate opening. In this case, a rating curve must
be supplied when the gate acts like a weir. Figure 41 shows the flow rate dependence on the gate opening.
4
Gate:Q=2B0.72h0.62
Weir:Q=1.75h1.5
3
Free flow condition
Gate opening, BGATE, m
weir flow
Gate flow equation
0 0.5 1 1.5 2
Flow rate, m3 s-1
Figure 41. Flow rate variation with gate opening.
In some reservoir systems, an outlet valve is connected to the reservoir and a head-discharge relationship
is used based on the gate opening or number of gate turns. In this case, the outlet level is usually at a

different elevation than the withdrawal elevation. The above gate formulation can still be used if no reverse
flow occurs through the needle valve. This situation is illustrated in Figure 42. In this case, the elevation of
the outflow is required in addition to the elevation at which the outflow is taken if a rating curve is used in
the model. This use is described in the section on gates in the control file.
Figure 42. Selective withdrawal with outflow connected to a valve with a gate.
The user can insert weirs and/or spillways, specify connectivity to other model segments, and insert the
ratings curve parameters for each weir/spillway. The model treats each spillway, weir, or gate as a selective
withdrawal outflow and uses the selective withdrawal algorithm for determining water flow from each ver-
tical layer adjacent to the structure. Inflows from hydraulic control structures are treated as tributary in-
flows where the user must specify whether the inflow is placed according to density, equally distributed
between all vertical layers, or distributed between a given elevation range.
Weir equations are used when gates are open and the open gate does not interfere with the flow (when B
≥ 0.8Δh).
Gates can also be used to move flows around to different parts of the model domain – different branches.
In that case the specification of gate openings is interpreted as specification of flow rates.
Branch Momentum Exchange

Starting with CE-QUAL-W2 Version 3, CE-QUAL-W2 conserves longitudinal momentum at branch intersec-
tions (Figure 43). The vector component of velocity in the x-direction of the main channel, Ux, can be com-
puted from the channel orientations. The x-direction component is Ux=Ucos where U is the longitudinal
velocity of the tributary at the downstream segment that intersects the main branch and  is the difference
in the angle between the main stem and tributary segments.

AUXILIARY FUNCTIONS LATERAL INFLOWS
main
x
U Uy
x
Ux
trib
Figure 43. Schematic of branch connection.
The conservation of momentum about a control volume, the main stem segment, would result in an addi-
tional source of momentum. Lai (1986) shows that the correction to the x-momentum equation would be:
𝑞𝐵𝑈𝑥 (2-167)
where:
q = lateral inflow per unit length
This arises from re-deriving the momentum equations and assuming that all the fluid entering the segment
is moving at the velocity Ux. The correction to the x-momentum equation is:
 UB  UUB  WUB  g cos B 

z
t
+
x
+
z
x
−
  x dz

(A-2-168)
1 B xx 1  B  x z
+ + + qBU x
 x  z 
side tributary
momentum
Lateral Inflows
Wells (1997) proposed accounting for the cross-shear as a result of the y component of the velocity of a
side branch in the computation of the vertical eddy viscosity. This was implemented by increasing the cross-
shear velocity gradient. In Version 2, wind shear across the lateral axis of a segment also increased the
vertical mixing by affecting the computation of Az. Analogous to wind shear, an additional side shear is
included in the calculation of the vertical eddy viscosity:
66 Appendix A Hydrodynamics and Transport

BRANCH MOMENTUM EXCHANGE AUXILIARY FUNCTIONS
2
 l 2    U    wy e-2 kz +  trib  (-CRi )
2
Az =      +   e (2-169)
2  z    Az 
where:
𝑓𝑖 2
𝜏𝑡𝑟𝑖𝑏 ≅ 𝜌 𝑈
8 𝑦
fi = is an interfacial friction factor,  0.01
∑ 𝑄𝑖𝑛𝑦
𝑈𝑦 = 𝛥𝑧𝛥𝑥
∑ 𝑄𝑖𝑛𝑦 = [𝑈𝑏𝑟 𝛥𝑧𝐵] + [∑ 𝑄𝑡𝑟𝑖𝑏 ]
Ubr = Ubrsin(main-branch)
z = inflow cell layer height
B = inflow cell width
x = inflow cell segment length
Qtrib = tributary flow rate assumed to be at right angles to the main channel
This side shear effect is only computed when the vertical mixing algorithm chosen by the user is W2 or
W2N.

AUXILIARY FUNCTIONS
Density
Accurate hydrodynamic calculations require accurate water densities. Water densities are affected by var-
iations in temperature and solids concentrations given by:
 = T +   S (2-170)
where:
 = density, kg m-3
T = water density as a function of temperature, kg m-3
S = density increment due to solids, kg m-3
A variety of formulations has been proposed to describe water density variations due to temperatures. The
following relationship is used in the model (Gill, 1982):
 T = 999.842594 + 6.793952 x 10-2 Tw

w
− 9.095290x10-3 T2w + 1.001685 x 10-4 T3w (2-171)

− 1.120083x 10-6 T4w + 6.536332 x 10-9 T5w
Suspended and dissolved solids also affect density. For most applications, dissolved solids will be in the
form of total dissolved solids. For estuarine applications, salinity should be specified. The effect of dissolved
solids on density is calculated using either of these variables with the choice specified by the variable
[WTYPE]. Density effects due to total dissolved solids are given by Ford and Johnson (1983):
  TDS = (8.221 x 10 -4 - 3.87 x 10 -6 T w + 4.99 x 10 -8 T 2w ) TDS (A-2-172)
where:
TDS = TDS concentration, g m-3
and for salinity (Gill, 1982):
  sal = (0.824493 - 4.0899 x 10-3 T w + 7.6438 x 10-5 T 2w

- 8.2467 x 10-7 T 3w + 5.3875 x 10- 9 T 4w )  sal
(2-173)
+ (-5.72466 x 10-3 + 1.0227 x 10-4 T w
- 1.6546 x 10-6 T 2w ) 1.5
sal + 4.8314 x 10  sal
-4 2
where:
sal = salinity, kg m-3
The suspended solids effects are given by Ford and Johnson (1983):
 1 
  ss =  ss  1 -  x 10 -3 (2-174)
 SG 

SELECTIVE WITHDRAWAL AUXILIARY FUNCTIONS
where:
ss = suspended solids concentration, g m-3

SG = specific gravity of suspended solids
Assuming a specific gravity of 2.65, the above relationship is simplified to:
  ss = 0.00062  ss (2-175)
The total effect of solids is then:
  s = (   sal or   tds ) +   ss (2-176)
Selective Withdrawal
Selective withdrawal is applied to all structure outflows where layer locations and outflows at each layer
are calculated based on the total outflow [QOUT], structure type [SINKC], elevation [ESTR], and computed
upstream density gradients. The selective withdrawal computation uses these values to compute vertical
withdrawal zone limits and outflows. It also sums the outflows for multiple structures. In addition, all out-
flows – such as withdrawals and hydraulic structures – use selective withdrawal theory to compute the
withdrawal envelope.
Outflow distribution is calculated in the subroutine SELECTIVE_WITHDRAWAL. This routine first calculates
limits of withdrawal based on either a user specified point or line sink approximation for outlet geometry
[SINKC]. The empirical expression for point sink withdrawal limits is:
d = ( cbi Q/N )0.3333 (2-177)
and for a line sink:
d = ( cbi 2 q/N )0.5 (2-178)
where:
d = withdrawal zone half height, m
Q = total outflow, m3 s-1
N = internal buoyancy frequency, Hz
q = outflow per unit width, m2 s-1
cbi = boundary interference coefficient
The width is the outlet width. The point sink approximation assumes approach flow is radial both longitu-
dinally and vertically while the line sink approximation assumes flow approaches the outlet radially in the
vertical. The boundary interference coefficient is two near a physical boundary and one elsewhere.
Velocities are determined using a quadratic shape function:

AUXILIARY FUNCTIONS
2
(  k -  o ) 
Vk =1 -   (2-179)
( l - o ) 
where:
Vk = normalized velocity in layer k
k = density in layer k, kg m-3
o = density in the outlet layer, kg m-3
l = density of the withdrawal limit layer, kg m-3
The shape function generates a maximum velocity at the outlet level with velocities approaching zero at
withdrawal limits. During non-stratified periods, outflow from top to bottom is uniform. Uniform flows
also result from large outflows during periods of mild stratification. As stratification develops, withdrawal
limits decrease and outflow is weighted towards the outlet elevation.
Withdrawal limits can be varied by specifying a line sink and changing the effective width. Small outlet
widths result in nearly uniform outflows, while large widths limit outflows to the outlet layer.
Sediment Resuspension
This algorithm is based on work from Kang et al. (1982) where the bottom shear stress is computed based
on wind speed, wind fetch and depth. The wind blowing across a water surface creates wind waves that
have orbital motion that decays with depth. The model user inputs a critical shear stress for detachment
of the particles. If the critical shear stress is exceeded, then particles are resuspended. The approach of
Kang et al. (1982) consists of the following steps:
1. Computation of the wave height, Hs in m,
𝑔𝐹 0.42
𝑊2 𝑔𝐻 0.75 0.0125( 2)
𝑊
𝐻𝑠 = 𝑔
0.283 𝑡𝑎𝑛ℎ [0.53 ( 𝑊2
) ] 𝑡𝑎𝑛ℎ [ 𝑔𝐻 0.75
] (2-180)
𝑡𝑎𝑛ℎ[0.53( ) ]
𝑊2
where W is the wind velocity (m/s), F is the fetch (m), H is the mean depth (m)
2. Computation of wave period, Ts in s,
𝑔𝐹 0.25
2𝜋𝑊 𝑔𝐻 0.375 0.077( 2)
𝑊
𝑇𝑠 = 𝑔
1.2 𝑡𝑎𝑛ℎ [0.833 (𝑊2
) ] 𝑡𝑎𝑛ℎ [ 𝑔𝐻 0.375
] (2-181)
𝑡𝑎𝑛ℎ[0.833( ) ]
𝑊2
3. Computation of the wavelength, L in m, iteratively from the following equation:
𝑔𝑇𝑠2 2𝜋𝐻
𝐿= 2𝜋
𝑡𝑎𝑛ℎ [ 𝐿 ] (2-182)
4. Computation of the orbital velocity, 𝑈̄in cm/s,
𝜋𝐻 100
𝑈̄ = 𝑇 𝑠 𝑠𝑖𝑛ℎ(2𝜋𝐻/𝐿) (2-183)
𝑠
70 Appendix A Hydrodynamics and Transport

AUXILARY FUNCTIONS PARTICLE TRANSPORT
5. Computation of bottom shear stress,  in dynes/cm2,
𝜏 = 0.003𝑈̄ 2 (2-184)
6. Computation of actual bottom scour rate of suspended solids,  in mass of sediments scoured
per area or g/m2,
𝜀=0 𝜏 ≤ 𝜏𝑐 (2-185)
𝛼𝑜
𝜀= (𝜏 − 𝜏𝑐 )3 𝜏 > 𝜏𝑐 (2-186)
𝑡𝑑2
where o is an empirical constant=0.008, td=7, and c is the user-defined critical shear stress in dynes/cm2.
Resuspension supposedly only occurs during the first hour of the wind shear greater than the critical
shear stress. Hence, the rate of resuspension per time, E in g/m2/hour, would be
𝜀
𝐸= for the first hour and nothing after that. The resulting concentration of suspended solids in the
1ℎ𝑟
entire water column, c in mg/l, if distributed evenly over the entire volume would then be
10000𝜀𝐴𝑏𝑜𝑡𝑡𝑜𝑚 10000𝜀
𝑐= = .
𝑉 𝐻
where Abottom is the surface area of the bottom and V is the volume of the water column above the bot-
tom (=HAbottom). Chapra (1997) also uses this approach and provides an example calculation.
3. Particle Tracking Algorithm

The original particle tracking algorithm in CE-QUAL-W2 was developed by Goodwin et al. (2001) for use in
a smart-particle tracking algorithm predicting movement of fish in a southern reservoir in the USA. A de-
scription from Goodwin et al. (2001) of the algorithm is reproduced below:
The particle-tracking algorithm used was developed by Chapman et al. (1994) to predict transport
and fate of floatables and suspended and dissolved materials in three dimensions. The algorithm
was simplified for use in CE-QUAL-W2 by deleting portions of the model relating to surface
transport of floatables and suspended and dissolved materials. In addition, transport equations
for lateral movement were deleted because CE-QUAL-W2 computes forcing functions only in the
longitudinal and vertical directions. The following equations are used for particle tracking in CE-
QUAL-W2 based on velocity:
Xt+1 = Xt + Ut(Zt,Xt) * t (3-1)

Zt+1 = Zt + Wt(Zt,Xt) * t (3-2)
where
Xt+1 = longitudinal position of particle at time t+1 (m); defined as
xNode(k,i) + x at time t+1,
Zt+1 = vertical position of particle at time t+1 (m); defined as
zNode(k,i) + z at time t+1,
71
Node(k,i) = node above, upstream, and closest to particle’s position defined as the
upper-left node (Figure 1) of the rectangular cell formed by
intersection of vertical layer k and longitudinal segment i,
xNode(k,i) = longitudinal position of Node(k,i) (m),
x = offset (downstream) from Node(k,i) to particle’s position (m),
zNode(k,i) = vertical position of Node(k,i) (m),
z = offset (below) from Node(k,i) to particle’s position (m),
Xt = longitudinal position of particle at time t (m); defined as x Node(k,i) +
x at time t,
Zt = vertical position of particle at time t (m); defined as
zNode(k,i) + z at time t,
Ut(Zt,Xt) = interpolated horizontal flow velocity (m/s) at location (Zt,Xt),
Wt(Zt,Xt) = interpolated vertical flow velocity (m/s) at location (Zt,Xt), and
t = time interval between time t and time t+1 (seconds).
Figure 44. From Goodwin et al. (2001) illustrating the particle transport through the CE-QUAL-W2 grid.
After Xt+1 and Zt+1 are calculated, boundary checks are performed to determine if the particle’s
new position exceeds either the longitudinal or vertical boundaries of the cell containing Xt and Zt
or any system boundary. An algorithm is activated to determine the appropriate cell in which to
place the particle if the particle’s new position exceeds the boundaries of its cell from the previ-
ous time step. Once the appropriate cell is found, k, i, Nodek,i, xNode(k,i), zNode(k,i), x, and z are up-
dated for the particle at t+1 and stored. Since the displacement of a particle is proportional to
the time step used, it’s possible the displacement calculated in equations (1) and/or (2) could po-
sition the particle outside system boundaries. To correct for this, the model places a particle
near the boundary at a user-specified distance if Xt+1 or Zt+1 exceeds the system boundaries.
In addition, a random movement of the particle is added to the velocity movement. The randomness
added to the particle movement is based on the intensity of turbulence in a system. Following the work of
Dunsbergen and Stalling (1993) and Chapman et al. (1994), the transport in x and z including the transla-
tion by velocity and random turbulent transport are shown below:
𝑋 𝑡+1 = 𝑋 𝑡 + 𝑈∆𝑡 + 𝑅√6𝐷𝑥 ∆𝑡 (3-3)

𝑍 𝑡+1 = 𝑍 𝑡 + 𝑊∆𝑡 + 𝑅√6𝐷𝑧 ∆𝑡 (3-4)
Particle Transport 72
AUXILARY FUNCTIONS PARTICLE TRANSPORT
where R represents a random number having a uniform distribution with zero mean and unit variance, Dx
and Dz are the longitudinal and vertical mass/heat dispersion coefficients. The use of the uniform rather
than a Gaussian distribution saves computational time and does not affect the resulting random distribu-
tion (Hathhorn, 1997).
The value of Dx is specified in the control file, w2_con.npt, whereas the value of Dz is computed based on
the turbulence closure model used, such as the k turbulence model (for example in w2_con.npt the user
can set AZC=TKE). The choice of Dx can lead to high values of random movement in x if Dx is high. In all
cases the maximum movement of a particle is constrained by the grid cell height, z, and segment length,
x.
Also the particles move in the lateral direction based on the following equation assuming that the lateral
dispersion coefficient is the same as the longitudinal coefficient:
𝑌 𝑡+1 = 𝑌 𝑡 + 𝑉∆𝑡 + 𝑅√6𝐷𝑥 ∆𝑡 (3-5)
The value of V is based on any lateral inflows or outflows at a segment.
The location of the particles in the grid are identified by its segment (I) and layer (K) location as well as X,
Y, and Z coordinates within each layer. Figure 45 and Figure 46 show the x, y, and z coordinate reference
locations within the cells for the particle algorithm for the plan view and side view, respectively. Figure 47
shows the definition sketch for lateral velocities in a layer. The standard CE-QUAL-W2 designation for lon-
gitudinal (positive downstream or increasing segment number) and vertical velocities (positive downward)
in followed in this algorithm. Initial x, y, and z are given for each particle released into a segment and
layer. The user specifies x and z, but y is chosen as the middle of a layer, i.e., y=B/2 where B is the layer
width.
Figure 45. Plan view of x and y coordinates within a layer (K) and segment (I). DLX is the segment length
and B is the segment width at the given layer K.
73
Figure 46. Side view of z and y coordinates within a layer (K) and segment (I). DZ is the layer thickness
and B is the layer width.
Figure 47. Definition sketch for lateral and longitudinal velocities within the cells. Withdrawals are al-
ways assumed to be on the LHS (left hand side) of the segment.
Particle Transport 74
4. Temperature and Water Quality
Modeling
Authors and contributors: Scott A. Wells, Tom M. Cole, Chris Berger, Stewart Rounds, Annette
Sullivan, S. Prakash, J. A. Vandenberg, E. M. Buchak, Lillian Jeznach, John Tobiason, Zhong Zhang
This section of the User’s Manual documents the governing equation for constituent mass and energy
(temperature) and the source/sink terms for temperature and each water quality state variable. There is
also a section on the computation of vertical and longitudinal dispersion coefficients.
Turbulent Advection-Diffusion Equation

As in the momentum equation, time-averaged variables for velocity are introduced (Figure 48) and concen-
tration (Figure 49).
u
u
time
t t+T
Figure 48. Velocity variability with time.
A75
THEORY HYDRODYNAMICS
c
c
time
t t+T
Figure 49. Concentration variability with time.
The instantaneous velocity and concentration are decomposed into a mean and an unsteady component:
1 𝑡+𝑇
𝑇
∫𝑡 𝑢(𝑡) 𝑑𝑡 (4-1)
Similarly for w, v, and c:
𝑣 = 𝑣 + 𝑣′
𝑤 = 𝑤 + 𝑤′
𝑐 = 𝑐 + 𝑐′ (4-2)
Substituting these into the 3D governing equation and time averaging:
c c c c  2 c 2 c 2 c 
+u +v +w = D  2 + 2 + 2
 t x  yz  
x y z
 
transport by mean advection molecular diffusive transport
(4-3)
−


x
(
uc -


)
y
vc -


( )
z
wc + S ( )
  
turbulent mass tranport
The new terms in the governing equation represent mass transport by turbulent eddies. As the intensity of
turbulence increases, turbulent mass transport increases. Notice also that all velocities and concentrations
are time averaged. The following turbulent mass fluxes are defined as:
𝐽𝑡 = ( 𝑢′ 𝑐 ′ , 𝑣 ′ 𝑐 ′ , 𝑤 ′ 𝑐 ′ ) (4-4)
where:

HYDRODYNAMICS THEORY
( uc ) = - E c
x
x
( )
vc = - E y
c
y
( wc ) = - E z
c
z
Substituting into the above equation:
c c c c   c    c 
+u +v +w =  ( Ex + D ) +  ( Ey+ D ) 
t x y z x  x   y  y 
(4-5)
  c 
+  ( Ez + D ) +S
z  z 
In turbulent fluids, Ex, Ey, and Ez >> D, and D can be neglected except at interfaces where turbulence goes
to zero. The turbulent diffusion coefficients can be thought of as the product of the velocity scale of turbu-
lence and the length scale of that turbulence. These coefficients are related to the turbulent eddy viscosity.
One is turbulent mass transport, the other is turbulent momentum transport between adjacent control
volumes. In general, these turbulent diffusion coefficients are non-isotropic and non-homogeneous.

AUXILIARY FUNCTIONS HEAT EXCHANGE
Temperature and Water Quality Transport

Spatial averages across the lateral dimension of the channel of the turbulent time-averaged quantities can
now be introduced:
𝑐 = 𝑐 + 𝑐”
𝑢 = 𝑢 + 𝑢” (4-6)
𝑤 = 𝑤 + 𝑤"
where the double overbar is a spatial average over y and the double prime is the deviation from the spatial
mean as illustrated in Figure 50 for velocity and Figure 51 for constituent concentrations.
y=y2
u
y
u
x
y=y1
Figure 50. Lateral average of the velocity field.
y=y2
c y
c
x
y=y1
Figure 51. Lateral average of the concentration field.
These are substituted into the governing equation and then the governing equation is integrated over the
width such that:

HEAT EXCHANGE AUXILIARY FUNCTIONS
mass transfer at side boundaries

𝜕𝐵𝑐̄̄ 𝜕𝐵𝑢̄̄ 𝑐̄̄ 𝜕𝐵𝑤 ̄̄ 𝑐̄̄ ⏞̄ ̄ |
𝜕𝑡
+ 𝜕𝑥 + 𝜕𝑧 = −𝐵 (𝑣̄ 𝑐̄ 𝑦2 − 𝑣̄̄ 𝑐̄̄ |𝑦1 + 𝑐′𝑣′|𝑦2 − 𝑐′𝑣′|𝑦1)
𝜕 𝜕𝑐̄̄ 𝜕 𝜕𝑐̄̄ 𝜕𝐵𝑢"𝑐" 𝜕𝐵𝑤 ″ 𝑐 ″
+ 𝜕𝑥 ((𝐷 + 𝐸𝑥 )𝐵 𝜕𝑥) + 𝜕𝑧
((𝐷 + 𝐸𝑧 )𝐵 𝜕𝑧) − [ 𝜕𝑥 + 𝜕𝑧 ] + 𝑟𝐵 (4-7)
Note how the following terms are simplified:
1 𝑦2 𝜕(𝑐̄̄ + 𝑐 ″ ) 1 𝑦2 𝜕(𝑐̄̄ ) 1 𝑦2 𝜕(𝑐 ″ )

∫ 𝑑𝑦 = ∫ 𝑑𝑦 + ∫ 𝑑𝑦
𝐵 𝑦1 𝜕𝑡 𝐵 𝑦1 𝜕𝑡 𝐵 𝑦1 𝜕𝑡
1 𝜕 𝑦2 1 𝜕 𝑦2 1 𝜕𝐵𝑐̄ ̄
= 𝐵 𝜕𝑡 ∫𝑦1 𝑐̄̄ 𝑑𝑦 + 𝐵 𝜕𝑡 ∫𝑦1 𝑐 ″ 𝑑𝑦 = 𝐵 𝜕𝑡 (4-8)
1 𝑦2 𝜕(𝑢̄̄ + 𝑢″ )(𝑐̄̄ + 𝑐 ″ ) 1 𝑦2 𝜕(𝑐̄̄ 𝑢̄̄ ) 1 𝑦2 𝜕(𝑐 ″ 𝑢″ )

∫ 𝑑𝑦 = ∫ 𝑑𝑦 + ∫ 𝑑𝑦
𝐵 𝑦1 𝜕𝑥 𝐵 𝑦1 𝜕𝑡 𝐵 𝑦1 𝜕𝑡
1 𝜕 𝑦2 1 𝜕 𝑦2 ″ ″ 1 𝜕𝐵𝑢̄̄ 𝑐̄̄ 1 𝜕𝐵𝑢″ 𝑐 ″
= ∫ 𝑐̄̄ 𝑢̄̄
𝐵 𝜕𝑥 𝑦1
𝑑𝑦 + ∫ 𝑐 𝑢 𝑑𝑦
𝐵 𝜕𝑥 𝑦1
=
𝐵 𝜕𝑡
+
𝐵 𝜕𝑥
(4-9)
The spatial average of any double primed variable goes to zero by definition. The turbulent dispersion co-
efficients are defined as:
c
u" c" = − D x
x (4-10)
c
w" c" = − D z
z
The dispersion terms are a result of lateral averaging of the velocity field. In general, except at an interface,
Dx >> Ex >> D and similarly for Dz >> Ez >> D. Substituting in for the dispersion coefficients and using q to be
the net mass transport from lateral boundaries, this equation becomes:
𝜕𝐵𝑐̿ ̿ 𝑐̿
𝜕𝐵𝑢 ̿ 𝑐̿
𝜕𝐵𝑤 𝜕 𝜕𝑐̿ 𝜕 𝜕𝑐̿
𝜕𝑡
+ 𝜕𝑥 + 𝜕𝑧 = 𝑞𝐵 + 𝜕𝑥 (𝐷𝑥 𝐵 𝜕𝑥) + 𝜕𝑧 (𝐷𝑧 𝐵 𝜕𝑧) + 𝑟𝐵 (4-11)
If the overbars are dropped and replaced with capitals, c is replaced with , then the following equation is
obtained:
𝜕𝐵𝛷 𝜕𝑈𝐵𝛷 𝜕𝑊𝐵𝛷 𝜕 𝜕𝛷 𝜕 𝜕𝛷

𝜕𝑡
+ 𝜕𝑥
+ 𝜕𝑧
− 𝜕𝑥 [𝐵𝐷𝑥 𝜕𝑥 ] − 𝜕𝑧 [𝐵𝐷𝑧 𝜕𝑧 ] = 𝑞𝛷 𝐵 + 𝑆𝛷 𝐵 (4-12)
where:
 = laterally averaged constituent concentration, g m-3
Dx = longitudinal temperature and constituent dispersion coefficient, m2 sec-1
Dz = vertical temperature and constituent dispersion coefficient, m2 sec-1
q = lateral inflow or outflow mass flow rate of constituent per unit volume, g m-3 sec-1
S = laterally averaged source/sink term, g m-3 sec-1
Note that this can be concentration or temperature since the concentration of heat can be determined to
be cpT where  is the fluid density, cp is the specific heat of water, and T is the temperature.

The following must be known in order to solve the equation:
1. laterally-averaged velocity field the from momentum equations

2. appropriate boundary and initial conditions
3. Dx and Dz
4. laterally-averaged source/sink terms
Determination of Dz and Dx
The specification of Dx, the longitudinal dispersion coefficient, is a user-defined input. The prepro-
cessor though does a check on the magnitude of this value by computing D x based on an approach
by Okubo (1971):
𝐷𝑥 = 5.84𝑋10−4 ∆𝑥 1.1
where Dx is in m2/s and the longitudinal grid spacing x is in m. Dx can also be set to be a function of the
mean velocity or this value can be space and time invariant. Dx can be computed dynamically using
̿∆𝑧
𝐷𝑥 = 𝛽𝑈
Where β is a user input value, U with double bars are the time-averaged, lateral averaged longitudinal ve-
locity at a segment and layer. Hence, the dispersion coefficient scales with the average longitudinal veloc-
ity.
Dz is internally computed within the CE-QUAL-W2 model using the Reynold’s analogy, where D z is
computed from Az, the vertical eddy viscosity, from
𝐷𝑧 = 0.14𝐴𝑧
Temperature Auxiliary Functions

Heat Exchange
Surface Heat Exchange
Surface heat exchange can be formulated as a term by-term process (see Figure 52) using the explicit adja-
cent cell transport computation as long as the integration timestep is shorter than or equal to the frequency
of the meteorological data. Surface heat exchange processes depending on water surface temperatures
are computed using computations from the prior time step.

Figure 52. Surface heat exchange term-by-term formulation.
Term-by-term surface heat exchange is computed as:
 n =  s −  sr +  a −  ar − br −  e −  c (4-13)
where:
n: net surface heat flux, W m-2
s: incoming short-wave solar radiation, W m-2
sr: reflected short-wave solar radiation, W m-2
sn: net short-wave solar radiation, s-sr, W m-2
a: incoming long-wave atmospheric radiation, W m-2
ar: reflected atmospheric long-wave radiation, W m-2
br: back long-wave radiation, W m-2
e: evaporative heat loss, W m-2
c: conductive heat loss, W m-2
The short-wave solar radiation is either measured directly or computed from sun angle relationships and
cloud cover. Annear and Wells (2007) provide an overview of formulae for short wave solar radiation. If the
model user chooses the internally computed short wave solar, the model computes the net clear sky solar
radiation 𝜙𝑠 (net short-wave solar radiation in Btu/ft2/day) using the EPA (1971) relationship:
(4-14)
𝜙𝑠 = 24(2.044𝐴𝑜 + 0.1296𝐴2𝑜 − 1.941𝐸 − 3𝐴3𝑜 + 7.591𝐸 − 6𝐴4𝑜 )*0.1314
Note that this equation includes the average reflection loss of short-wave solar radiation. The total clear
sky solar radiation was calculated using a least-squares fit polynomial regression of the solar altitude𝐴𝑜

(degrees), which is the angle of inclination of the sun relative to the horizon from an observer's perspective
(Wunderlich, 1972) using
(4-15)
𝐴𝑜 = 𝐴𝑆𝑖𝑛[𝑆𝑖𝑛(𝐿𝑎𝑡)𝑆𝑖𝑛(𝛿) + 𝐶𝑜𝑠(𝐿𝑎𝑡)𝐶𝑜𝑠(𝛿)𝐶𝑜𝑠(𝐻)]
where Lat is the latitude, 𝛿 is the solar declination, and H is the local hour angle. The local hour angle H
(radians), is the angular position of the sun for a given location at a specific time during the day and was
calculated from Ryan and Stolzenbach (1972) using
2𝜋 24 (4-16)
𝐻= [𝐻𝑂𝑈𝑅 − (𝐿𝑜𝑛𝑔 − 𝜑) + 𝐸𝑄𝑇 − 12.0]
24 360
where HOUR is the local hour, 𝜑 is standard meridian, Long is the longitude, and EQT is the equation of
time. The equation of time EQT (hours), represents the difference between true solar time and mean solar
time due to seasonal variations in the orbital velocity of the earth (Ryan and Stolzenbach, 1972). DiLaura
(1984) calculated EQT as
𝐸𝑄𝑇 = 0.170𝑆𝑖𝑛[4𝜋(𝐼𝑁𝑇(𝐽𝑑𝑎𝑦) − 80)/373] (4-17)

−0.129𝑆𝑖𝑛[2𝜋(𝐼𝑁𝑇(𝐽𝑑𝑎𝑦) − 8)/355]
where Jday is the Julian day as a floating-point value on a scale of 1 to 365 days for a year (366 for a leap
year) and INT is the integer function. The local hour HOUR (hours), was calculated using the time during
the day (Wunderlich, 1972), such as
(4-18)
𝐻𝑂𝑈𝑅 = 24(𝐽𝑑𝑎𝑦 − 𝐼𝑁𝑇(𝐽𝑑𝑎𝑦))
EPA (1971) calculated the standard meridian 𝜑 (degrees), as
𝐿𝑜𝑛𝑔 (4-19)
𝜑 = 15.0 ∗ 𝐼𝑁𝑇 ( )
15.0
where Long is the longitude. The time zones calculate a more appropriate standard meridian than the
longitude, so the variable TZ (hours), the time zone relative to Greenwich Mean Time (GMT) was used to
improve the calculation of the standard meridian in the above equation, such that
(4-20)
𝜑 = −15.0 ∗ 𝐼𝑁𝑇(𝑇𝑍)
The solar declination angle 𝛿 (radians) was calculated by Spencer (1971) as:
𝛿 = 0.006918 − 0.399912𝐶𝑜𝑠(𝜏𝑑 ) + 0.070257𝑆𝑖𝑛(𝜏𝑑 )

(4-21)
−0.006758𝐶𝑜𝑠(2𝜏𝑑 ) + 0.000907𝑆𝑖𝑛(2𝜏𝑑 )
−0.002697𝐶𝑜𝑠(3𝜏𝑑 ) + 0.001480𝑆𝑖𝑛(3𝜏𝑑 )
where𝜏𝑑 is the angular fraction of the year which Spencer (1971) calculated as

2𝜋(𝐼𝑁𝑇(𝐽𝑑𝑎𝑦) − 1)
𝜏𝑑 = (4-22)
365
Cloud cover reduction of clear sky solar radiation uses the following relationship (Wunderlich, 1972):
𝜙𝑠_𝑛𝑒𝑡 = 𝜙𝑠_𝑐𝑙𝑒𝑎𝑟𝑠𝑘𝑦 (1 − 0.65𝐶 2 ) where C is the cloud cover fraction between 0 and 1.
The long wave atmospheric radiation is computed from air temperature and cloud cover or air vapor pres-
sure using Brunt’s formula. The right-hand terms are all water surface temperature dependent.
Water surface back radiation is computed as:
𝜙𝑏𝑟 = 𝜀𝜎(𝑇𝑠 + 273.15)4 (4-23)
where:
 = emissivity of water, 0.97
 = Stephan-Boltzman constant, 5.67 x 10-8 W m-2 oK-4
Ts = water surface temperature, oC
Like the remaining terms, it is computed for each surface layer cell on each iteration timestep.
Evaporative heat loss is computed as:
𝜙𝑒 = 𝑓(𝑊𝑧 )(𝑒𝑠 − 𝑒𝑎 ) (4-24)
where:
f(Wz) = evaporative wind speed function at wind height of z, W m-2 mm Hg-1
es = saturation vapor pressure at the water surface, mm Hg
ea = atmospheric vapor pressure, mm Hg
Evaporative heat loss depends on air temperature and dew point temperature or relative humidity. Surface
vapor pressure is computed from the surface temperature for each surface cell on each iteration.
Surface heat conduction is computed as:
𝜙𝑐 = 𝐶𝑐 𝑓(𝑊𝑧 )(𝑇𝑠 − 𝑇𝑎 ) (4-25)
where:
Cc = Bowen's coefficient, 0.47 mm Hg oC-1
Ta = air temperature, oC
Short wave solar radiation penetrates the surface and decays exponentially with depth according to Bear’s
Law (see Figure 53):
𝜙𝑠 (𝑧) = (1 − 𝛽)𝜙𝑠𝑛 𝑒 −𝜂𝑧 (4-26)
where:
s(z) =short wave radiation at depth z, W m-2
 =fraction absorbed at the water surface
 =extinction coefficient, m-1
sn =net short-wave radiation reaching the surface, W m-2

Figure 53. Short-wave solar penetration in a water body.
Evaporation Models for f(W)
Aside from the problems of measuring meteorological data and obtaining representative meteorological
data for a waterbody, evaporative wind speed function, f(Wz), chosen by the model user can affect surface
temperatures especially in the summer. Various formulations of f(Wz) have been catalogued and examined
in Edinger et al. (1974).
The model user can include different evaporation formulations using an evaporation wind speed formula
of the form
𝑓(𝑊𝑧 ) = 𝑎 + 𝑏𝑊𝑧𝑐 (4-27)
where:
f(Wz) = wind speed function with wind measured at height z, W m-2 mm Hg-1
a = empirical coefficient, 9.2 default
b = empirical coefficient, 0.46 default
c = empirical coefficient, 2 default
Wz = wind speed measured at 2 m above the ground, m s-1
The function is used in computing both evaporative water and heat loss. The default values for a, b, and c
are the ones suggested in Edinger et. al. (1974). The model assumes that the wind is measured at a 2m
height. The following equation converts b from any measurement height to 2 m:
𝑏2𝑚 = 𝛼 𝑐 𝑏𝑧 (4-28)
where bz is b measured at z m and  is the conversion factor between the wind at z and the wind at 2m
using

2
ln 
W 2m  z0  1
= = (4-29)
Wz z  
ln 
 z0 
where:
W2m = wind speed at elevation 2 m, m s-1
Wz = wind speed at height z, m s-1
z0 = wind roughness height
Note that wind roughness height, zo, is a model input parameter since Version 3.6. Typical roughness values
include 0.003 ft (0.001 m) for wind < 5 mph (2.3 m/s) and 0.015 ft (0.005 m) for wind > 5 mph (2.3 m/s). The
roughness values can range from 0.0005 ft to 0.03 ft (0.00015-0.01 m).
The Ryan-Harleman (1974) evaporation model formulation has also been included in the model using:
𝑓(𝑊𝑧 ) = 𝑎 + 𝑏𝑊𝑧 (4-30)
where:
b = 4.1 W m-2 mm Hg-1 m-1 s-1
a =  ( T sv - T av ) , where s is surface temperature and a is air temperature
1/ 3
 = 3.59 W m2 mm Hg C-1/3
−1
  e 
T v = T  1 - 0.378   , K
*
  p 
p = atmospheric pressure, assumed to be 760 mm Hg
For the Lake Hefner model, a = 0 and b = 4.99 W m-2 mm Hg-1 m-1 s-1.
If the virtual temperature difference, Tv, is negative or less than that computed using the Lake Hefner model,
f(W) reverts to the Lake Hefner evaporation model. Figure 54 shows a comparison of the Ryan-Harleman
model with the model’s default formulation.
Adams et al. (1981) recommended that the Lake Hefner model be used for natural lakes (Table 6).

200
Ryan-Harleman formulation
Default W2 formulation afw=9.2, bfw=0.46, cfw=2
Lake Hefner Model
f(W), W/m2/mm Hg 160
120
80
40
0 4 8 12 16 20
Wind speed, m/s
Figure 54. Comparison of the wind speed formualtion for Ryan-Harleman and W2 default (for Tair=15ºC,
Tdew=-5ºC, Tsurface=25ºC).
Summaries of several evaporation formulations are shown below in Table 6 as adapted from Adams, et al.
(1981).
Table 6. Typical Evaporation Formulae for Lakes and Reservoirs

Name Time Waterbody e f(W) Remarks
good agreement with
3 hrs
Lake Lake Hefner, lake data from sev-
and 17.2W2(es-e2) 5.06W2
Hefner OK, 2587 acres eral lakes in US and
day
Russia
essentially the same
Lake Hefner OK,
Kohler day 17.5W2(es-e2) 5.14W2 as Lake Hefner for-
2587 acres
mula
ponds and 0.1708+ based on Russian
Zaykov - (1.3+14W2) (es-e2)
small reservoirs 4.11W2 work
ea obtained daily
from mean morning
small lakes and 10.512+2.94W and evening meas-
Meyer month (80+10W2) (es-e2)
reservoirs 2 urements of air tem-
perature and relative
humidity

Name Time Waterbody e f(W) Remarks

data from meteoro-
Mor- logical stations,
month Class A pan (73.5+14.7W2) (es-e2) 9.658+4.32W2
ton measurement
heights assumed
extensive pan meas-
pans, 85 ft dia urements using dif-
Rohwe
day tank, 1300 acre (67+10W2) (es-e2) 8.8+2.94W2 ferent pans, corre-
r
reservoir lated with tank and
reservoir data
e = evaporation at sea-level with wind corrected to 2 m, BTU ft-2 day-1
W = wind, mph
e = vapor pressure, mm Hg
W m-2 = 0.1314*BTU ft-2 day-1
10 mb = 7.5006151 mm Hg
The mass evaporation rate is computed by dividing evaporative heat loss by the latent heat of evaporation
of water. Surface heat exchange always includes evaporative heat loss in the heat budget, but the user may
choose to exclude it in the water budget. For many reservoirs, inflow rates are determined from storage
estimates that implicitly include evaporation.
Equilibrium Temperature
Another option in lieu of the term-by-term formulation presented above is to use a linearized form of the
heat flux expression that is a linear function of surface temperature. The equilibrium temperature, TE, is
defined as the temperature at which the net rate of surface heat exchange is zero (see Figure 55) or is the
temperature a water body is seeking if it is starting at a surface temperature of Ts.
Figure 55. Concept of the equilibrium temperature, TE, where the net surface heat flux is defined as zero.

Linearization of the term-by-term heat balance along with the definition of equilibrium temperature allows
expressing the net rate of surface heat exchange, n, as:
𝜙𝑛 = −𝐾(𝑇𝑠 − 𝑇𝐸 ) (4-31)
where:
n = net rate of surface heat exchange, W m-2
K = coefficient of surface heat exchange, W m-2 oC-1
Tw = water surface temperature, oC
TE = equilibrium temperature, oC
Seven separate heat exchange processes are summarized in the coefficient of surface heat exchange and
equilibrium temperature. The linearization is examined in detail by Brady, et al. (1969) and Edinger et al.
(1974).
How to compute TE and K?

• Compute TE directly from n=0
• Compute K directly from K=-n/(Ts-TE)
Or these can be approximated as shown below:
Estimating TE and K:
TE: compute from heat flux equation where n=0 or approximate technique (Brady et al., 1969):
𝑠𝑛𝜑
𝑇𝐸 = 23+𝑓(𝑊)(𝛽+0.255) + 𝑇𝑑
where:
sn: net short wave solar in Btu/ft2/day
Td: dew point temperature oF
f(W): wind speed function = 17W2 Lake Hefner model, in Btu/ft2/day/mm Hg
𝛽 = 0.255 − 0.0085𝑇 ∗ + 0.000204𝑇 ∗2
T*=0.5(Tw+Td)
Tw: water surface temperature
Td: dew point temperature
W2: wind speed at 2 m in mph
K: compute from slope of net flux vs temperature or use approximate formula (Brady et al.,
1969):
𝐾 = 23 + (𝛽𝑤 + 0.255)17𝑊2
where: 𝛽𝑤 = 0.255 − 0.0085𝑇𝑤 + 0.000204𝑇𝑤 2 and units of K are in Btu/ft2/day/oF and are converted to
W/m2/oC.
Sediment Heat Exchange
Sediment heat exchange with water is generally small compared to surface heat exchange and many previ-
ous modelers have neglected it. Investigations on several reservoirs have shown the process must be in-
cluded to accurately reproduce hypolimnetic temperatures primarily because of the reduction in numerical

diffusion that previously swamped the numerical solution. The formulation is similar to surface heat ex-
change:
H sw = − K sw ( T w − T s ) (4-32)
where:
Hsw = rate of sediment/water heat exchange, W m-2
Ksw = coefficient of sediment/water heat exchange, W m-2 oC-1
Tw = water temperature, oC
Ts = sediment temperature, oC
Previous applications used a value of 0.3 W m-2 ºC-1 for Ksw that is approximately two orders of magnitude
smaller than the surface heat exchange coefficient. Average yearly air temperature is a good estimate of
Ts.
Dynamic Shading
Solar Altitude and Azimuth
The declination angle, δ, is computed from Spencer (1971):
𝛿 = 0.006918 − 0.399912 𝑐𝑜𝑠(𝜏𝑑 ) + 0.070257 𝑠𝑖𝑛(𝜏𝑑 ) − 0.006758 𝑐𝑜𝑠(2𝜏𝑑 ) +

0.000907 𝑠𝑖𝑛(2𝜏𝑑 ) − 0.0022697 𝑐𝑜𝑠(3𝜏𝑑 ) + 0.001480 𝑠𝑖𝑛(3𝜏𝑑 ) (4-33)
where:
2𝜋(𝐽𝐷𝑖 −1)
𝜏𝑑 =
365
JDi = Julian date integer value
The local hour is calculated as:
𝐻𝑂𝑈𝑅 = 24(𝐽𝐷𝑟 − 𝐽𝐷𝑖 ) (4-34)
where:
JDr = Julian date floating-point value
An equation of time, EQT, correction is needed to calculate the local hour angle. The equation of time
represents the difference between true and mean solar time due to seasonal variations in the earth’s orbital
velocity (DiLaura, D.L 1984) and is given as:
4𝜋(𝐽𝐷𝑖 −80) 2𝜋(𝐽𝐷𝑖 −8)

𝐸𝑄𝑇 = 0.17 𝑠𝑖𝑛 − 0.129 𝑠𝑖𝑛 (4-35)
373 355
where:

AUXILIARY FUNCTIONS SHADING
The local hour angle, H, is calculated as:
2𝜋 24
𝐻= 24
[𝐻𝑂𝑈𝑅 + (𝐿𝑜𝑛𝑔 − 𝜑) 360 + 𝐸𝑄𝑇 − 12] (4-36)
The solar altitude, Ao, is the angle of inclination of the sun relative to the horizon from an observer's per-
spective as shown in Figure 56.
Ao
Az
Figure 56. Schematic of solar altitude, Ao, and azimuth, AZ.
Ao is calculated from Wunderlich (1972):
𝜋 𝜋
𝐴𝑜 = 𝐴 𝑠𝑖𝑛 [𝑠𝑖𝑛 (𝑙𝑎𝑡 ∗ ) 𝑠𝑖𝑛(𝛿) + 𝑐𝑜𝑠 (𝑙𝑎𝑡 ∗ ) 𝑐𝑜𝑠(𝛿) 𝑐𝑜𝑠(𝐻)] (4-37)
180 180
The solar azimuth is the direction of the sun with respect to a North-South axis measured clockwise from
the North as shown in Figure 57. The solar azimuth is computed as (Nautical Almanac, 2001):
𝐿𝑎𝑡∗𝜋 𝐿𝑎𝑡∗𝜋
𝑠𝑖𝑛(𝛿) 𝑐𝑜𝑠( )−𝑐𝑜𝑠(𝛿) 𝑐𝑜𝑠(𝐻) 𝑠𝑖𝑛( )
180 180
𝑋= 𝑐𝑜𝑠 𝐴𝑜
(4-38)
If 𝑋 > 1, 𝑋 = 1
If 𝑋 < −1, 𝑋 = −1 (4-39)
𝐴 = 𝐴 𝑐𝑜𝑠 𝑋 (4-40)
𝐴𝑍 = 2𝜋 − 𝐴
If 𝐻 < 0, 𝐴𝑍 = 𝐴 (4-41)

SHADING AUXILIARY FUNCTIONS
After computing the solar altitude and azimuth, the impact of shading (Figure 57) on short-wave solar radi-
ation is computed as follows.
ST
Sw
0
E B/2
Cl
Figure 57. Schematic of topographic and vegetative shading, solar altitude (0), and vegetation height
(T) and their effect on shadow length.
Topographic Shading
The algorithm uses the position of the sun to determine which topographic inclination angle coincides with
the direction of incoming solar radiation. The algorithm determines the closest two inclination angles in
the direction of the incoming solar radiation and uses them to linearly interpolate an inclination angle for
the specific direction of the incoming solar radiation. The calculated inclination angle is then used to de-
termine if vegetative or topographic shading dominates at that time. If the solar altitude is below the cal-
culated topographic inclination angle, then topographic shading dominates and the short-wave solar radi-
ation is reduced by 90% for complete shade. This allows for 10% of the incoming solar radiation as a result
of diffuse radiation even when in the shade. If the solar altitude is above the calculated inclination angle,
then vegetative shading dominates.
Vegetative Shading
If the topographic angle is less than the solar altitude, vegetative shade dominates and the algorithm cal-
culates the shading influence by determining how far the shadow extends over the water. Figure 58 and
Figure 59 show schematics of the azimuth angle, segment orientation and computed shadow lengths. The
tree shadow length, ST, is calculated using the tree height, T, and solar altitude, A0, using Equation 206.
Then the length of the shadow cast over the water is calculated using Equation 207 where E is the distance
between the tree and the edge of water and 0 is the segment orientation. The shadow length, SN, perpen-
dicular to the edge of the water is then calculated using Equation 208. Refer to Figure 58 and Figure 59 for
diagrams showing the distance calculated in Equations 206 to 208.
𝑇
𝑆𝑇 = 𝑡𝑎𝑛 𝐴 (4-42)
0
𝐸
𝑆𝑤 = 𝑆𝑇 − 𝑠𝑖𝑛(𝛩 (4-43)
0 −𝛼𝐴 )

𝑆𝑁 = 𝑆𝑊 𝑠𝑖𝑛(𝛩0 − 𝛼𝐴 ) (4-44)
Simplifying Equations 206 to 208:
𝑇∗𝑠𝑖𝑛(𝛩0 −𝛼𝐴 )
𝑆𝑁 = 𝑡𝑎𝑛 𝐴0
− 𝐸 (4-45)
A shading reduction factor is applied in cases where a model segment has potential shading along only part
of its segment length or the vegetation density is low. For example, if shade-producing vegetation exists
along only half the length of a segment and is 100% opaque, a shade reduction factor of 0.5 is used. If
shading is due to vegetation along only half of a segment with 80% opaqueness, a value of 0.4 is used.
The shade factor, 𝑠𝑓𝑎𝑐𝑡, is the shadow length perpendicular to the edge of the water, SN, multiplied by the
shade reduction factor, SRF, and divided by the segment width:
𝑆𝑁
𝑠𝑓𝑎𝑐𝑡 = 𝑆𝑅𝐹 (4-46)
𝐵
The amount of shade that should be applied to the incoming short-wave solar radiation is calculated as:
𝑆ℎ𝑎𝑑𝑒 = (1 − 𝑠𝑓𝑎𝑐𝑡) (4-47)
The short-wave solar radiation [SRO] computed by the model is reduced by the shade variable:
𝑆𝑅𝑂𝑛𝑒𝑡 = 𝑆𝑅𝑂 ∗ 𝑆ℎ𝑎𝑑𝑒 (4-48)
Figure 58. Azimuth angle, A, and stream orientation, 0.

SHADING AUXILIARY FUNCTIONS
Figure 59. Relationship between azimuth, stream orientation, and shadow length.
Data Requirements
Topography and vegetation data are stored in a user-defined input file. An example input file is shown in
Appendix C. The file includes tree top elevations for both stream banks. The file also includes the distance
from the centerline of the river to the controlling vegetation and the shade reduction factor (explained
below) for both stream banks. The shade file has vegetation characteristics recorded by the left and right
banks of the stream. The convention used for defining left or right bank is dependent on looking down-
stream in the system, the ‘Left Bank’ is on the left and the ‘Right Bank’ is to the right.
The shade algorithm reads in 18 topographic inclination angles surrounding each segment center-point.
The inclination angles can be determined using topographical maps, Digital Elevation Models (DEM), or
contour plots. The steepest inclination angle for each of the 18 locations surrounding a segment should be
selected since this angle will control the topographic shading. The first inclination angle is taken from di-
rectly North of a segment (orientation angle 0.0) and moves clockwise to the East with increasing orienta-
tion angles around the segment in 20 increments.
How far away from the centerline of the river the topography should be analyzed will depend on the system.
Wide flat river systems will utilize longer distances for identifying influencing topography than a narrow
river canyon. In addition, rather than restricting the code to orientation angles only toward the south ap-
propriate for the Northern hemisphere, using orientation angles that surround a segment allows the algo-
rithm to be used in both the northern and southern hemispheres.
In addition, the user can specify dynamic shading reduction factors as a function of time. Usually, these
would correspond to the times for leaf growth and fall for deciduous trees.

The topography and vegetation information is first read in from the input file. Then, using the segment
orientation angle, 0, and the solar azimuth, A, the bank that has the sun behind it is computed. The
criteria used for determining the bank with the sun behind it was modified from Chen (1996) because the
segment orientation angle is determined differently in CE-QUAL-W2. Table 7 shows the criteria used in the
model.
Table 7. Criteria for determining sunward bank

Sunward 0° < 𝛩0 ≤ 180° 180° < 𝛩0 ≤ 360°
Bank
Left 𝛼𝐴 < 𝛩0 or 𝛼𝐴 > 𝛩0 + 180° 𝛩0 − 180° < 𝛼𝐴 < 𝛩0
Right 𝛩0 < 𝛼𝐴 < 𝛩0 + 180° 𝛼𝐴 > 𝛩0 or 𝛼𝐴 < 𝛩0 − 180°

Ice Cover
Ice thickness, onset, and loss of ice cover play an important role in the heat budget of northern waterbodies.
At high latitudes, ice cover may remain until late spring or early summer and prevent warming due to ab-
sorption of short-wave solar radiation.
The ice model is based on an ice cover with ice-to-air heat exchange, conduction through the ice, conduc-
tion between underlying water, and a "melt temperature" layer on the ice bottom (Ashton, 1979). The
overall heat balance for the water-to-ice-to-air system is:
h
i L f = h ai ( T i - T e ) - h wi ( T w - T m ) (4-49)
t
where:
i = density of ice, kg m-3
Lf = latent heat of fusion of ice, J kg-1
h/t = change in ice thickness (h) with time (t), m sec-1
hai = coefficient of ice-to-air heat exchange, W m-2 oC-1
hwi = coefficient of water-to-ice heat exchange through the melt layer, W m-2 oC-1
Ti = ice temperature, oC
Te = equilibrium temperature of ice-to-air heat exchange, oC
Tw = water temperature below ice, oC
Tm = melt temperature, 0oC
The ice-to-air coefficient of surface heat exchange, hai, and its equilibrium temperature, Tei, are computed
the same as for surface heat exchange in Edinger, et al. (1974) because heat balance of the thin, ice surface
water layer is the same as the net rate of surface heat exchange presented previously. The coefficient of
water-to-ice exchange, hwi, depends on turbulence and water movement under ice and their effect on melt
layer thickness. It is a function of water velocity for rivers but must be empirically adjusted for reservoirs.
Ice temperature in the ice-heat balance is computed by equating the rate of surface heat transfer between
ice and air to the rate of heat conduction through ice:
- ki ( T i - T m )
h ai ( T i - T ei ) = (4-50)
h
where:
ki = molecular heat conductivity of ice, W m-1 oC-1
When solved for ice temperature, Ti, and inserted in the overall ice-heat balance, the ice thickness relation-
ship becomes:
i L f  h ( T M - T ei )
= - h wi ( T w - T m ) (4-51)
t h
+
1
ki h ia
95 Water Quality
OVERVIEW
from which ice thickness can be computed for each longitudinal segment. Heat from water to ice trans-
ferred by the last term is removed in the water temperature transport computations.
Variations in the onset of ice cover and seasonal growth and melt over the waterbody depend on locations
and temperatures of inflows and outflows, evaporative wind variations over the ice surface, and effects of
water movement on the ice-to-water exchange coefficient. Ice will often form in reservoir branches before
forming in the main pool and remain longer due to these effects.
A second, more detailed algorithm for computing ice growth and decay has been developed for the model.
The algorithm consists of a series of one-dimensional, quasi steady-state, thermodynamic calculations for
each timestep. This algorithm is similar to those of Maykut and Untersteiner (1971), Wake (1977) and Pat-
terson and Hamblin (1988). The detailed algorithm provides a more accurate representation of the upper
part of the ice temperature profile resulting in a more accurate calculation of ice surface temperature and
rate of ice freezing and melting.
The ice surface temperature, Ts, is iteratively computed at each timestep using the upper boundary condi-
tion as follows. Assuming linear thermal gradients and using finite difference approximations, heat fluxes
through the ice, qi, and at the ice-water interface, qiw, are computed. Ice thickness at time t, θ(t), is deter-
mined by ice melt at the air-ice interface, Δθai, and ice growth and melt at the ice-water interface, Δθiw.
The series of computations for ice cover is presented below.
Initial Ice Formation
Formation of ice requires lowering the surface water temperature to the freezing point by normal surface
heat exchange processes. With further heat removal, ice begins to form on the water surface. This is indi-
cated by a negative water surface temperature. The negative water surface temperature is then converted
to equivalent ice thickness and equivalent heat is added to the heat source and sink term for water. The
computation is done once for each segment beginning with the ice-free period:
- T wn  w C P w h
0= (4-52)
i L f
where:
0 = thickness of initial ice formation during a timestep, m
Twn = local temporary negative water temperature, oC
h = layer thickness, m
w = density of water, kg m-3
Cpw = specific heat of water, J kg-1 oC-1
i = density of ice, kg m-3
Lf = latent heat of fusion, J kg-1
Air-Ice Flux Boundary Condition and Ice Surface Temperature Approximation
The ice surface temperature, Ts, must be known to calculate the heat components, Hbr, He, Hc, and the
thermal gradient in the ice since the components and gradient all are either explicitly or implicitly a function
of Ts. Except during the active thawing season when ice surface temperature is constant at 0ºC, Ts must be
computed at each timestep using the upper boundary condition. The approximate value for Ts is obtained
by linearizing the ice thickness across the timestep and solving for Ts.
Water Quality 96
 H 
n-1
T
n
s
n
sn
n
+ Han - Hbr (Tsn ) - He (Tsn ) - Hc (Tsn ) (4-53)
Ki
d  ai
H sn + H an - H br - H e - H c + qi =  i L f , for T s = 0 o C (4-54)
dt
T f - T s (t)
qi = K i (4-55)
 (t)
where:
Ki = thermal conductivity of ice, W m-1 oC -1
Tf = freezing point temperature, oC
n = time level
 = thickness of ice, m
-2
qi = heat flux through ice, W m
Absorbed Solar Radiation by Water Under Ice
Although the amount of penetrated solar radiation is relatively small, it is an important component of the
heat budget since it is the only heat source to the water column when ice is present and may contribute
significantly to ice melting at the ice-water interface. The amount of solar radiation absorbed by water
under the ice cover may be expressed as:
𝐻𝑝𝑠 = 𝐻𝑠 (1 − 𝐴𝐿𝐵)(1 − 𝛽𝑖 )𝑒 −𝛾𝑖𝜃(𝑡) (4-56)
where:
Hps = solar radiation absorbed by water under ice cover, W m-2
Hs = incident solar radiation, W m-2
ALBi = ice albedo
i = fraction of the incoming solar radiation absorbed in the ice surface
i = ice extinction coefficient, m-1
Ice Melt at Air-Ice Interface
The solution for Ts holds as long as net surface heat exchange, Hn(Ts), remains negative corresponding to
surface cooling, and surface melting cannot occur. If Hn(Ts) becomes positive corresponding to a net gain
of heat at the surface, qi must become negative and an equilibrium solution can only exist if Ts > Tf. This
situation is not possible as melting will occur at the surface before equilibrium is reached (Patterson and
Hamblin, 1988). Because of a quasi-steady approximation, heat, which in reality is used to melt ice at the
surface, is stored internally producing an unrealistic temperature profile. Stored energy is used for melting
at each timestep and since total energy input is the same, net error is small. Stored energy used for melting
ice is expressed as:
(t)
 i C p T s  (t) =  i L f   ai
i
(4-57)
2
97 Water Quality
OVERVIEW
where:
Cpi = specific heat of ice, J kg-1 oC-1
ai = ice melt at the air-ice interface, m-1
Ice-Water Flux Boundary Condition Formulation
Both ice growth and melt may occur at the ice-water interface. The interface temperature, Tf, is fixed by
the water properties. Flux of heat in the ice at the interface therefore depends on Tf and the surface tem-
perature Ts through the heat flux qi. Independently, heat flux from the water to ice, qiw, depends only on
conditions beneath the ice. An imbalance between these fluxes provides a mechanism for freezing or melt-
ing. Thus,
d  iw
qi - qiw =  i L f (4-58)
dt
where:
iw = ice growth/melt at the ice-water interface
The coefficient of water-to-ice exchange, Kwi, depends on turbulence and water movement under the ice
and their effect on melt layer thickness. It is known to be a function of water velocity for rivers and streams
but must be empirically adjusted for reservoirs. The heat flux at the ice-water interface is:
qiw = h wi ( T w (t) - T f ) (4-59)
where:
Tw = water temperature in the uppermost layer under the ice, oC
Finally, ice growth or melt at the ice-water interface is:
1  T f - T ns 
 n
= K i - h wi ( T nw - T f ) (4-60)
i L f 
iw
 n-1

Freezing Temperature of Ice
In general, the temperature at which water freezes, Tf, is set to 0oC for fresh water. If the model is set to
SALT [See WTYPEC input], the freezing temperature is affected by the salinity. In the model, when
WTYPEC=SALT, the TDS state variable is in units of ppt and the equations used for computing the freezing
temperature are computed from the following equations:
For TDS < 35 ppt,

𝑇𝑓 = −0.0545𝑇𝐷𝑆 (4-61)
For TDS > 35 ppt:
𝑇𝑓 = −0.3146 − 0.0417𝑇𝐷𝑆 − −0.000166𝑇𝐷𝑆 2 (4-62)
Water Quality 98
Gain and Loss of Water
As ice is formed, water is removed from the waterbody. There is a removal of 0.917 m 3 of water for every
1 m3 of ice formed. The reverse occurs as ice melts and comes back into the water body. Currently, the
model does not account for ablation nor on snow accumulation even though this would be easy to add to
the code.
Water is conserved in forming and melting ice from the waterbody. Hence, dissolved substances in the
water are concentrated as ice forms. When the ice melts, the model assumes the ice temperature is 0oC
and is pure ice with no dissolved constituents. This process is shown below in Figure 60.
Figure 60. Ice formation and melting water balance.
Overview of Kinetic Source/Sink Term

In order to solve the 2D advection-diffusion equation, the source/sink term, S, must be specified. The
model solves for temperature and a user specified number of water quality variables. Water quality state
variables along with their kinetic source/sink terms are shown in Table 8. The user can specify any number
of generic constituents [NGC], suspended solids groups [NSS], CBOD groups [NBOD], algal groups [NAL],
macrophyte groups [NMC], zooplankton groups [NZP], and periphyton/epiphyton groups [NEP].
99 Water Quality
OVERVIEW
Table 8. CE-QUAL-W2 Water Quality State Variables

Constituent Internal Source Internal Sink
Total dissolved solids
generic constituent, no interac-
tions with other state variables
bacteria settling
tracer 0 and 1st order increase 0 and 1st order decay
water age reaeration gas transfer (volatilization)
contaminants photodegradation
N2 gas (Total Dissolved Gas)
volatile contaminant
Inorganic suspended solids sedimentation
bioavailable P measured as one of algal/epiphyton respiration
the following labile/refractory particulate/dissolved
algal/epiphyton growth
ortho-P organic matter
adsorption onto inorganic suspended solids
dissolved P sediment release
SRP CBOD decay
sediment release
algal/epiphyton/macrophyte excretion
algal/epiphyton/macrophyte growth
ammonium labile/refractory dissolved/ particulate
nitrification
organic matter decay
CBOD decay
denitrification
nitrate-nitrite Nitrification
algal/epiphyton/macrophyte growth
anoxic sediment release algal/epiphyton growth
dissolved silica
particulate biogenic silica decay adsorption onto suspended solids
settling
particulate biogenic silica algal/epiphyton mortality
decay
Iron anoxic sediment release oxic water column settling
algal/epiphyton/macrophyte
labile dissolved organic matter mortality Decay
excretion
refractory dissolved organic matter labile dissolved organic matter decay Decay
settling
labile particulate organic matter algal/epiphyton/macrophyte mortality
decay
refractory particulate organic mat- settling
labile particulate organic matter decay
ter decay
Total P in labile dissolved organic
mortality Decay
matter
excretion
Total P in refractory dissolved or-
labile dissolved organic matter decay Decay
ganic matter
Total P in labile particulate organic settling
algal/epiphyton/macrophyte mortality
matter decay
Total P in refractory particulate or- settling
ganic matter decay
Total N in labile dissolved organic
mortality Decay
matter
excretion
Total N in refractory dissolved or-
labile dissolved organic matter decay Decay
ganic matter
Total N in labile particulate organic settling
algal/epiphyton/macrophyte mortality
matter decay
Total N in refractory particulate or- settling
ganic matter decay
CBOD decay, settling
CBOD-P (Total P in organic matter
decay, settling
represented by CBOD)
CBOD-N (Total N in organic matter
decay, settling
represented by CBOD)
Water Quality 100

Constituent Internal Source Internal Sink
respiration
excretion
algae algal growth
mortality
settling
respiration
excretion
Epiphyton epiphyton growth
mortality
settling
transport
settling
zooplankton zooplankton growth
excretion
mortality
respiration
macrophytes macrophyte growth mortality
excretion
surface exchange
algal/epiphyton/macrophyte/zooplankton respira-
tion
surface exchange nitrification
dissolved oxygen
algal/epiphyton growth CBOD decay
0 and 1st order SOD
labile/refractory dissolved/ particulate organic
matter decay
labile/refractory dissolved/ particulate
organic matter decay surface exchange
total inorganic carbon sediment release algal/epiphyton growth
surface exchange CBOD decay
algal respiration
[dynamic alkalinity can also be computed based
alkalinity on biological reactions otherwise it is treated as
conservative]
Methane Sediment diagenesis Surface exchange
Hydrogen sulfide Sediment diagenesis Surface exchange
Mature fine tailings (about 70%
water and 30% fine clay – can take
centuries to consolidate; from oil consolidation
sands mining operation; another
inorganic suspended solids group)
Sediment (1st order sediment settling of algae, LPOM, RPOM, epiphy-

decay, focusing of sediments
model) ton burial
Sediment diagenesis model with Organic matter settling to the sedi-
Release to overlying water (C, N, P)
state variables in the sediment ment and sediment diagenesis
In addition to these water quality state variables, the model also solves for pH and the carbonate cycle (CO 2,
HCO3, H2CO3) as part of the set of derived water quality variables.
The constituent transport relationships described in this section of the User’s Manual document the
transport of constituents with their kinetic reaction rates expressed in source and sink terms. All
sources/sinks (both internal and external) for water temperature are contained in the internal code array
[TSS]. The sources/sinks for constituents are separated into two arrays, [CSSB] and [CSSK]. [CSSB] contains
boundary sources/sinks. [CSSK] contains internal sources/sinks due to kinetic interactions. The division of
terms allows kinetic sources/sinks to be updated at different frequencies than boundary sources/sinks -
consistent with coarser time scales associated with biological and chemical processes as opposed to hydro-
dynamics. Computational time is also reduced. The frequency at which kinetic sources/sinks [CSSK] are
updated is specified by the parameter [CUF].
101 Water Quality

OVERVIEW
The source/sink term [CSSK] represents a mass rate of change (model code internal units of grams m-3 sec-
1
) of a constituent due to kinetic reactions where concentrations are expressed as grams meter-3 or mg/l.
The kinetic reactions can be depicted graphically by considering each constituent as a compartment. Ar-
rows represent mass transfer rates between compartments. All of the rate terms in the following discussion
are in units of sec-1 and these are the units used in the code. However, all rate units input into the model
from the control file are in units of day-1 and are then converted to sec-1 before being used in the code.
Both graphical and mathematical descriptions of the rate equation for each constituent are provided.
Generic Constituent
Any number of generic constituents [NGC] can be defined that can settle, decay (using either zero order
and/or first order decay and/or photochemical reactions), and/or transfer across the air/water interface.
The user supplies a zero and/or 1st order decay coefficient with or without an Arrhenius temperature de-
pendence function, and/or a settling velocity and/or a photodegradation coefficient and/or coefficients for
volatilization. Gas transfer (or volatilization) requires setting the reaeration coefficient (by making it a frac-
tion of the dissolved oxygen gas transfer rate) and a saturated gas concentration. Generic constituents do
not interact with the hydrodynamics nor any other water quality state variables.
Figure 61. Internal flux for generic constituent compartment.
Referring to Figure 61, the source/sink term for a generic constituent is:
Water Quality 102

 g
- I o (1 −  )e −z  g + sur K L ( s −  g )
A
Sg = − K0  g − K 1 g g − g
(T - 20) (T - 20)

 
0-order decay
  
1st -order decay z 
 photodegradation
 Vsur
  
settling gas transfer
(4-63)
where:
g = temperature rate multiplier
T = water temperature, oC
α = photodegradation parameter, m2/J
Io = radiation at surface, W/m2
 = light extinction coefficient, m-1
β= fraction of short wave solar absorbed on the surface
 g = settling velocity, m s-1
K0 = zero order decay coefficient, g m-3 s-1 at 20oC
K1 = first order decay coefficient, s-1 at 20oC
g = generic constituent concentration, g m-3
s = generic constituent concentration gas saturation in the atmosphere, g m-3
Asur = surface area, m2
Vsur = surface volume, m3
KL = surface gas transfer coefficient, m/s
Photodegradation can be an important mechanism for constituents such as fecal coliforms (Mitchell and
Chamberlain, 1978; Auer and Niehaus, 1993) and for dissolved organic matter (Moran and Zepp, 1996; Re-
che et al., 1999). The photo decay rate can be modeled as a pseudo 1st order decay (klight) rate, as described
by
𝛿𝐶
= −𝑘𝑙𝑖𝑔ℎ𝑡 × 𝐶
𝑑𝑡
Where klight is the first-order decay coefficient (sec-1) and C is the constituent concentration (g/m3). Cham-
berlin and Mitchell (1978) defined the light induced decay of coliforms based on Beer’s Law and a propor-
tionality constant as
𝑘𝑙𝑖𝑔ℎ𝑡 =∝ 𝐼0 𝑒 −𝜆𝑧
Where α is the user defined proportionality constant for the effect of incident light on a specific constituent
or organism in the case of fecal coliforms (cm 2/cal), I0 is the incident light energy at the water surface
(cal/cm2-day),  is the light attenuation coefficient per unit depth (m-1) and z is the depth below the water
surface (m). Table 9 shows a selection of α values reported in the literature for coliforms (Bowie et al.,
1985). The CE-QUAL-W2 model uses units of W/m2 for radiation, hence the corresponding units of alpha as
used in the CE-QUAL-W2 model are m2/J.
Table 9. Comparison of α estimates based on Chamberlin and Mitchell (1978) with additional values
(Bowie et al., 1985)
Organism Study α (cm2/cal) α (m2/J) Data Source

Gameson and
Coliform Group 14 field studies (mean) 0.481
1.15E-05 Gould (1975)
Foxworthy and
Coliform Group 24 field studies (mean) 0.168
4.01E-06 Kneeling (1969)
103 Water Quality

GENERIC CONSTITUENTS KINETICS
Organism Study α (cm2/cal) α (m2/J) Data Source

Gameson and
Coliform Group 61 laboratory studies (mean) 0.136
3.25E-06 Gould (1975)
Estimated from field and la-
Fecal Coliform boratory studies (mean of 0.042 Mancini (1978)
salt and freshwater) 1.00E-06
22 chamber studies in fresh-
Fecal Coliforms 0.005 Lantrip (1983)
water (mean) 1.19E-07
22 chamber studies in fresh-
Total Coliforms 0.004 Lantrip (1983)
water (mean) 9.55E-08
A conservative tracer, coliform bacteria, N2 gas (and Total Dissolved Gas), a volatile generic constituent, and
water age are some of the state variables that can be modeled using the generic constituent and are de-
scribed further below.
Conservative Tracer
A conservative constituent is included to allow dye study simulations, movements of conservative materials
through the waterbody, and as an aid in calibrating and testing flow regimes. As a conservative material,
this constituent has no internal sources or sinks and the rate term [CSSK] is set to zero.
𝑆𝑡𝑟𝑎𝑐𝑒𝑟 = 0 (4-64)
Coliform Bacteria with decay and photodegradation

Coliform bacteria are commonly used as an indicator of pathogen contamination. Safety standards and
criteria for drinking and recreational purposes are based upon coliform concentrations. Predictions of col-
iform bacteria are important because of their impact on recreation and water supply.
Figure 62. Internal flux for coliform bacteria.
Total coliform, fecal coliform, fecal streptococci, and/or any other type of bacteria that do not interact with
other state variables can be simulated with this generic constituent formulation. Referring to Figure 62,
 col the rate equation for coliform bacte-
S g = − K col  (T -20) col −  col

 - I o (1 −  )e −z col ria is (including photodegradation):
1st -order decay
z 


photodegradation
settling
Water Quality 104

(4-65)
where:
 = temperature factor (Q10)
T = water temperature, oC
Kcol = coliform mortality rate, sec-1 at 20oC
col = coliform concentration, g m-3
The Q10 formulation arises from a doubling of the reaction rate with each 10ºC increase in temperature.
This doubling rate has not been found at lower temperatures (Hargrave 1972) and is quite variable for var-
ious reactions (Giese 1968). Modeling coliform bacteria is discussed in detail in Zison, et al. (1978).
N2 dissolved gas and TDG

The generic constituent can model N2 gas as a state variable by
• setting the zero-order and first-order decay rate to 0 day-1
• setting the photodegradation parameter to 0
• setting the N2 gas saturation to “-1.0”
By setting these parameters, the W2 model internally computes the N2 gas saturation as a function of
temperature and air humidity and internally computes the gas transfer coefficient. The source sink term is
shown in Figure 63 and is
𝐴𝑠𝑢𝑟
𝑆𝑁2 = 𝐾 (𝛷𝑠𝑁2 − 𝛷𝑁2 ) (4-66)
𝑉𝑠𝑢𝑟 𝐿𝑁2
where:
N2 = N2 generic constituent concentration, g m-3
sN2 =N2 gas saturation concentration in the atmosphere, g m-3
Asur = surface area, m2
Vsur = surface volume, m3
KLN2 = surface gas transfer coefficient for N 2, m/s
Nitrogen gas is a slightly soluble gas, similar to oxygen. Its gas transfer characteristics are similar to oxygen
in being controlled by turbulence on the liquid film (Chapra, 1997).
The value of KLN2 can be determined from the reaeration formulae presented in the preceding section. The
basic physics of gas transfer are the same for N2 and O2 since they are both liquid-controlled gases. Using
the Higbie-Penetration theory (Chapra, 1997),
𝐷
𝐾𝐿𝑁2 = 𝐾𝐿𝑂2 √𝐷𝑁2
𝑂2
where DN2 and DO2 are the molecular diffusion coefficients for oxygen and nitrogen, respectively. Using
the relationship that
𝐷𝑁2 𝑀𝑊𝑂2
=√
𝐷𝑂2 𝑀𝑊𝑁2
𝑀𝑊𝑂2 0.25
𝐾𝐿𝑁2 = 𝐾𝐿𝑂2 ( ) = 1.034𝐾𝐿𝑂2
𝑀𝑊𝑁2
105 Water Quality

where MW is the molecular weight and MWO2=32 g/mole and MWN2=28 g/mole.
Surface exchange
N2 gas
Figure 63. N2 gas exchange.
The value of N2 saturation, S N2 , is determined using Henry's Law. Henry's Law can be written as
𝑥𝑁2 = 𝐾𝐻 𝑃𝑁2
where XN2 is the mole fraction of Nitrogen gas, KH is Henry's Law constant in moles/atm and PN2 is the par-
tial pressure of N2 gas in the atmosphere in atm. Nitrogen gas is assumed to be 79% of the total gas com-
position in the atmosphere for dry air. Also, in order to account for atmospheric pressure changes with
altitude, the following relationship is used in CE-QUAL-W2 to correct the atmospheric pressure:
𝐴𝑙𝑡𝑖𝑡𝑢𝑑𝑒, 𝑘𝑚 5.25
𝑃(𝑎𝑡𝑚) = [1 − ]
44.3
where altitude is the altitude of the water body in km. Since atmospheric pressure is not an input variable
to CE-QUAL-W2 (except when using the SYSTDG algorithm for TDG generation at hydropower facilities –
see under Dissolved Oxygen/Dam Reaeration), the atmospheric pressure is assumed to be 1 atm at sea
level.
The mole fraction of N2 gas is related to the mass concentration as follows:

𝑛𝑁2
𝑥𝑁2 =
𝑛𝐻2𝑂 + 𝑛𝑁2 +. . .
where nN2 is the number of moles of N2, nH2O is the number of moles of water (the dominant term). Neglect-
ing all terms in the denominator except nH2O, which is 55.6 moles/liter (=1000 g/18 g/mole), then the con-
centration of N2 is CN2:
28𝑔 103 𝑚𝑔
𝐶𝑁2 = 𝑛𝑁2
𝑚𝑜𝑙𝑒 𝑔
Combining the last 2 equations with Henry's Law, we obtain
𝐶𝑁2 = 55.6 ∗ 28 ∗ 103 𝑃𝑁2 𝐾𝐻 = 1.5568𝐸6𝑃𝑁2 𝐾𝐻
The Henry's Law constant is temperature dependent. Using the data of K H versus temperature in
Tchobanoglous and Schroeder (1987),
Water Quality 106

KH(atm-1)= 1.8816E-005 - 4.116E-007 * T + 4.6E-009 * T2
where T is the surface temperature in oC. This relationship is shown in Figure 64.
2E-005
1.8E-005
1.6E-005
KH, atm-1
1.4E-005
Polynomial
Equation Y = 1.8816E-005 - 4.116E-007 * X + 4.6E-009 * pow(X,2)
Degree = 2
1.2E-005 Number of data points used = 5
Average X = 20
Average Y = 1.3344E-005
Coefficients:
1E-005 Degree 0 = 1.8816E-005
Degree 1 = -4.116E-007
Degree 2 = 4.6E-009
Coef of determination, R-squared = 0.998568

8E-006
0 10 20 30 40
Temperature, oC
Figure 64. Henry's Law constant dependence on temperature.
The %N2 in the atmosphere needs to be corrected for water vapor in the atmosphere. The 79% N 2 is
based on dry air. The total pressure of dry air can be determined by taking the total atmospheric pressure
minus the contribution of water vapor.
The vapor pressure is computed using the air temperature in oC as (Chapra, 1997)
17.27𝑇
𝑃𝑤𝑎𝑡𝑒𝑟𝑣𝑎𝑝𝑜𝑟 (𝑚𝑚𝐻𝑔) = 4.596 𝑒𝑥𝑝 ( )
237.3 + 𝑇
Then Pwatervapor is converted to atm by multiplying by the conversion factor 0.00136. Since this pressure is
at saturation or at 100% humidity, the actual Pwatervapor is then multiplied by the relative humidity fraction.
Since the CE-QUAL-W2 model does not use relative humidity, the relative humidity (RH) was computed
based on dew point temperature, Td, and air temperature, T: (Singh, 1992)
112 − 0.1𝑇 + 𝑇𝑑 8
𝑅𝐻 = [ ]
112 + 0.9𝑇
where all temperatures are in oC.
Figure 65 shows the Pwatervapor as a function of air temperature and relative humidity.
107 Water Quality

0.080
0.070
0.060
Vapor pressure, atm
0.050
P,
RH=100%
0.040
P, RH=50%
0.030
0.020
0.010
0.000
0 20 40
Air temperature, C
Figure 65. Vapor pressure as a function of air temperature and relative humidity.
The %N2 is then 0.79(1-RH*Pwatervapor). The term (1-RH*Pwatervapor) is shown in Figure 66 as a function of air
temperature and relative humidity.
Also, the increase of salinity can affect the N2 saturation value. This may be important for estuaries, but
this effect is not included in the model at this time. The higher the salinity, the less amount of N 2 the wa-
ter can hold.
Once N2 gas is used in the model, the derived constituent TDG% is also turned ON and used in the any de-
rived constituent output. TDG, % is computed from
0.79 ∗ 𝑁2 𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛 0.21 ∗ 𝑂2 𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛

𝑇𝐷𝐺, % = ( + ) 100
𝑁2 𝑠𝑎𝑡𝑢𝑟𝑎𝑡𝑖𝑜𝑛 𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛 𝑂2 𝑠𝑎𝑡𝑢𝑟𝑎𝑡𝑖𝑜𝑛 𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛
N2 gas can be entrained at hydropower facilities and at dams. Algorithms for adding dissolved gas at these
facilities is outlined under Dissolved Oxygen/Dam Reaeration.
Water Quality 108

1.00
0.99
Atmospheric pressure, atm

0.98
0.97
P, RH=100%
0.96 P, RH=50%
P, RH=25%
0.95
0.94
0.93
0.92
0 20 40
Air temperature, C
Figure 66. Atmospheric pressure correction as a result of vapor pressure correction assuming original
atmospheric pressure is 1 atm.
Volatile Generic Constituent

The loss of a volatile generic constituent, such as a hydrocarbon, by volatilization from the water surface
is simulated using the same algorithms as used for N2.
The rate at which a generic constituent molecule enters and leaves the water surface is described by
𝐴𝑠𝑢𝑟 1
𝑆𝑣𝑜𝑙 = 𝐾 (ɸ𝑠𝑔 − ɸ𝑔 ) = 𝐾 (ɸ𝑠𝑔 − ɸ𝑔 )
𝑉𝑜𝑙𝑐𝑒𝑙𝑙 𝐿𝑔 ℎ𝐾𝑇 𝐿𝑔
where:
KLg = the surface gas transfer coefficient for a generic constituent (m/s)
Asur = the surface area of the surface layer computational cell (m2)
Volcell = the volume of the surface layer computational cell (m3)
hKT = depth of the surface layer (K=KT) computational cell (m)
ɸg = the liquid phase generic constituent concentration (g/m3)
ɸsg = the hypothetical saturation liquid phase generic constituent concentration if in equilibrium with the
gas concentration (g/m3) based on the gas phase concentration and the Henry’s law constant.
The gas transfer rate coefficient (KLg) is an overall gas transfer rate coefficient, reflecting the composite
effect of the individual gas transfer rate constants in both interfacial liquid and gas zones and of Henry’s
law constant. The gas transfer rate of a generic constituent at the water surface can be estimated based
on the reaeration rate for oxygen, as calculated using
𝐷𝑔
𝐾𝐿𝑔 = 𝐾𝐿𝑂2 √( )
𝐷𝑂2
109 Water Quality

where KLO2 is the reaeration coefficient for oxygen (m/s), Dg is the molecular diffusion coefficient for a ge-
neric constituent, and DO2 is the molecular diffusion coefficient for oxygen. The molecular diffusion coeffi-
cients can be related to their molecular weights as described by
𝐷𝑔 𝑀𝑊
𝐷𝑂2
= √ 𝑀𝑊𝑂2
𝑔
The above 2 equations can be combined to generalize the reaeration coefficient for any generic constitu-
ent based on its molecular weight and the reaeration coefficient for oxygen
𝑀𝑊
𝐾𝐿𝑔 = 𝐾𝐿𝑂2 ( 𝑀𝑊𝑂2 )0.25 = 𝐾𝐿𝑂2 × 𝐶𝐺𝐾𝐿𝐹
𝑔
where CGKLF is the unitless user defined gas transfer factor that indicates the fraction of the surface repa-
ration coefficient, KLO2, for dissolved oxygen gas transfer, as calculated in
𝑀𝑊
( 𝑀𝑊𝑂2 )0.25 = 𝐶𝐺𝐾𝐿𝐹
𝑔
Therefore, the only values required to estimate hydrocarbon volatilization are the molecular weights of
oxygen and the generic constituent of interest as well as the reaeration coefficient for oxygen, which is
internally calculated in the gas transfer subroutine.
CGKLF values for common hydrocarbons as well as for nitrogen and carbon dioxide are shown in Table 10.
The values are calculated based on their respective molecular weights and the molecular weight of oxy-
gen (32.00 g/mol).
Table 10. CGKLF input values for hydrocarbons, CO2, and N2.
Chemical MW (g/mole) CGKLF

Nitrogen 28.01 1.034
Carbon Dioxide 44 0.923
Benzene 78.11 0.800
MTBE 88.15 0.776
Toluene 92.13 0.768
Ethylbenzene 106.16 0.741
For a volatile hydrocarbon, it can often be assumed that the saturation liquid phase concentration (ɸ’g) is
negligible, since there is no significant concentration of the hydrocarbon in the gas phase (atmosphere).
Therefore the modified source/sink equation for a volatile generic constituent that includes volatilization
from the top layer (water surface) along with possible zero and first order decay is described by
(𝑇−20) (𝑇−20) 1
𝑆𝑔𝑉𝑜𝑙 = −𝑘0 𝜃𝑔 − 𝑘1 𝜃𝑔 ɸ𝑔 − 𝐾𝐿𝑂2 𝐶𝐺𝐾𝐿𝐹ɸ𝑔
ℎ𝐾𝑇
Where SgVol is the source/sink term for the volatile generic constituent concentration (gm-3s-1), ɸg is the
volatile generic constituent concentration (g/m3), and other terms are as defined previously.
Water Age or Residence time

Setting the zero-order decay rate to –1 day-1 and zeroing out all other generic constituent kinetic parame-
ters results in a state variable that increases by 1 day-1, which is an exact representation of water age or
hydraulic residence time. This is a very useful state variable when looking at hydrodynamics.
Water Quality 110

KINETICS INORGANIC SUSPENDED SOLIDS
𝑆𝑎𝑔𝑒 = −1 𝑑𝑎𝑦 −1 (4-67)
An example of water age for a reservoir in Washington, USA is shown Figure 67, where new water comes
in the main inflow but acts as an interflow through the reservoir.
Figure 67. Water age in Chester Morse Lake, WA.
Inorganic Suspended Solids

Inorganic suspended solids [ISS] are important in water quality simulations because of their influence on
density, light penetration, and nutrient availability. Increased solids concentrations reduce light penetra-
tion in the water column thus affecting temperature that in turn affects biological and chemical reaction
rates. Dissolved phosphorus and silica concentrations can also be affected by solids through sorption and
settling. Light and nutrient availability largely control algal production.
The settling velocity of each inorganic suspended solids compartment is a user-defined parameter. Usually
this is determined from Stoke’s settling velocity for a particular sediment diameter and specific gravity. Any
number of inorganic suspended solids groups can be modeled.
111 Water Quality

LABILE DOM KINETICS
Silica Phosphorus
sorption
Inorganic Suspended Solids
settling
system loss
Figure 68. Internal flux for inorganic suspended solids.
Referring to Figure 68, the rate equation for inorganic suspended solids is:
𝜕𝛷𝐼𝑆𝑆
𝑆𝐼𝑆𝑆 = 𝜔𝐼𝑆𝑆
𝜕𝑧
(4-68)
where:
z = layer thickness, m
ISS = settling velocity, m sec-1
ISS = inorganic suspended solids concentration, g m-3
In the finite difference representation of suspended solids concentrations, solids settling from layer [K]-1
serve as a source for the layer below it [K]. Lateral averaging results in homogeneous solids concentrations
laterally. Concentrations of suspended solids, though, generally decrease with distance away from the
dominant flow path, and this effect is not included in the model. The rate term for inorganic suspended
solids is evaluated in the subroutine SUSPENDED_SOLIDS.
Total Dissolved Solids or Salinity

Total dissolved solids (TDS) affect water density and ionic strength, thereby affecting water movements,
pH, and the distribution of carbonate species. Dissolved solids are normally expressed as TDS in freshwater
applications. Estuarine applications normally use salinity. Either TDS or salinity can be used with the choice
indicated by the parameter [WTYPEC] specified in the control file. The choice is then reflected in the com-
putation of density and ionic strength. If TDS is used, the units are g m-3, while salinity is kg m-3. It is
important to keep in mind that TDS and salinity are not necessarily equivalent. In the model, both are
treated conservatively with the rate term set to zero.
Labile DOM
Because of the importance of dissolved oxygen in aquatic systems, all constituents exerting an oxygen de-
mand must be included in kinetic formulations. This demand is often measured in rivers as the biochemical
oxygen demand (BOD), which includes microbial respiration and metabolism of various organic and inor-
ganic compounds. However, production of these materials occurs as well as decomposition, requiring the
major components of BOD be modeled individually. One of these constituents is dissolved organic matter
Appendix B Water Quality 112

KINETICS REFRACTORY POM
(DOM), which is composed of labile and refractory components. DOM is modeled as two separate com-
partments because of the different decomposition rates of the two groups.
Epiphyton
Macrophytes Inorganic C
excretion mortality
Phosphate
mortality
Algae excretion Labile DOM decay
Ammonium
decay
RDOM
Dissolved Oxygen
Figure 69. Internal flux between labile DOM and other compartments
Referring to Figure 69, the rate equation for labile DOM is:
S LDOM =  K ae a +  (1 − Pam ) K am  a +  K ee  e
     
algal excretion algal mortality epiphyton excretion
+  (1 − Pem ) K em  e −  OM K LDOM  LDOM − K L→ R  LDOM +  (1 − Pmm ) K mm  macro

            
epiphyton mortality labiole DOM decay labile to refractory macrophyte mortality/excretion
DOM decay
(4-69)
where:
Pam = pattern coefficient for algal mortality
Pem = pattern coefficient for epiphyton mortality
Pmm = partitioning coefficient for macrophyte mortality
OM = temperature rate multiplier for organic matter decay
Kae = algal excretion rate, sec-1
Kam = algal mortality rate, sec-1
Kee = epiphyton excretion rate, sec-1
Kem = epiphyton mortality rate, sec-1
Kmm = macrophyte mortality rate, sec-1
KLDOM = labile DOM decay rate, sec-1
KL→R = labile to refractory DOM transfer rate, sec-1
a = algal concentration, g m-3
e = epiphyton concentration, g m-3
LDOM = labile DOM concentration, g m-3
macro = marophyte concentration, g m-3
113 Water Quality

LABILE POM KINETICS
Refractory DOM
Refractory DOM is composed of compounds in the aquatic environment that slowly decompose exerting
oxygen demand over long periods. Internally, refractory DOM is produced from the decomposition of labile
DOM.
decay Inorganic C
Labile DOM
Phosphate
Refractory DOM
decay
Ammonium
Dissolved Oxygen
decay
Figure 70. Internal flux between refractory DOM and other compartments.
Referring to Figure 70, the rate equation for refractory DOM is:
S RDOM = K → R  LDOM −  OM K RDOM  RDOM

L     (4-70)
labile to refractory decay
DOM decay
where:
OM = temperature rate multiplier
KRDOM = refractory DOM decay rate, sec-1
KL R = transfer rate from labile DOM, sec-1
RDOM = refractory DOM concentration, g m-3
and the rate terms are evaluated in subroutine REFRACTORY_DOM.
Labile Particulate Organic Matter

Labile particulate organic matter (LPOM) represents particulate organic material in the water column.
When decaying, particulate organic matter is a source of refractory particulate organic matter, nitrogen,
phosphorus, and inorganic carbon. A stoichiometric relationship is used for mineralization of ammonium,
phosphorus, and inorganic carbon, and an oxygen demand is exerted as LPOM decomposes. When LPOM
settles to the bottom, it accumulates and decays in the sediment compartment or the sediment diagenesis
compartment if they are activated for the simulation.
Water Quality 114

KINETICS CBOD
Figure 71. Internal flux between Labile POM and other compartments.
Referring to Figure 71, the rate equation for LPOM is:
S LPOM =  Pam K am  a +  Pem K em e − K LPOM  OM  LPOM

      
algal mortality epiphyton mortality decay

− K L→ R  LPOM −  POM  LPOM +  Pmm K mm  macro +  K zm  zoo +
z

labile to refractory  macrophyte
 
mortality zooplankton mortality
POM decay settling
  pom 
 K (1 − Z ) − K
 
   LPOM  zoo
zg effic zoo
    +   +  
zg 

zoo zoo a lg a pom LPOM 
zooplankton excretion   
zooplankton ingestion of LPOM
(4-71)
where:
Pam = partition coefficient for algal mortality
Pem = partition coefficient for epiphyton mortality
Pmm = partition coefficient for macrophyte mortality
Pzoopp = partition coefficient for zooplankton LPOM grazing
OM = temperature rate multiplier for organic matter
POM = POM settling rate, m sec-1
Kmz = zooplankton mortality rate, sec-1
Kzg = zooplankton growth rate, sec-1
KLPOM = labile POM decay rate, sec-1
115 Water Quality

LABILE POM KINETICS
Keffic = Zooplankton ingestion efficiency

KL→R = transfer rate from labile POM to refractory POM, sec-1
e = periphyton/epiphyton concentration, g m-3
macro = macrophyte concentration, g m-3
LPOM = detritus concentration, g m-3
zoo = zooplankton concentration, g m-3
pom = zooplankton preference fraction for particulate organic matter
alg = zooplankton preference fraction for algae
zoo = zooplankton preference fraction for zooplankton
and the rate terms are evaluated in subroutine POM. POM settling and accumulation in the sediment com-
partment is handled identically to the algal compartment.
Refractory Particulate Organic Matter

Refractory POM is slowly decaying non-living, organic matter that settles. The source/sink terms are first
order decay, the conversion of LPOM to RPOM, and sedimentation. The settled organic matter becomes
part of the first order sediment model or the sediment diagenesis model if they are active.
Labile POM
decay
Inorganic C
decay Phosphate
Dissolved Oxygen Refractory POM decay
Ammonium
settling
Sediment
Figure 72. Internal flux between refractory POM and other compartments.
Referring to Figure 72, the rate equation for labile POM is:
 RPOM
S RPOM = K L → R  LPOM −  OM K RPOM  RPOM −  RPOM (4-72)
  z
labile to refractory decay 
POM decay settling
where:
KL→R = transfer rate from labile POM to refractory POM, sec-1
KRPOM = refractory POM decay rate, sec-1
POM = POM settling velocity, m/sec-1
LPOM = labile POM concentration, g m-3
Water Quality 116

KINETICS CBOD
RPOM = refractory POM concentration, g m-3
Carbonaceous Biochemical Oxygen Demand (CBOD)

Any number of CBOD groups with varying decay rates and either particulate or dissolved can be modeled
allowing the user to more accurately characterize various CBOD sources. Additionally, different CBOD
sources can be tracked separately in the model to determine what affect they have on the receiving water.
Care must be taken when including CBOD in the simulation to ensure that CBOD, DOM, POM, and algal
biomass are properly accounted for. CBOD is typically specified as allochthonous inputs and the forms of
autochthonous organic matter are kept track of in the various organic matter compartments. This ensures
that no “double dipping” occurs. This group can also model dissolved or particulate CBOD by specifying a
zero or a finite settling velocity, respectively, for the group. The settled CBOD goes into the 1st order sedi-
ment compartment.
In Version 3.6 and earlier, Figure 73 described the CBOD cycle with fixed stoichiometry of the CBOD in terms
of C:N:P. Starting with Version 3.7, code changes were made such that for each CBOD group, the nitrogen
and phosphorus corresponding to the CBOD group were modeled as separate constituents. One can still
though use the V3.6 and earlier fixed N and P stoichiometry for the BOD groups if one wishes. CBOD-C
though is still based on a fixed C stoichiometry.
Inorganic C
Phosphate
CBOD decay
Ammonium
decay
Dissolved Settling
Oxygen
Figure 73. Internal flux between CBOD and other compartments in Version 3.6 and earlier.
The following figure illustrates the CBOD cycle for Version 3.7 and later.
117 Water Quality

LABILE POM KINETICS
CBOD Inorganic C
decay
decay
Dissolved Settling
Oxygen
Figure 74. Internal flux between CBOD and other compartments in Version 3.7 and later.
The rate equation for CBOD is:

R 
S CBOD = −  T − 20 K BOD RBOD CBOD−CBOD BOD CBOD
   z (4-73)
decay  
settling
where:
 = BOD temperature rate multiplier
T = temperature, C
CBOD = CBOD settling velocity, m/sec-1
KBOD = CBOD decay rate, sec-1
RBOD = CBOD conversion from input CBOD to CBOD-ultimate (CBODu)
BOD = CBOD concentration, g m-3
Note that the user can enter the 5-day CBOD5 or any other day CBODday into the CE-QUAL-W2 model. The
RBOD term converts to CBODu (ultimate or 28 day) internally in the model. The model though will output
CBOD5 if you entered CBOD5 as input - not CBODu. If you use CBODu as the input CBOD, then RBOD=1 and the
output would be CBODu.
𝐶𝐵𝑂𝐷𝑢
Typical determination of RBOD is based on the following standard CBOD equation: 𝑅𝐵𝑂𝐷 = =
𝐶𝐵𝑂𝐷𝑡
1
(1−𝑒𝑥𝑝[−𝐾𝐵𝑂𝐷 𝑡])
where t is the time in days (5 days, for example), KBOD is the BOD decay rate in day-1, and
CBODt is the CBOD at that time.
Carbonaceous Biochemical Oxygen Demand - Phos-

phorus (CBODP)
The phosphorus associated with a specific CBOD group is modeled as a separate constituent.
Sources and sinks for carbonaceous biochemical oxygen demand - phosphorus (CBODP) are
shown in Figure 75.
Water Quality 118

KINETICS CBOD
CBODP Phosphate
decay
Settling
Figure 75. Internal flux between CBODP and other compartments.
The rate equation for CBODP is:  CBODP

BOD CBODP− CBOD
T − 20
S CBODP = − 
 K
  
decay  z
settling
where:
CBOD = CBOD settling velocity, m/sec-1
CBODP = CBODP concentration, g m-3
Carbonaceous Biochemical Oxygen Demand - Nitro-

gen (CBODN)
The nitrogen associated with a specific CBOD group is modeled as a separate constituent. Figure 76 shows
the CBODN sources and sinks.
CBODN Ammonium
decay
Settling
Figure 76. Internal flux between CBODN and other compartments.
119 Water Quality

LABILE POM KINETICS
The rate equation for CBODN is:
 CBODN
BOD CBODN − CBOD
T − 20
S CBODN =−
 K
 
where:
 z
 = BOD temperature rate decay
settling
multiplier
CBOD = CBOD settling velocity, m s-1
CBODN=CBODN concentration, g m-3
Algae
An algal community can be represented as a single assemblage or as multiple algae groups. Often, the algal
groups might include diatoms, greens, and cyanobacteria. The user has complete freedom in how many
and what kinds of algal groups can be included in the simulation through careful specification of the kinetic
rate parameters that define the characteristics of each algal group.
mortality
Zooplankton excretion Labile DOM
photosynthesis grazing
mortality Labile POM
photosynthesis
Dissolved Oxygen Algae
settling Inorganic C
respiration
Ammonium
Sediment
Phosphate
Silica
Figure 77. Internal flux between algae and other compartments.
Referring to Figure 77, the rate equation for each algal group is:
 a
S a = K ag  a − 
K ar  a − K ae  a − K am  a −  a
       (4-74)

growth respiration excretion mortality z
settling
  a lg  a 
−   Z   zoo 

   a lg  a

+  pom  lpom +   
zoo  zoo 
net loss to grazing
Water Quality 120

KINETICS ALGAE
where:
z = cell height
Z  = net growth rate of a zooplankton species
𝜎𝑎𝑙𝑔 , 𝜎𝑧𝑜𝑜 = zooplankton grazing preference factors for algae (alg) or zooplankton (zoo)
Kag = algal growth rate, sec-1
Kar = algal dark respiration rate, sec-1
a = algal settling rate, m sec-1
Chlorophyll a (chl a) is most commonly available as an estimate of algal biomass. To convert chl a to algal
biomass, chl a is typically multiplied by the given algae (as g m-3 or mg/l dry weight OM)/chl a (as µg chlo-
rophyll a/l) ratio. This value can vary widely depending on the makeup of the algal population. Some pre-
vious studies determined the conversion factor by regressing particulate organic matter with chlorophyll a.
Algal growth rate is computed by modifying a maximum growth rate affected by temperature, light, and
nutrient availability:
K ag =  ar  af  min K ag max (4-75)
where:
ar = temperature rate multiplier for rising limb of curve
af = temperature rate multiplier for falling limb of curve
min = multiplier for limiting growth factor (minimum of light, phosphorus, silica, and nitrogen) be-
tween 0 and 1
Kagmax = maximum algal growth rate, sec-1
Rate multipliers for algal growth are computed based upon available light, phosphorus, nitrogen, and silica.
The rate multiplier, , between 0 and 1 for light is based upon the Steele (1962) function:
𝐼
𝐼 − +1
𝜆𝑙 = 𝐹 ( 𝐼 ) = 𝑒 𝐼𝑠 (4-76)
𝐼𝑠
where:
I = available light, W m-2
Is = saturating light intensity at maximum photosynthetic rate, W m-2
l = light limiting factor or F(I)
The above expression allows for simulation of photoinhibition at light intensities greater than the saturation
value. However, light penetration decreases with depth:
𝐼(𝑧) = (1 − 𝛽)𝐼0 𝑒 −𝛼𝑧 (4-77)

where:
I0 = solar radiation at the water surface, W m-2
α = attenuation coefficient, m-2
z = depth, m
121 Water Quality

ALGAE KINETICS
(1-β) = fraction of solar radiation absorbed at the water surface
The average effect of light on algal growth in a particular model cell can be obtained by combining the
above two expressions and integrating over the cell depth to obtain (Chapra and Reckhow, 1983):
l =
e
e - 2
- e- 1 (4-78)
 z where:
(1 -  ) I O
 1= e
- d
Is
(1 -  ) I O
 2= e
 (d +  z)
Is
d = depth at top of model cell, m

The fraction of solar radiation, (1-β)Io, is added directly to the surface layer. The attenuation coefficient, α,
consists of a baseline value [EXH2O] to which the effects of inorganic [EXINOR] and organic [EXORG] sus-
pended solids, and algae [EXA] are added.
Rate multipliers limiting maximum algal growth due to nutrient limitations are computed using the Monod
relationship:
i
i = (4-79)
Pi +  i
where:
i = phosphorus or nitrate + ammonium concentration, g m-3
Pi = half-saturation coefficient for phosphorus or nitrate + ammonium, g m-3
The algal nitrogen preference for ammonium is based upon the following (Thomann and Fitzpatrick, 1982)
equation:
 NOx
PNH 4 =  NH 4
(K NH 4 + Φ NH 4 )(K NH 4 + Φ NOx )
K NH 4
+  NH 4
(Φ NH 4 + Φ NOx )(K NH 4 + Φ NOx )
(4-80)
where:
PNH4 = ammonium preference factor
KNH4 = ammonia preference half-saturation coefficient, g m-3
NH4 = ammonium concentration, g m-3
NOx = nitrate-nitrite concentration, g m-3
Water Quality 122

KINETICS ALGAE
This allows algae to use primarily ammonium and gradually switch to nitrate as ammonium concentrations
decrease.
Algal dark respiration is computed using the rising limb of the temperature function:
K ar =  ar  af K armax (4-81)
where:
Karmax = maximum dark respiration rate, sec-1
Algal photorespiration (excretion) is evaluated using an inverse relation to the light rate multiplier:
Kae = (1 − λl )  ar  af Kaemax
(4-82)
where:
Kaemax = maximum excretion rate constant, sec-1
l= light limiting factor
Excretion rates increase at both low and high light intensities, with excretion products contributing to labile
DOM.
Algal mortality is defined as:
Kam =  ar  af Kammax
(4-83)
where:
Kammax = maximum mortality rate, sec-1
This mortality rate represents both natural and predator mortality. Algal growth does not occur in the
absence of light. Algal growth is not allowed to exceed the limit imposed by nutrient supply over a given
timestep. Algal excretion is not allowed to exceed algal growth rates.
Similar to inorganic solids, settling algae serve as a source for the layer below. Unlike inorganic solids, algae
passing to the sediments accumulate within the sediment compartment. POM is also accumulated in this
sediment compartment.
Periphyton/Epiphyton
Any number of user defined periphyton/epiphyton groups can be modeled. Similar to the 1st order sedi-
ment compartment, periphyton/epiphyton are not transported in the water column.
123 Water Quality

EPIPHYTON KINETICS
mortality
excretion Labile DOM
photosynthesis
mortality Labile POM
photosynthesis
Dissolved Oxygen Epiphyton
burial Inorganic C
respiration
Ammonium
Sediment
Phosphate
Silica
Figure 78. Internal flux between epiphyton and other compartments.
Referring to Figure 78, the rate equation for each epiphyton/periphyton group is:
S e = Keg e − K e r  e − K e e  e − K e m  e − K eb  e
     
growth respiration excretion mortality burial (4-84)
where:
Keg = epiphyton growth rate, sec-1
Ker = epiphyton dark respiration rate, sec-1
Keb = epiphyton burial rate, sec-1
Epiphyton/periphyton growth rate is computed by modifying a maximum growth rate affected by epiphy-
ton/periphyton biomass, temperature, and nutrient availability:
𝐾𝑒𝑔 = 𝛾𝑒𝑟 𝛾𝑒𝑓 𝜆𝑚𝑖𝑛 𝐾𝑒𝑔 𝑚𝑎𝑥 (4-85)
where:
er = temperature rate multiplier for rising limb of curve
ef = temperature rate multiplier for falling limb of curve
min = multiplier for limiting growth factor (minimum of phosphorus, silica, nitrogen, and epiphy-
ton/periphyton biomass) between 0 and 1
Keg = epiphyton/periphyton growth rate, sec-1
Kegmax = maximum epiphyton/periphyton growth rate, sec-1
Rate multipliers for epiphyton/periphyton growth are computed based upon available light, phosphorus,
nitrogen, silica, and epiphyton biomass. Epiphyton/periphyton biomass is included as a surrogate for light
limited epiphyton/periphyton self-shading and will be discussed in greater detail below.
The rate multiplier for light is based upon the Steele (1962) function:
Water Quality 124

KINETICS EPIPHYTON
𝐼
𝐼 − +1
𝜆𝑙 = 𝐹 ( 𝐼 ) = 𝑒 𝐼𝑠 (4-86)
𝐼𝑠
where:
I = available light, W m-2
Is = saturating light intensity at maximum photosynthetic rate, W m-2
l = light limiting factor or F(I)
The above expression allows for simulation of photoinhibition at light intensities greater than the saturation
value. However, light penetration decreases with depth:
𝐼(𝑧) = (1 − 𝛽)𝐼0 𝑒 −𝛼𝑧 (4-87)
where:
I0 = solar radiation at the water surface, W m-2
α = attenuation coefficient, m-2
z = depth, m
β = fraction of solar radiation absorbed at the water surface
The average effect of light on epiphyton/periphyton growth in a model cell can be obtained by combining
the above two expressions and integrating over the cell depth to obtain (Chapra and Reckhow, 1983):
l =
e
e - 2
- e- 1 (4-88)
 z (1 -  ) IO
where:
1= e
- d
Is
(1 -  ) IO
2= e
 (d +  z)
Is
d =depth at the top of computational cell, m
The attenuation coefficient, λ, is computed from a baseline value [EXH2O] to which the effects of inorganic
[EXINOR] and organic [EXORG] suspended solids, as well as the extinction of each algal group, are added.
Epiphyton/periphyton self-shading are accounted for in the biomass limitation formulation.
Rate multipliers limiting epiphyton growth due to nutrient limitations are computed using the Monod rela-
tionship:
i
i = (4-89)
Pi +  i
where:
i = phosphorus or nitrate + ammonium concentration, g m-3
Pi = half-saturation coefficient for phosphorus or nitrate + ammonium, g m-3
The epiphyton/periphyton preference for ammonium can be modeled using the following (Thomann and
Fitzpatrick, 1982):
125 Water Quality

EPIPHYTON KINETICS
𝛷𝑁𝑂𝑥 𝐾𝑁𝐻4
𝑃𝑁𝐻4 = 𝛷𝑁𝐻4 (𝐾 + 𝛷𝑁𝐻4 (𝛷 (4-90)
𝑁𝐻4 +𝛷𝑁𝐻4 )(𝐾𝑁𝐻4 +𝛷𝑁𝑂𝑥 ) 𝑁𝐻4 +𝛷𝑁𝑂𝑥 )(𝐾𝑁𝐻4 +𝛷𝑁𝑂𝑥 )

KNH4 = ammonia preference half-saturation coefficient, g m-3
NOx = nitrate-nitrite concentration, g m-3
Epiphyton/periphyton dark respiration is computed using the rising limb of the temperature function:
𝐾𝑒𝑟 = 𝛾𝑒𝑟 𝛾𝑒𝑓 𝐾𝑒𝑟 𝑚𝑎𝑥 (4-91)
where:
er = temperature rate multiplier for rising limb of the curve
er = temperature rate multiplier for falling limb of the curve
Karmax = maximum dark respiration rate, sec-1
Epiphyton/periphyton excretion is evaluated using an inverse relation to the light rate multiplier:
K ee = (1 -  l ) er  ef K ee max (4-92)
where:
l = light limiting factor
Keemax = maximum excretion rate constant, sec-1
Excretion rates increase at both low and high light intensities, with excretion products contributing to labile
DOM.
Epiphyton/periphyton mortality is defined as:
𝐾𝑒𝑚 = 𝛾𝑒𝑟 𝛾𝑒𝑓 𝐾𝑒𝑚 𝑚𝑎𝑥 (4-93)
where:
Kemmax = maximum mortality rate, sec-1
This mortality rate represents both natural and predator mortality. Epiphyton/periphyton growth does not
occur in the absence of light. Epiphyton/periphyton growth is not allowed to exceed the limit imposed by
nutrient supply over a given timestep. Epiphyton/periphyton excretion is not allowed to exceed epiphy-
ton/periphyton growth rates.
The epiphyton/periphyton burial rate represents the burial of dead epiphyton/periphyton to the organic
sediment compartment. The epiphyton/periphyton become part of the 1st-order sediment compartment.
Epiphyton/periphyton though that die (mortality) become a part of the labile particulate organic matter
and the labile dissolved organic matter pool. The user defines the fraction of the dead epiphyton (EPOM)
that goes to the LPOM pool. This POM is then transported in the water column. Currently, there is no
Water Quality 126

KINETICS EPIPHYTON
sloughing of epiphyton/periphyton into the water column as a function of velocity shear. This is a function
of the biomass limitation term.
Figure 79. How epiphyton/periphyton transfer organic matter to the water column at death and to the
sediment during burial.
The epiphyton/periphyton biomass is controlled by a biomass limitation equation based on Monod kinetics.
The biomass limitation function, f [BLIM], varies from 0 to 1 and is multiplied with the growth rate. This
function is defined as
𝐵
𝑓 = [1 − 𝐵+𝐾 ] (4-94)
𝐵
where:
B = epiphyton areal biomass, g/m2 [EPD]
Kb = epiphyton areal biomass half-saturation coefficient, g/m2 [EHS]
𝑉
The areal biomass is calculated as 𝐵 = 𝛷𝑒 (4-95)
𝐴
where:
A = computational cell surface area including side walls and bottom area, m2
V = computational cell volume, m3
The biomass limitation is a surrogate calibration parameter for light limitation due to self-shading.
Macrophytes
The macrophyte model consists of two parts: a section describing the water quality compartment and a
section describing the hydrodynamic compartment. Bother of these sections are from Berger (2000) and
Berger and Wells (2008). Chris Berger is the primary author of this section.
127 Water Quality

KINETICS
The macrophyte model was designed to simulate multiple submerged macrophyte species. It does not
differentiate between plant parts. The nutrient fluxes for the water quality component of the macrophyte
compartment are shown in Figure 80. Light, temperature, carbon dioxide, ammonia-nitrogen (only ammo-
nia is used as a N source for macrophytes), and ortho-phosphorus may limit growth. Depending on the
macrophyte species, nitrogen and phosphorus may be obtained from the sediments or the water column.
If they are obtained from the sediments, the sediments are assumed to be an infinite pool that cannot limit
growth. Plants grow upwards from the sediment through model layers. Growth upward is accomplished by
moving the growth of a layer to the layer above if the concentration in the layer is greater than a threshold
concentration and the concentration in the upper layer is less than the same threshold concentration. Mac-
rophyte shading is modeled by making light attenuation a function of macrophyte concentration.
drawdown
Inorganic
Carbon
Labile Dissolved PO4-P

Organic Matter
mortality
NH4-N
photosynthesis
Macrophytes
respiration
Sediment P
respiration
Oxygen
Sediment N
mortality photosynthesis
Labile Particulate Refractory Particulate

Organic Matter Organic Matter
Figure 80. Nutrient fluxes for the macrophyte compartment in CE-QUAL-W2.
The macrophyte growth rate is modeled as follows:

S macro =  m max f ( I , N , P, C ) 1 2  macro − K mr  1 macro − K mm  macro
   
growth respiration mortality (4-96)
where:
𝑆𝑚𝑎𝑐𝑟𝑜 : macrophyte growth rate density (g/m3/s)
I : solar radiation (W/m2)
𝑓(𝐼, 𝑁, 𝑃, 𝐶): Growth limiting function between 0 and 1
𝜇𝑚 𝑚𝑎𝑥 : maximum macrophyte growth rate (day-1)
Kmr: maximum respiration rate (day-1)
Water Quality 128

KINETICS
Kmm: mortality/excretion rate (day-1)

𝛾1 : ascending temperature rate multiplier
𝛾2 : descending temperature rate multiplier
𝛾1 : growth limiting factor due to photosynthesis
macro: macrophyte concentration (g/m3)
Growth rate, respiration rate, and mortality/excretion rate are temperature dependent. Temperature ef-
fects are modeled using the equations developed by Thornton and Lessem (1978) which are currently used
in the phytoplankton compartment of CE-QUAL-W2. The growth limiting function f(I,N,P,C) is the minimum
of the light f(I), nitrogen f(N), phosphorus f(P), and carbon f(C) limiting functions such that 𝑓(𝐼, 𝑁, 𝑃, 𝐶) =
𝑚𝑖𝑛𝑖𝑚𝑢𝑚(𝑓(𝐼), 𝑓(𝑁), 𝑓(𝑃), 𝑓(𝐶)).
All the limiting functions are unit-less and have a value between 0 and 1. The limiting functions for the
nutrients have the following Michaelis-Menten form
𝑆
𝑓(𝑆) =
𝐾𝑆 +𝑆
(4-97)
where S (mg/l) is the nutrient concentration and Ks (mg/l) is the half-saturation concentration. Light limi-
tation was modeled with a hyperbolic equation which as the same form as the Michaelis-Menten func-
tion:
𝐼
𝑓(𝐼) = 𝐼 +𝐼ℎ
(4-98)
2 2
where I: solar radiation (W/m ), Ih: half-saturation coefficient for solar radiation (W/m ). This function is
frequently used in the absence of photo-inhibition (Carr et al., 1997).
The net light extinction coefficient𝛾 (m-1) was modeled as a function of macrophyte plant tissue concen-
tration giving
𝛾 = 𝜀𝐻2 𝑂 + 𝜀𝑖𝑠𝑠 𝛷𝑖𝑠𝑠 + 𝜀𝑜𝑠𝑠 𝛷𝑜𝑠𝑠 + ∑of 𝜀𝑚𝑎𝑐 𝛷𝑚𝑎𝑐 + ∑ of 𝜀𝑎 𝑙𝑔 𝑎𝑒 𝛷𝑎 𝑙𝑔 𝑎𝑒 (4-99)
macrophyte algae
groups groups
where:
𝜀𝐻2𝑂 = light extinction exclusive of suspended solids, m-1
𝜀𝑖𝑠𝑠 = light extinction due to inorganic suspended solids, m3 m-1 g-1
𝜀𝑜𝑠𝑠 = light extinction due to non-living organic suspended solids, m3 m-1 g-1
𝜀𝑚𝑎𝑐 = light extinction due to macrophytes, m3 m-1 g-1
𝜀𝑎 𝑙𝑔 𝑎𝑒 = light extinction due to algae, m3 m-1 g-1
𝛷𝑖𝑠𝑠 = inorganic suspended solids concentration, g m-3
𝛷𝑜𝑠𝑠 = organic suspended solids concentration, g m-3
𝛷𝑚𝑎𝑐 = macrophyte plant tissue concentration, g m-3
𝛷𝑎 𝑙𝑔 𝑎𝑒 = algae concentration, g m-3
Modeling of flow through macrophytes incorporated the following concepts:
• Porosity of the macrophytes was calculated through determination of the blockage area of the vege-
tation normal to the direction of flow
• The drag of individual stems and leaves were totaled to determine the total drag force in a model cell
• The effective Mannings 𝑛 was calculated by combining the effect of bed shear and the drag force on
the plants
129 Water Quality

KINETICS
Modeling Frictional Force

The total frictional force f was partitioned into a bottom friction component 𝑓𝑏 and a vegetation drag
component 𝑓𝑣 giving
𝑓 = 𝑓𝑏 + 𝑓𝑣 (4-100)
Bottom shear 𝜏𝑏 was simulated using Manning’s friction factor
𝜌𝑤 𝑔𝑛2
𝜏𝑏 = 1 𝑈|𝑈| (4-101)
𝑅 ⁄3
where:
𝜌𝑤 : density of water
𝑔: gravitational constant
𝑈: water velocity
𝑅: hydraulic radius
n: Manning’s friction factor.
The Columbia Slough model has already been calibrated for the no macrophytes condition. Manning’s
friction factor for the model segments was typically around 0.03.
Vegetative drag caused by macrophytes was modeled in a manner similar to that used by Petryk and
Bosmajian (1975) where the drag force 𝐷𝑖 on the ith plant is
𝜌𝑤 𝑈 2
𝐷𝑖 = 𝐶𝑑 𝐴𝑖 ( )
2
where:
𝐴𝑖 : area of plant projected normal to the direction of flow
𝐶𝑑 : drag coefficient.
The total drag force in a model cell due to vegetation is

𝜌𝑤 𝑈 2
∑𝐷𝑖 = 𝐶𝑑 ( ) ∑𝐴𝑖
2
The total plant area normal to the direction of flow ∑𝐴𝑖 was estimated using biomass to surface area ra-
tios from Sher-Kaul et al. (1995) and surface area to volume ratios from Sand-Jensen and Borum (1991).
The projected area normal to flow would then be 𝜀∑𝐴𝑖 and the total drag was then
𝜌𝑤 𝑈 2
∑𝐷𝑖 = 𝐶𝑑 ( ) ∑𝐴𝑖
2
The drag coefficient 𝐶𝑑 was a calibration parameter but has been shown to be of the order of 1.0 for vege-
tation (Hsi and Nath, 1968; Hoerner, 1965; Petryk, 1969).
The effective Mannings 𝑛 of each model cell was calculated in the manner used by Petryk and Bosmajian
(1975) with
𝐶𝑑 ∑𝐴𝑖 1 4⁄
𝑛 = 𝑛𝑏 √1 + 2
2𝑔𝐴𝐿 𝑛𝑏
𝑅 3 (4-102)
where
𝑛𝑏 : Mannings friction factor due to bed shear only.
This derivation is shown in more detail in Berger and Wells (2008).
Water Quality 130

KINETICS
Modeling Porosity
The volume 𝑉𝑚 of macrophytes can be estimated by dividing the macrophyte mass in a model cell 𝑚 by
density 𝜌𝑚
𝑚
𝑉𝑚 =
𝜌𝑚
The porosity 𝜑 was estimated by dividing volume within a cell free of macrophytes by the cell total vol-
ume 𝑉 giving
𝑉 − 𝑉𝑚
𝜑=( )
𝑉
The cross-sectional area of each model cell was multiplied by the porosity to calculate the effective cross-
sectional area. The porosity affected both the continuity and momentum equations.
Changes to Governing Equations

Several of the governing equations have been altered to account for porosity and the frictional effects of
macrophytes. Equations affected include the x-momentum equation, the continuity equation, the free
water surface equation, and the constituent transport equation. The new x-momentum equations is
𝜕𝑈𝜑𝐵 𝜕𝑈𝑈𝜑𝐵 𝜕𝑊𝑈𝜑𝐵 𝜑𝐵 𝜕𝑃 1 𝜕𝜑𝐵𝜏𝑥𝑥 1 𝜕𝜑𝐵𝜏𝑥𝑧

+ + =− + + (4-103)
𝜕𝑡 𝜕𝑥 𝜕𝑧 𝜌 𝜕𝑥 𝜌 𝜕𝑥 𝜌 𝜕𝑧
where:
𝑈 :x-direction velocity, m/sec
𝑊 :z–direction velocity, m/sec
𝐵 :channel width, meters
𝜌 :density, mg/l
𝑃 :Pressure, Newtons/m2
𝜏𝑥𝑧 :vertical shear stress, Newtons/m2
𝜏𝑥𝑥 : longitudinal shear stress, Newtons/m2
: porosity
Vertical shear stress 𝜏𝑥𝑧 is a function of interfacial shear stress, shear stress due to wind, and bottom and
plant shear stress. It was determined from
𝜏𝑥𝑧 𝜕𝑈 𝜏𝑤𝑥 −𝑑𝑘𝑧 𝜏𝑏𝑚

𝜌
= 𝐴𝑧 𝜕𝑧 + 𝜌
𝑒 + 𝜌
(4-104)
where:
𝜏𝑏𝑚 -bottom and plant shear stress
𝐴𝑧 - turbulent eddy viscosity
𝜏𝑤𝑥 -wind shear stress
k-wave number.
Bottom and plant shear stress is calculated using the effective Mannings 𝑛 determined above
𝜌𝑔𝑛2
𝜏𝑏𝑚 = 1 𝑈|𝑈| (4-105)
𝑅 ⁄3
The continuity equation was changed to
131 Water Quality

KINETICS
𝜕𝑈𝜑𝐵 𝜕𝑊𝜑𝐵
𝜕𝑥
+ 𝜕𝑧
= 𝑞𝜑𝐵 (4-106)
where:
𝑞 - lateral inflow/outflow per unit volume (T-1).
The new free water surface equation is
𝜕𝐵𝜂 𝜑𝜂 𝜕 ℎ ℎ
𝜕𝑡
= 𝜕𝑥 ∫𝜂 𝑈𝜑𝐵𝑑𝑧 − ∫𝜂 𝑞𝜑𝐵𝑑𝑧 (4-107)
with
𝐵𝜂 – time and spatially varying surface width, meters
𝜂 - free water surface elevation, meters
The constituent transport equation could also be affected by the reduction of cross-sectional area due to
macrophytes giving
𝜕𝜑𝐵𝐶 𝜕𝑈𝜑𝐵𝐶 𝜕𝑊𝜑𝐵𝐶 𝜕 𝜕𝐶 𝜕 𝜕𝐶

+ + − 𝜑𝐵𝐷𝑥 − 𝜑𝐵𝐷𝑧 = 𝑞𝜑 𝜑𝐵 + 𝑆𝐾 𝜑𝐵 (4-108)
𝜕𝑡 𝜕𝑥 𝜕𝑧 𝜕𝑥 𝜕𝑥 𝜕𝑧 𝜕𝑧
where:
𝐶 - constituent concentration, mg/l
𝐷𝑥 - longitudinal temperature and constituent dispersion coefficient, m 2/sec
𝐷𝑧 - vertical temperature and constituent dispersion coefficient, m2/sec
𝑞𝜑 - lateral inflow or outflow mass flow rate of constituent per unit volume, mg/l/sec
𝑆𝐾 - kinetics source/sink term for constituent concentration, mg/l/sec
The parameter coefficients used in the first application of this model to the Columbia Slough are shown
below in Table 11. Descriptions of the input data fields are included in Part 3 of the User Manual.
Table 11. Parameters and values used for macrophytes in the Columbia Slough model.
Variable CE-QUAL-W2 Coefficient definition (units)

name in input Model value
data file
PRNMC ON or OFF Macrophyte snapshot output on or off
MACROC ON or OFF Macrophyte compartment on or off
EXM 0.01 Light extinction coefficient for organic particles (m/mg/l)
MBMP 40.0 Threshold macrophyte concentration for which growth is moved to the above
layer (mg/l)
MMAX 500.0 Maximum macrophyte concentration (mg/l)
MG 0.30 Maximum macrophyte growth rate (day-1)
MR 0.05 Macrophyte respiration rate (day-1)
MM 0.05 Macrophyte mortality rate (day-1)
SATM 20.0 Macrophyte half-saturation light intensity at the maximum photosynthetic
rate (Watts/m2)
MT1 7 Lower temperature bound for macrophyte growth (Celsius)
MT2 15 Lowest temperature at which macrophyte growth processes are near the max-
imum rate (Celsius)
Water Quality 132

KINETICS
Variable CE-QUAL-W2 Coefficient definition (units)

name in input Model value
data file
MT3 24 Upper temperature at which macrophyte growth processes are near the maxi-
mum rate (Celsius)
MT4 34 Upper lethal temperature for macrophytes(Celsius)
MK1 0.1 Temperature rate multiplier for MT1
MPOM 0.9 Fraction of dead macrophytes which becomes POM, the fraction (1-MPOM)
becomes labile DOM
LRPMAC 0.2 Fraction of POM which originates as dead macrophytes becoming labile POM
PSED 0.5 Fraction of phosphorus uptake by macrophytes obtained from sediments
NSED 0.5 Fraction of nitrogen uptake by macrophytes obtained from sediments
MHSP 0 Half-saturation constant for P uptake by macrophytes (mg/l)
MHSN 0 Half-saturation constant for N uptake by macrophytes (mg/l)
MHSC 10 Half-saturation constant for carbon uptake by macrophytes (mg/l)
MACP 0.005 Stoichiometric equivalent between macrophyte biomass and orthophosphate
MACN 0.08 Stoichiometric equivalent between macrophyte biomass and nitrogen
MACC 0.45 Stoichiometric equivalent between macrophyte biomass and carbon
02MR 1.1 Dissolved oxygen requirement for macrophyte respiration
02MG 2.0 Stoichiometric equivalent for dissolved oxygen production during macrophyte
photosynthesis
CD 3.0 Macrophyte drag coefficient
Zooplankton
A multiple zooplankton compartment was adapted from the U.S. Army Corps of Engineers reservoir model
CE-QUAL-R1 (Environmental Laboratory, 1995) for the CE-QUAL-W2 model. Zooplankton are assumed to
be non-motile and are transported only by advection and dispersion. Zooplankton can graze algae, detritus
(POM), and other zooplankton. Losses occur through mortality and respiration. The source/sink term for
zooplankton is shown below:
(∑ 𝜎alg 𝛷𝑎 + 𝜎𝑝𝑜𝑚 𝛷𝑙𝑝𝑜𝑚 + ∑ 𝜎zoo 𝛷zoo ) − 𝑍𝐿

𝑆𝑧𝑜𝑜 = 𝛾1 𝛾2 𝑍𝑒 𝐾zma𝑥 ( (1 − 𝛾2 )𝐾𝑧𝑚 𝛷𝑧𝑜𝑜
) 𝛷𝑧𝑜𝑜 − ⏟
⏟ (∑ 𝜎alg 𝛷𝑎 + 𝜎𝑝𝑜𝑚 𝛷𝑙𝑝𝑜𝑚 + ∑ 𝜎zoo 𝛷zoo ) + 𝑍1⁄2 mortality
growth
−⏟
𝛾1 𝐾𝑧𝑟 𝛷𝑧𝑜𝑜
respiration
where:
𝛷lpom = Labile particulate organic matter concentration (mg/l)
𝛷rpom = Refractory particulate organic matter concentration(mg/l)
𝛷𝑎 = Algae concentration (mg/l)
𝛷𝑧𝑜𝑜 = Zooplankton concentration (mg/l)
𝐾𝑧𝑚 = Zooplankton mortality rate (d-1)
𝐾𝑧𝑚𝑎𝑥 = Maximum ingestion rate for zooplankton (d-1)
133 Water Quality

KINETICS
𝐾𝑧𝑟 =Zooplankton respiration rate (d-1)

𝑍1⁄2 =Half-saturation coefficient for zooplankton ingestion (mg/l)
𝑍𝑒 =Zooplankton ingestion efficiency
𝑍𝐿 =Low threshold concentration for zooplankton feeding (mg/l)
𝛾1 =Temperature coefficient for rising limb of curve for zooplankton
𝛾2 =Temperature coefficient for falling limb of curve for zooplankton
𝜎𝑎𝑙𝑔 =Zooplankton preference fraction for algae
𝜎𝑧𝑜𝑜 =Zooplankton preference fraction for zooplankton
𝜎𝑝𝑜𝑚 =Zooplankton preference fraction for particulate organic matter
The zooplankton source/sinks are also illustrated in Figure 81. The growth rate is a function of temperature,
the maximum growth rate, and a modified Michaelis-Menten equation which includes a low threshold con-
centration below which zooplankton do not feed. At dissolved oxygen concentrations below 2 mg/l feeding
stops and the mortality rate is doubled. The zooplankton model coefficients described in Table 12.
Figure 81. Zooplankton source/sinks.
Table 12. Parameters used in the zooplankton model.
Name Description
ZMAX Maximum ingestion rate for zooplankton (1/day)
ZMORT Maximum nonpredatory mortality rate for zooplankton (1/day).
ZEFFIC Zooplankton assimilation efficiency or the proportion of food assimilated to food consumed (dimension-
less).
PREFA Preference factor of zooplankton for algae (dimensionless).
PREFP Preference factor of zooplankton for detritus (dimensionless).
ZRESP Maximum zooplankton respiration rate (1/day).
ZOOMIN Threshold food concentration at which zooplankton feeding begins (mg/l).
ZS2P Zooplankton half-saturation constant for food concentration (mg/l)
ZOOT1 Lower temperature for zooplankton growth (Celsius)
ZOOT2 Lower temperature for maximum zooplankton growth (Celsius)
ZOOT3 Upper temperature for maximum zooplankton growth (Celsius)
Water Quality 134

KINETICS
Name Description
ZOOT4 Upper temperature for zooplankton growth (Celsius)
ZOOK1 Fraction of zooplankton growth rate at ZOOT1 (dimensionless)
EXZ Zooplankton light extinction (m-1)
O2ZR Oxygen stoichiometry for zooplankton respiration
ZP Stoichiometric equivalent between zooplankton biomass and phosphorus
ZN Stoichiometric equivalent between zooplankton biomass and nitrogen
ZC Stoichiometric equivalent between zooplankton biomass and carbon
Phosphorus
Phosphorus is an important element in aquatic ecosystems since it serves as one of the primary nutrients
for phytoplankton growth. In many fresh waters, phosphorus is considered to be the nutrient limiting max-
imum production of phytoplankton biomass (Schindler, 1971; Schindler et al., 1973; Vollenweider, 1968,
1976).
Phosphorus is assumed to be completely available as ortho-phosphate (PO4) for uptake by phytoplankton.

Measurements of soluble reactive phosphorus are closest to the form used in the model. Macrophytes are
specified as either taking P from the sediments or from the water column.
Macrophytes
Algae Epiphyton Sediment
LDOM
photosynthesis respiration
respiration
decay RDOM
Phosphate
Zooplankton
LPOM
anaerobic adsorption/
release settling
RPOM
Sediments System Loss
CBOD
Figure 82. Internal flux between phosphorus and other compartments.
Referring to Figure 82, the rate equation for phosphorus is:
135 Water Quality

KINETICS
S p =  ( Kar - Kag )  P a a +  ( Ker - Keg )  P e e + K LDOM  POM  OM  LDOM

    
algal net growth epiphyton net growth labile DOM decay
+ K RDOM  P OM  OM  RDOM + K LPOM  P OM  OM  LPOM + K RPOM  P OM  OM  RPOM

        
refractory DOM decay labile POM decay refractory POM decay
A
+  KCBOD RCBOD P −CBOD T −20 CBOD + Ks  P OM  OM s + SOD  OM sed
   V
CBOD decay 1st -order sediment release
0-order sediment release
( ISS +  F e  F e )PP

−  P +  ( Kmr - (1 − f psed ) Kmg )  Pm macro +  K zr  P z  zoo
ISS
 z
     
macrophyte net growth zooplankton respiratio n
inorganic solids adsorption
(4-109)
where:
z = model cell thickness, m
Ased = sediment surface area, m2
V = cell volume, m3
PP = adsorption coefficient, m3 g-1
fpsed = fraction of macrophyte phosphorus uptake from sediments
Pe = periphyton/epiphyton stoichiometric coefficient for phosphorus
Pa = algal stoichiometric coefficient for phosphorus
Pm = macrophyte stoichiometric coefficient for phosphorus
Pz = zooplankton stoichiometric coefficient for phosphorus
POM = organic matter stoichiometric coefficient for phosphorus
P-CBOD = phosphorus/CBOD stoichiometric ratio
 = temperature rate multiplier for CBOD decay
RBOD = conversion ratio for 5-day CBOD to CBOD ultimate
ISS = inorganic suspended solids settling velocity, m sec-1
Fe = particulate organic matter settling velocity, m sec-1
Keg = periphyton/epiphyton growth rate, sec-1
Ker = periphyton/epiphyton dark respiration rate, sec-1
Kmg = macrophyte growth rate, sec-1
Kmr = macrophyte respiration rate, sec-1
Kzr = macrophyte respiration rate, sec-1
KCBOD = CBOD decay rate, sec-1
Ksed = sediment decay rate, sec-1
SOD = anaerobic sediment release rate, g m-2s-1
P = phosphorus concentration, g m-3
Fe = total iron concentration, g m-3
Water Quality 136

KINETICS

CBOD = CBOD concentration, g m-3
sed = organic sediment concentration, g m-3
and the rate terms are evaluated in subroutine PHOSPHORUS. In the model, the PO4 concentration is in
units of PO4 as P.
The contribution of algae, POM, and DOM to phosphorus is given in the rate equations. However, effects
due to settling and contribution from sediments require some additional explanation.
Dissolved inorganic phosphorus adsorbs onto inorganic particulates under oxic conditions and is lost when
these materials settle. Loss may be rapid in the upper end of reservoirs in the riverine and transition zones
due to greater concentrations of allochthonous particulates. A Langmuir isotherm describes this process.
Since phosphorus concentrations are generally small, only the isotherm's linear region is utilized and is rep-
resented by the product PPΦP. The adsorbed solids settle at a rate equal to the solids' settling velocity.
Adsorption is not allowed to occur if dissolved oxygen concentrations approach anaerobic conditions.
Phosphorus adsorption onto inorganic suspended solids should be used cautiously. In most systems, avail-
able phosphorus sites for adsorption onto inflowing inorganic suspended solids are generally already in use,
so little adsorption takes place when inorganic suspended solids enter into a reservoir or estuary. The
phosphorus formulation needs to be recast with inorganic phosphorus as the state variable that is then
partitioned between dissolved and particulate forms. However, phosphorus sorption onto iron hydroxides
that form when anoxic waters come into contact with oxygenated water can be an important mechanism
of phosphorus removal from the water column for certain waterbodies.
Sediment contribution of phosphorus to overlying waters can be simulated in three ways:

(1) The sediment compartment accumulates particulate organic matter and algae, which then decay
under oxic conditions. This is modeled as a 1st-order process. However, sediment phosphorus
release depends upon sediment age, chemistry, overlying phosphorus concentrations, and other
factors not included in the sediment compartment.
(2) Under anaerobic conditions using the zero-order sediment oxygen model, sediments can be as-
signed a release rate for phosphorus that is a function of the assigned sediment oxygen demand
rate. This rate only changes based on temperature. Phosphorus release is only allowed to occur if
the overlying water dissolved oxygen concentration is approaching anaerobic conditions.
(3) Sediment diagenesis predicts P release under anaerobic conditions. This is a truly predictive model
based on organic matter loading to the sediments.
Note that when the sediment diagenesis model is not used, a combination of the first two can be used
where organic materials accumulate and decay in the sediments under aerobic conditions and are released
based on the SOD zero-order decay rate under anaerobic conditions.
137 Water Quality

AMMONIUM KINETICS
Ammonium
Ammonium is a product of the decay of organic matter, and algae can use ammonium during photosynthe-
sis to form proteins. In many estuarine applications, nitrogen is the limiting nutrient for algal growth. Mac-
rophytes can prefer taking N from the sediments versus the water column. Ammonia can decay in the pres-
ence of nitrifying bacteria to Nitrite and Nitrate.
Sediment
Nitrate-Nitrite
photosynthesis
RDOM
nitrification
Algae
decay LDOM
Macrophytes respiration Ammonium
LPOM
Epiphyton
anaerobic
release respiration
RPOM
photosynthesis
Sediment Zooplankton CBOD
Figure 83. Internal flux between ammonium and other compartments.
Referring to Figure 83, the rate equation for ammonium is:
S NH 4 =  K ar  N a s −  K ag  N a  a PNH 4 +  K er  N e  e −  K eg  N e  e PNH 4
      
algal respiration algal growth epiphyton respiration epiphyton growth
+ K LDOM  NOM  OM  LDOM + K RDOM  NOM  OM  RDOM + K LPOM  NOM  OM  LPOM

       
labile DOM decay refractory DOM decay labile POM decay
A
+ K RPOM  N OM  OM  RPOM + K s  N OM  OM s + S OD NH 4  OM sed
  
refractory POM decay 1st - order sediment release V
 
0 - order sediment release
+  K CBODRCBOD  N −CBOD  CBOD − K

T − 20
 NH 4  NH 4
  NH 4 
CBOD decay nitrification
+  (K mr − (1 − f nsed )K mg ) N m  macro +  K zr  N z  zoo

 
net macrophyte growth zooplankton respiration
(4-110)
Water Quality 138

KINETICS AMMONIUM
where:
Ased = sediment area, m2
V = volume of cell, m3
fnsed = fraction of macrophyte nitrogen uptake from sediments
Na = algal stoichiometric coefficient for nitrogen
Ne = periphyton/epiphyton stoichiometric coefficient for nitrogen
Nm = macrophyte stoichiometric coefficient for nitrogen
Nz = zooplankton stoichiometric coefficient for nitrogen
NOM = organic matter stoichiometric coefficient for nitrogen
N-CBOD = CBOD stoichiometric coefficient for nitrogen
NH4 = temperature rate multiplier for nitrification
NOx = temperature rate multiplier for denitrification
 = temperature rate multiplier for CBOD decay
RCBOD = ratio of 5-day CBOD to ultimate CBOD
KNOx = nitrate-nitrogen decay rate, sec-1
KNH4 = ammonium decay rate, sec-1
Kmr = macrophyte respiration rate, sec-1
Kzr = zooplankton respiration rate, sec-1
SODNH4 = sediment ammonium release rate, g m-2 sec-1
and the rate terms are evaluated in subroutine AMMONIUM. As with phosphorus, 0-order sediment re-
lease only occurs when dissolved oxygen is less than a minimum value as determined by the transition from
aerobic to anaerobic processes. Either a 0- or 1st-order process or a combination of both may be used for
sediment ammonium release. Also, the sediment diagenesis model computes ammonium release during
anaerobic conditions. In the model, the ammonium concentration is in units of NH4 as N.
139 Water Quality

NITRITE-NITRATE KINETICS
Nitrate-Nitrite
This compartment represents nitrate plus nitrite. Nitrite is an intermediate product in nitrification between
ammonium and nitrate. Nitrate can be used as a source of nitrogen for algae and periphyton/epiphyton
during photosynthesis based on the ammonia-nitrate preference factor for set for each algal and periphy-
ton group.
Nitrogen may be the limiting nutrient for algae in systems with high phosphorus loadings or in estuaries.
Some species of cyanobacteria can fix atmospheric nitrogen for use in photosynthesis. This process can be
included by setting the nitrogen half-saturation concentration for algal growth to zero.
Ammonium
nitrification
Algae
photosynthesis
Nitrate-Nitrite
Epiphyton
denitrification
Sediment Water Column
Figure 84. Internal flux between nitrate + nitrite and other compartments.
Referring to Figure 84, the rate equation for nitrate-nitrite is:
 NOx
S NOx = K NH 4  NH 4  NH 4 − K NO x  NO x  NO x −  NOx
  z

nitrificat ion water column denitrific ation
sediment denitrific ation
(4-111)
−  K ag N a  a (1 - PNH 4 ) −  K e g N e e (1 - PNH 4 )
     
algal uptake epiphyton uptake
where:
NOx = temperature rate multiplier for denitrification
Ne = epiphyton stoichiometric coefficient for nitrogen
KNH4 = nitrification rate, sec-1
KNOx = denitrification rate, sec-1
NOx = sediment transfer velocity, m sec-1
NH4= ammonia-nitrogen concentration, g m-3
NOx = nitrate-nitrogen concentration, g m-3
Appendix B Water Quality 140

KINETICS AMMONIUM
and the rate terms are evaluated in subroutine NITRATE. Nitrification is only allowed to occur if oxygen is
present, and denitrification is allowed based on the transition from aerobic to anaerobic processes (see
Dissolved Oxygen). In the model, the nitrate+nitrite concentration is in units of NO3 + NO2 as N.
Dissolved Silica
Dissolved silica is an important component of diatoms, providing the structural skeleton. In many cases
diatoms can be silica limited. Dissolved silica is taken up by algae based on stoichiometric relationships and
is produced by the decay of organic matter containing particulate biogenic silica. Also, dissolved silica is
adsorbed onto inorganic suspended solids based on a partitioning coefficient.
Algae Epiphyton
respiration photosynthesis respiration
Dissolved Particulate
Silica dissolution biogenic
silica
anaerobic decay adsorption/

release settling
Sediment System Loss
Figure 85. Internal flux between dissolved silica and other compartments.
Referring to Figure 85, the rate equation for dissolved silica is:
( )
S DSi = −  K ag − K ar  Sia  a −  K eg − K er  Sie e ( )
algal uptake epiphyton uptake
Ased
+ K sed  SiOM  OM  s ed + SOD Si  OM + K PSi OM  PSi
V (4-112)
1st order sediment release particulate biogenic
0-order sediment release decay
P Si (   ss ss +  POM  POML +  POM  POMR +  F e F e )

−  DSi
z
inorganic solids adsorption and settling
where:
 = temperature rate factor for BOD decay
Si = fraction of SOD for silica release
z = computational cell height, m
141 Water Quality

DISSOLVED OXYGEN KINETICS
Sie = epiphyton stoichiometric ratio for silica

Sia = algal stoichiometric ratio for silica
SiOM = sediment organic matter stoichiometric ratio for silica
Ased= sediment area, m2
V = computational cell volume, m3
PSi= silica adsorption coefficient, m3 g-1
SOD = sediment oxygen demand, g m-2 sec-1
ISS = inorganic suspended solids settling velocity, m sec-1
sed = organic sediment mass, g
DSi = dissolved silica concentration, g m-3
Organic Matter Variable Stoichiometry

The CE-QUAL-W2 model includes a feature starting with V3.5 which allows for the variable stoichiometry
of organic matter. Past versions of W2 models have used fixed stoichiometric constants for the ratios of
nitrogen and phosphorus to organic matter.
Eight constituents were required to simulate the amount of nitrogen and phosphorus in labile dissolved
organic matter (LDOM), refractory organic matter (RDOM), labile particulate organic matter (LPOM), and
refractory particulate organic matter (RPOM). These constituents are summarized in Table 13. Hence, all
inputs of organic matter accumulate N and P according to the stoichiometry of the incoming organic matter.
Table 13. Constituents used for variable stoichiometry of organic matter.
Abbreviation Constituent
LDOM-P Labile Dissolved Organic Matter – Phosphorus
RDOM-P Refractory Dissolved Organic Matter – Phosphorus
LPOM-P Labile Particulate Organic Matter – Phosphorus
RPOM-P Refractory Particulate Organic Matter – Phosphorus
LDOM-N Labile Dissolved Organic Matter – Nitrogen
RDOM-N Refractory Dissolved Organic Matter – Nitrogen
LPOM-N Labile Particulate Organic Matter – Nitrogen
RPOM-N Refractory Particulate Organic Matter – Nitrogen
Labile Dissolved Organic Matter – Phosphorus (LDOM-P)

LDOM-P is the amount of phosphorus in labile dissolved organic matter. The rate equation of LDOM-P is:
∑ 𝐾𝑎𝑒 𝛿Pa 𝛷𝑎 + ⏟
𝑆LDOM-P = ⏟ ∑(1 − 𝑃𝑎𝑚 )𝐾𝑎𝑚 𝛿Pa 𝛷𝑎 + ⏟
∑ 𝐾𝑒𝑒 𝛿Pe 𝛷𝑒
Water Quality 142

KINETICS DISSOLVED OXYGEN
+ ∑(1 − 𝑃𝑒𝑚 )𝐾𝑒𝑚 𝛿Pe 𝛷𝑒 + ∑(1 − 𝑃𝑚𝑚 )𝐾𝑚𝑚 𝛿Pm 𝛷𝑚 − ⏟

𝐾𝐿𝐷𝑂𝑀 𝛾OM 𝛷𝐿𝐷𝑂𝑀−𝑃 − ⏟
𝐾𝐿→𝑅 𝛾OM 𝛷𝐿𝐷𝑂𝑀−𝑃
⏟ ⏟
labile DOM-P decay labile to refractory
epiphyton mortality macrophyte mortality
DOM-P decay
where:
Pem = pattern coefficient for periphyton/epiphyton mortality
𝛿P-LDOM =LDOM stoichiometric ratio for phosphorus
Pe = periphyton/epiphyton stoichiometric coefficient for phosphorus
Kee = periphyton/epiphyton excretion rate, sec-1
Kem = periphyton/epiphyton mortality rate, sec-1
m = macrophyte concentration, g m-3
LDOM-P= labile DOM-P concentration, g m-3
Refractory Dissolved Organic Matter – Phosphorus (RDOM-P)

RDOM-P is the amount of phosphorus in refractory dissolved organic matter. The rate equation of RDOM-
P is:
S RDOM − P = K → R  LDOM − P −  OM K RDOM  RDOM − P
L   
DOM decay
where:
LDOM-P= labile DOM-P concentration, g m-3
RDOM-P= refractory DOM-P concentration, g m-3
KLROM= refractory POM decay rate, sec-1
KL→R= transfer rate from labile POM to refractory POM, sec-1
Labile Particulate Organic Matter – Phosphorus (LPOM-P)

LPOM-P is the amount of phosphorus in refractory dissolved organic matter. The rate equation of LPOM-P
is:
143 Water Quality

S LPOM − P =  Pam K am Pa  a +  Pem K em Pe e +  Pmm Pmpom K mm Pm  m +  K zm Pz  zoo
     
algal mortality epiphyton mortality macrophyte mortality zooplankton mortality
− K LPOM  OM  LPOM − P − K L→ R  LPOM − P +  K zg (1 − Z effic ) Pz  zoo

     
decay labile to refractory zooplankton excretion
POM decay
 LPOM − P   pom LPOM 

 zoo −  POM   LPOM − P
−  K zg
 LPOM    zoo  zoo +   alg  a +  pom LPOM 

z
   settling
zooplankton ingestion
where:
Pam= partition coefficient for algal mortality
Pem= partition coefficient for epiphyton mortality
Pmm= partition coefficient for macrophyte mortality
𝑃𝑚𝑝𝑜𝑚 =partition coefficient for RPOM and LPOM from macrophyte mortality
alg= Zooplankton preference fraction for algae
pom= Zooplankton preference fraction for particulate organic matter
POM= POM settling rate, m sec-1
Kam= algal mortality rate, sec-1
Kem= epiphyton mortality rate, sec-1
Kzm= zooplankton mortality rate, sec-1
Kmm= macrophyte mortality rate, sec-1
KLPOM= labile POM decay rate, sec-1
a= algal concentration, g m-3
zoo= zooplankton concentration, g m-3
e= epiphyton concentration, g m-3
LPOM= LPOM concentration, g m-3
LPOM-P= LPOM-P concentration, g m-3
Refractory Particulate Organic Matter – Phosphorus (RPOM-P)

RPOM-P is the amount of phosphorus in refractory dissolved organic matter. The rate equation of RPOM-
P is:
S RPOM = K → R  LPOM − P −  OM K RPOM  RPOM − P +  Pmm (1 − Pmpom )K mm Pm  m

L      
labile to refractory decay macrophyte mortality
POM decay
 RPOM − P
−  POM
 z
settling
where:
Pmm= partition coefficient for macrophyte mortality
Water Quality 144

OM= temperature rate multiplier
KRPOM= refractory POM decay rate, sec-1
Kmm= macrophyte mortality rate, sec-1
𝜔𝑃𝑂𝑀 = POM settling velocity, m/ sec-1
LPOM-P= labile POM phosphorus concentration, g m-3
RPOM-P= refractory POM phosphorus concentration, g m-3
Labile Dissolved Organic Matter – Nitrogen (LDOM-N)

LDOM-N is the amount of nitrogen in labile dissolved organic matter. The rate equation of LDOM-N is:
∑ 𝐾𝑎𝑒 𝛿Na 𝛷𝑎 + ⏟
𝑆LDOM-N = ⏟ ∑(1 − 𝑃𝑎𝑚 )𝐾𝑎𝑚 𝛿Na 𝛷𝑎 + ⏟
∑ 𝐾𝑒𝑒 𝛿Ne 𝛷𝑒
+ ∑(1 − 𝑃𝑒𝑚 )𝐾𝑒𝑚 𝛿Ne 𝛷𝑒 + ∑(1 − 𝑃𝑚𝑚 )𝐾𝑚𝑚 𝛿Nm 𝛷𝑚 − 𝐾

⏟𝐿𝐷𝑂𝑀 𝛾OM 𝛷𝐿𝐷𝑂𝑀−𝑁
⏟ ⏟
labile DOM-N decay
epiphyton mortality macrophyte mortality
−⏟
𝐾𝐿→𝑅 𝛾OM 𝛷𝐿𝐷𝑂𝑀−𝑁
labile to refractory
DOM-N decay
where:
Pem = pattern coefficient for epiphyton mortality
𝛿P-LDOM =LDOM stoichiometric ratio for nitrogen
m = macrophyte concentration, g m-3
LDOM-N= labile DOM-N concentration, g m-3
Refractory Dissolved Organic Matter – Nitrogen (RDOM-N)

RDOM-N is the amount of nitrogen in refractory dissolved organic matter. The rate equation of RDOM-N
is:
S RDOM − N = K → R  LDOM − N −  OM K RDOM  RDOM − N

L   
DOM decay
145 Water Quality

where:
LDOM-N= labile DOM-N concentration, g m-3
RDOM-N= refractory DOM-N concentration, g m-3
Labile Particulate Organic Matter – Nitrogen (LPOM-N)

LPOM-N is the amount of nitrogen in refractory dissolved organic matter. The rate equation of LPOM-N is:
S LPOM − N =  Pam K am Na  a +  Pem K em Ne e +  Pmm Pmpom K mm Nm  m + K zm Nz  zoo
     
algal mortality epiphyton mortality macrophyte mortality zooplankton mortality
− K LPOM  OM  LPOM − N − K L→ R  LPOM − N + K zg (1 − Z effic ) Nz  zoo

    
decay labile to refractory zooplankton
POM decay
 LPOM − N   pom LPOM 

 zoo −  POM   LPOM − N
− K zg
 LPOM    alg  a +   zoo  zoo +  pom LPOM 

z
     settling
zooplankton ingestion
where:
Pem = partition coefficient for periphyton/epiphyton mortality
𝜎𝑎𝑙𝑔 = Zooplankton preference fraction for algae
𝜎𝑝𝑜𝑚 = Zooplankton preference fraction for particulate organic matter
POM = POM settling rate, m sec-1
Kzm = zooplankton mortality rate, sec-1
KL R = transfer rate from labile POM to refractory POM, sec-1
zoo = algal concentration, g m-3
LPOM= LPOM concentration, g m-3
LPOM-N= LPOM-N concentration, g m-3
Water Quality 146

Refractory Particulate Organic Matter – Nitrogen (RPOM-N)

RPOM-N is the amount of nitrogen in refractory dissolved organic matter. The rate equation of RPOM-N
is:
S RPOM = K → R  LPOM − N −  OM K RPOM  RPOM − N +  Pmm (1 − Pmpom )K mm Nm  m

L      
labile to refractory decay macrophyte mortality
POM decay
 RPOM − N
−  POM
 z
settling
where:
𝑃𝑚𝑝𝑜𝑚 = partition coefficient for RPOM and LPOM from macrophyte mortality
 OM = temperature rate multiplier
K L→R = transfer rate from labile POM to refractory POM, sec-1
K RPOM = refractory POM decay rate, sec-1
𝜔𝑅𝑃𝑂𝑀 = POM settling velocity, m/ sec-1
LPOM-N= labile POM-N concentration, g m-3
RPOM-N= refractory POM-N concentration, g m-3
Particulate Biogenic Silica

Particulate biogenic silica results from diatom mortality and settles and also dissolves to form dissolved
silica.
Epiphyton Particulate
mortality decay
Biogenic Dissolved Silica
Algae Silica
Silica
settling
System Loss
Figure 86. Internal flux between particulate biogenic silica and other compartments
Referring to Figure 86, the rate equation for particulate biogenic silica is:
S PSi = P am Sia K am a + P e m Sie K e m e − K PSi OM  PSi −

 PSi  PSi
(4-113)
algae mortality epiphyton mortality
z
decay
settling
147 Water Quality

where:
Sie = epiphyton stoichiometric coefficient for silica
Sia = algal stoichiometric coefficient for silica
Pem = partition coefficient for periphyton/epiphyton mortality
KPSi = particulate biogenic silica decay rate, sec-1
PSi = particulate biogenic silica settling rate, m sec-1
PSi = particulate biogenic silica concentration, g m-3
Total Iron
Total iron is should no longer be used in the model. It is only still a state variable for backwards compatibility
with earlier model versions. Since Version 4.0, new state variables for reduced and oxidized forms of Fe and
Mn should be used. This state variable was used in the model primarily because of its effect on nutrient
concentrations through adsorption and settling. Iron is commonly released from anoxic sediments and may
influence nutrient dynamics in many reservoirs. Iron may also contribute to dissolved oxygen depletions
which are accounted for in the new model state variables. This older state variable though does not include
these effects. Iron sediment release is modeled as a zero-order process.
Total Iron
anaerobic release adsorption/settling
Sediment
Figure 87. Internal flux between total iron and other compartments.
Referring to Figure 87, the rate equation for total iron is:

S Fe = SOD  om As −  F e  F e
 z (4-114)
0-order
sediment release oxic water column
settling
where:
As = sediment area, m2
Fe = settling velocity, m sec-1
Water Quality 148

SOD = sediment oxygen demand, g m-2 sec-1.
Dissolved Oxygen
Oxygen is one of the most important elements in aquatic ecosystems. It is essential for higher forms of life,
controls many chemical reactions through oxidation, and is a surrogate variable indicating the general
health of aquatic systems.
CE-QUAL-W2 includes both aerobic and anaerobic processes. The ability to model anaerobic periods is
important since it provides information on potential problems with water quality. Simulations can be used
to identify possibilities for both metalimnetic and hypolimnetic oxygen depletion and its impact on various
water control management alternatives. If a single variable were to be measured in aquatic systems that
would provide maximum information about the system state, it would be dissolved oxygen.
Figure 88. Internal flux between dissolved oxygen and other compartments.
Referring to Figure 88, the rate equation for dissolved oxygen is:
149 Water Quality

S DO =  (K ag − K ar ) OMa a +  (K eg − K er ) OMe e + A sur K L ( s DO -  DO )

  Vsur
algal net production epiphyton net production 
aeration
− K RPOM  OM  OM  RPOM − K LPOM  OM  OM  LPOM − K LDOM  OM  OM  LDOM

  
refractory POM decay labile POM decay labile DOM decay
A
− K RDOM  OM  OM  RDOM − Ks  OM  OM s ed − SOD  OM sed
  V
refractory DOM decay 1st -order sediment decay  (4-115)
0-order SOD
−  K CBOD RCBODT −20 CBOD − K NH 4  NH 4  NH 4  NH 4 +  (K mg − K mr ) OMmac  macro

  
CBOD decay nitrification macrophyte net production
−   zoo K zr OMzoo zoo −  H 2 S −O 2 K H 2 S  H 2 S −  CH 4−O 2 K CH 4 CH 4 −  Fe 2−O 2 K Fe 2 Fe 2

   
zooplankton respiration oxidation
−  Mn 2−O 2 K Mn 2 Mn 2
  
oxidation
where:
OMa = oxygen stoichiometric coefficient for algal organic matter
OMe = oxygen stoichiometric coefficient for periphyton/epiphyton organic matter
OMmac = oxygen stoichiometric coefficient for macrophyte organic matter
OM = oxygen stoichiometric coefficient for organic matter
NH4 = oxygen stoichiometric coefficient for nitrification
Fe2-O2 = oxygen stoichiometric coefficient for Fe+2
Mn2-O2= oxygen stoichiometric coefficient for Mn+2
H2S-O2 = oxygen stoichiometric coefficient for H2S
CH4-O2 = oxygen stoichiometric coefficient for CH4
OMzoo = oxygen stoichiometric coefficient for zooplankton
zoo = temperature rate multiplier for zooplankton
RBOD = conversion from CBOD in the model to CBOD ultimate
V = volume of computational cell, m3
Asur = water surface area, m2
Kmr = macrophyte dark respiration rate, sec-1
Kzr = zooplankton respiration rate, sec-1
KNH4 = ammonia decay (nitrification) rate, sec-1
KFe2 = Fe+2 oxidation rate, sec-1
KMn2 = Mn+2 oxidation rate, sec-1
KCH4 = CH4 oxidation rate, sec-1
Water Quality 150

KH2S = H2S oxidation rate, sec-1

KL = interfacial exchange rate for oxygen, m sec-1
NH4 = ammonia-nitrogen concentration, g m-3
macr = macrophyte concentration, g m-3
BOD = CBOD concentration, g m-3
DO = dissolved oxygen concentration, g m-3
Fe2 = reduced Fe concentration, g m-3
Mn2 = reduced Mn concentration, g m-3
CH4 = CH4 concentration, g m-3
H2S = H2S concentration, g m-3
sDO = saturation DO concentration, g m-3
and the rate terms are evaluated in subroutine DISSOLVED_OXYGEN. Decay is not allowed to occur when
dissolved oxygen concentrations are zero. A Monod formulation is used to move gradually from oxic to
anoxic conditions. This is accomplished by reducing temperature rate multipliers eventually to zero as dis-
solved oxygen concentrations are zero.
This reduction of oxic reactions as dissolved oxygen levels approach zero is based on specification of a dis-
solved oxygen half-saturation constant in the following equation:
𝛷𝐷𝑂
Rate Reduction = where DO is the concentration of dissolved oxygen and KDO is a half-saturation
𝐾𝐷𝑂 +𝛷𝐷𝑂
dissolved oxygen concentration when oxic reactions are half of their maximum without limitation of oxygen
conditions.
Since the river basin model will encompass waterbodies that are dependent on boundary shear in river
segments and wind stress for lake or reservoir segments for turbulence, the reaeration formulae for these
systems must be different. In the following sections, formulae for reaeration as a function of wind speed
and boundary shear are presented. The user can select a different formulation for each waterbody type.
The reason for selecting a waterbody type is to help the user select a system that best approximates the
appropriate theory. The possible water body types are RIVER, LAKE, or ESTUARY.
151 Water Quality

River Reaeration Equations

Reaeration equations for rivers are given in Table 14. Most of these equations in the table are based on
field studies of selected streams or laboratory channels. Equations 7 and 8 were developed from Melting
and Flores (1999) for a large data set of reaeration coefficients. These may be the best choice for rivers
even though other equations have been used extensively.
Recently, Moog and Jirka (1998) suggested that formulations that do not account for channel slope should
not be used. Therefore, equations 7 and 8 may again be the best selection of equations for river sections.
Thomann and Mueller (1987) suggested using Equation 1 except for small streams where Equation 3 for
flow less than 10 cfs should be used. They also suggested a minimum value of KL of 0.6 m day-1. This value
was has been implemented as a lower limit in the code.
Mills et al. (1985) in a review of water quality reaeration coefficients used a different formulation than
equation 3 in Table 14 based on a 1978 reference at 25ºC:
𝐾𝑎 = 7776𝑈𝑆 𝑓𝑜𝑟 𝑄 < 10𝑐𝑓𝑠

𝐾𝑎 = 4665.6𝑈𝑆 𝑓𝑜𝑟 10 < 𝑄 < 3000𝑐𝑓𝑠
𝐾𝑎 = 2592𝑈𝑆 𝑓𝑜𝑟 𝑄 > 3000𝑐𝑓𝑠 (4-116)
where S is the slope in ft/ft and U is velocity in ft/s
However, Mills et al. (1985) only recommend its use for shallow low-flow streams. Therefore, equation 3
in Table 14 is exactly the same as the above formulation in the low flow regime (note that the slopes in
Equation 3 are in ft/mile and in the above equation are unitless).
Covar (1976) used an approach where the equations of O'Connor-Dobbins, Churchill, and Owens were used
together based on the applicability of each equation. The applicability of each equation was based on the
velocity of the stream and its depth. This is equation 0 in the following table.
Table 14. River reaeration equations.

# Equation Comments Applicability Reference
Ka – evaluated based on
0 Either Eq 1, 2 or 4 applicability criteria of Covar (1976)
these equations
DO2 = H2O molecular dif-
fusion, m2 s-1
U = average velocity, m depths between 1-
1/2
𝐾𝐿 (𝐷𝑂2 𝑈) s-1 O’Connor and
1 𝐾𝑎 = = 30 ft and velocities
𝐻 𝐻 3/2 H = average channel Dobbins (1958)
between 0.5-1.6 fps
depth, m
U, ft s-1 depths between 2- Churchill, Elmore

𝐾𝐿 11.6𝑈
2 𝐾𝑎 = = 1.67 H, ft 11 ft and velocities and Buckingham
𝐻 𝐻 Ka, day-1 between 1.8-5 fps (1962)
S, ft mile-1
𝐾𝑎 = 0.88𝑈𝑆 𝑓𝑜𝑟 10 < 𝑄 < 300𝑐𝑓𝑠 suggested for use Tsivoglou and
3 U, ft s-1
𝐾𝑎 = 1.8𝑈𝑆 𝑓𝑜𝑟 1 < 𝑄 < 10𝑐𝑓𝑠 when Q < 10cfs Wallace (1972)
Ka, day-1
depths between
𝐾𝐿 21.6𝑈0.67 U, ft s-1 0.4-2.4 ft and veloc- Owens et al.
4 𝐾𝑎 = =
𝐻 𝐻1.85 H, ft ities between 0.1- (1964)
1.8 fps
Water Quality 152

# Equation Comments Applicability Reference

u* = shear velocity,
(HSg)0.5
𝐾𝐿 25𝑢∗ S = slope of energy Thackston and
5 𝐾𝑎 = = (1 + 𝐹 0.5 )
𝐻 𝐻 grade line Krenkel (1966)
F = Froude number,
U/(gH)0.5
𝐾𝐿 7.62𝑈 U, ft s-1 Langbien and Du-
6 𝐾𝑎 = = 1.33
𝐻 𝐻 H, ft rum (1967)
U, m s-1
𝐾𝑎 = 517(𝑈𝑆)0.524 𝑄−0.242 𝑓𝑜𝑟 ⥄ 𝑄 < 0.556 S, m m-1 for pool and riffle Melching and Flo-
7
𝐾𝑎 = 596(𝑈𝑆)0.528 𝑄−0.136 𝑓𝑜𝑟 ⥄ 𝑄 > 0.556 Q, m3 s-1 streams res (1999)
Ka, day-1
U, m s-1
S, m m-1
𝐾𝑎 = 88(𝑈𝑆)0.313 𝐷−0.353 𝑓𝑜𝑟𝑄 < 0.556
W = stream top width, for channel-control Melching and Flo-
8 𝐾𝑎 = 142(𝑈𝑆)0.333 𝐷−0.66 𝑊 −0.243
m streams res (1999)
𝑓𝑜𝑟𝑄 > 0.556
D = average depth, m
Ka, day-1
U, m s-1
H, m
w/ channel slope - 𝐾𝑎 = 𝐶1 𝑈𝐶2 𝐻 𝐶3 𝑆 𝐶4 S, non-dimensional User defined rela-
9
w/o channel slope - 𝐾𝑎 = 𝐶1 𝑈𝐶2 𝐻 𝐶3 Ka, day-1 tionship
C1, C2, C3, C4 = user de-
fined parameters
u* = shear velocity,
(HSg)0.5
S = slope of energy
1 𝐾𝐿 5.0𝑢∗ Thackston and
𝐾𝑎 = = (1 + 9𝐹 0.25 ) grade line
0 𝐻 𝐻 Dawson (2001)
F = Froude number,
U/(gH)0.5
Ka, day-1
Figure 89 shows the functional dependence of these formulae assuming the following relationship between
flow (Q, ft3/s), velocity (V, ft/s) and depth (H, ft) (St. John et al., 1984):
V = 0.033 Q0.5
H = 0.475 Q0.4
S = 5.2 ft/mile
153 Water Quality

Figure 89. Reaeration coefficient as a function of flow rate.
Lake Reaeration Equations

Wind effects rather than boundary shear more often control reaeration in lakes, reservoirs, and estuarine
systems. There have been many wind studies for lakes (e.g., O’Connor, 1983) and open ocean systems. A
summary of wind speed formulae for predicting reaeration is shown in Table 15 based on a 10 m wind
measuring height.
Table 15. Lake reaeration equations as a function of wind speed at 20oC.

# Equation Comments Reference
𝐾𝐿 0.864𝑊 W, m s-1 at 10 m
𝐾𝑎 = =
1 𝐻 𝐻 H, m Broecker et al (1978)
KL, m day-1
𝐾𝐿 𝛼𝑊 𝛽  = 0.2,  = 1 for W < 3.5 m s-1
𝐾𝑎 = =
2 𝐻 𝐻 a = 0.057,  = 2 for W > 3.5 m s-1 where W is Gelda et al (1996)
a daily average wind speed
𝐾𝐿
𝐾𝑎 = W, m s-1 at 10 m
3 𝐻 Banks and Herrera (1977)
0.728𝑊 0.5 − 0.317𝑊 + 0.0372𝑊 2 KL, m day-1
=
𝐻
𝐾 0.0986𝑊 1.64 The latter equation was the original equa-
𝐾𝑎 = 𝐿 = [at 20oC] or
𝐻 𝐻 tion used where W is measured at 10 m and
0.5
𝐾𝐿 0.0986𝑊 1.64 600
𝐾𝑎 = = ( ) Sc is the Schmidt number
𝐻 𝐻 𝑆𝑐 Sc=(/D)= 13750[0.10656 exp(-
4 Wanninkhof et al. (1991)
.0627T)+.00495]
T = temperature, oC
=kinematic viscosity
D=diffusivity
Water Quality 154


𝐷𝑜2 DO2 = oxygen molecular diffusivity, m2 s-1
5 𝐾𝐿 (200 − 60𝑊 0.5 )10−6 W, m s-1 Kanwisher (1963)
𝐾𝑎 = =
𝐻 𝐻 KL, m s-1
𝐾𝐿 0.5 + 0.05𝑊 2
6 𝐾𝑎 = = Cole and Buchak (1995)
𝐻 𝐻
𝐾𝐿 0.362√𝑊
𝐾𝑎 = = 𝑊 < 5.5𝑚/𝑠
7 𝐻 𝐻 Banks (1975)
𝐾𝐿 0.0277𝑊 2
𝐾𝑎 = = 𝑊 > 5.5𝑚/𝑠
𝐻 𝐻
𝐾𝐿 0.64 + 0.128𝑊 2 Recommended form for WQRSS reservoir
8 𝐾𝑎 = = Smith (1978)
𝐻 𝐻 model
𝐾𝐿 0.156𝑊 0.63
𝐾𝑎 = = 𝑊 ≤ 4.1𝑚/𝑠
9 𝐻 𝐻 Liss (1973)
𝐾𝐿 0.0269𝑊 1.9
𝐾𝑎 = = 𝑊 > 4.1𝑚/𝑠
𝐻 𝐻
𝐾𝐿 0.0276𝑊 2 Downing and Truesdale
10 𝐾𝑎 = =
𝐻 𝐻 (1955)
𝐾𝐿 0.0432𝑊 2
11 𝐾𝑎 = = Kanwisher (1963)
𝐻 𝐻
𝐾𝐿 0.319𝑊
12 𝐾𝑎 = = Yu et al (1977)
𝐻 𝐻
𝐾𝐿 0.398
𝐾𝑎 = = 𝑊 < 1.6
13 𝐻 𝐻 W = wind speed, m s-1 Weiler (1974)
𝐾𝐿 0.155𝑊 2
𝐾𝑎 = = 𝑊 ≥ 1.6
𝐻 𝐻
𝐾𝐿 𝐶1 + 𝐶2 𝑊 𝐶3 User defined relationship where:
𝐾𝑎 = = W, m s-1 at 10 m
14 𝐻 𝐻
Ka, day-1
C1, C2, C3 are user defined
Figure 90 shows how these formulations vary with wind speed.
The definition of wind speed was usually taken at an elevation of 10 m for these formulations. The wind
speed at 10 m elevation in the middle of a lake or reservoir, W10m, can be computed from that measured at
10 m on land by using an approach from Fang and Stefan (1994).
𝑊10𝑚 = 𝑊𝑧 𝑓(𝑓𝑒𝑡𝑐ℎ) (4-117)
where:
Wz = wind speed measured at 10 m height on land, m/s
10 𝛿
𝑙𝑛 𝑙𝑛 5𝑍𝐵+4.6052
𝑧𝑜2 𝑧𝑜1
𝑓(𝑓𝑒𝑡𝑐ℎ) = 10 𝛿 ≅ 3𝑍𝐵+9.2103 (4-118)
𝑙𝑛 𝑙𝑛
𝑧𝑜1 𝑧𝑜2
where:
zo1 = roughness of land (assume 0.01 m) (Kraus, 1972)
zo2 = roughness of water surface (assume 0.0001 m) (Ford and Stefan, 1980)
 = thickness of wind boundary layer over smooth surface that is a function of the fetch length (El-
liot, 1958), m
𝑓𝑒𝑡𝑐ℎ
ZB = 0.8 𝑙𝑛 2
− 1.0718
155 Water Quality

The function f(fetch) varies from 1.056 for small lakes to 1.123 for large lakes. The fetch is the length in m
of the wind over the water surface from one bank to the other.
Figure 90. Variation of wind speed and KL for lake/reservoir equations.
Equations for correcting the wind speed to 10 m and accounting for fetch dependence are included in the
model. This dependence on measuring height, fetch, and wind speed is shown in Figure 91 and Figure 92.
Water Quality 156

8.5
Wind corrected to 10 m, m s-1

7.5
6.5
5.5
5
0 5 10 15 20
Wind measuring height, m
Figure 91. Wind speed of 5 m s-1 and a fetch of 5 km corrected to 10 m as a function of measuring height
on land.
6.2
Wind corrected to 10 m, m s-1
5.8
5.6
5.4
5.2
5
0 5000 10000 15000 20000 25000
Fetch, m
Figure 92. Wind speed of 5 m s-1 corrected to 10 m as a function of fetch.
157 Water Quality

Wind corrected to 10 m, m s-1 25
20
15
10
0
0 5 10 15 20 25
Wind speed measured, m s-1
Figure 93. Wind corrected to 10 m based on wind measured on land.
Estuarine Equations
For estuarine systems, Thomann and Mueller (1987) and Chapra (1997) suggest using any of the wind for-
mulations in Table 16 or Equation 1 in Table 14 using the mean tidal velocity over a tidal cycle for the
horizontal velocity. Table 16 shows an additional formulation from Thomann and Fitzpatrick (1982) for es-
tuaries, as well as the approach of Covar (1976) for rivers. Since many texts suggest using the mean tidal
velocity, caution should be exercised in using these equations since they are based on the instantaneous
velocity in the model.
Table 16. Reaeration equations for estuarine waterbody at 20oC.

Ka is determined based on ap-
0 Either Eq 1, 2 or 4 from Table 14 plicability criteria of each of Covar (1976)
these 3 formulations
U, m s-1
0.5 2 this formula combines the ef-
𝐾𝐿 0.728𝑊 − 0.317𝑊 + 0.0372𝑊 √𝑈
𝐾𝑎 = = + 3.93 1.5 Thomann and Fitz-
1 𝐻 𝐻 𝐻 fect of wind from Banks and
patrick (1982)
Herrera (1977) and estuary
tidal flow
U, m s-1
H, m
0.5 + 𝐶4 𝑊 2 W, m s-1 at 10 m User defined rela-
2 𝐾𝑎 = 𝐶1 𝑈𝐶2 𝐻 𝐶3 +
𝐻 Ka, day-1 tionship
C1, C2, C3, and C4 - user defined
parameters
Chu and Jirka (2003) reported on wind and stream flow induced reaeration and showed that for cases where
the time scale tidal flow is >> the time scale for reaeration, then the reaeration process can be described as
a steady process and the impacts of stream flow and wind can be additive using:
Water Quality 158

𝐾𝐿 = 𝐾𝐿𝑏 + 𝐾𝐿𝑤
Where KLb is the reaeration coefficient (or transfer velocity) due to bottom shear and K Lw is the reaeration
coefficient due to wind in m/day.
3 0.25
𝑢∗𝑏
𝐾𝐿𝑏 = 2.150 ( )
ℎ
𝑓
Where h is the depth in cm, 𝑢∗𝑏 = √ 𝑈, U is the mean water velocity, f is approximately 0.04.
8
𝐾𝐿𝑤 = 𝛼𝑢∗𝑎 for smooth surfaces (u*a<20 cm/s)
𝛼 = 4.38𝐸 − 5when ua* is in units of cm/s
2
𝐾𝐿𝑤 = 𝛽𝑢∗𝑎 for rough surfaces
𝛽 = 1.83𝐸 − 3 when ua* is in units of cm/s
𝑢∗𝑎 = 0.01𝑈10 (8 + 0.65𝑈10 )1/2 from Wu (1980) and U10 is wind speed at 10 m in m/s and u*a is in m/s.
Reaeration Temperature Dependence

Reaeration temperature dependence is usually based upon an Arrhenius formulation:
𝐾𝑇 = 𝐾20 𝛩𝑇−20 (4-119)
where  is 1.024. This dependency is based on the variation in molecular diffusivity as a function of tem-
perature. The molecular diffusivity of oxygen varies according to temperature based on the following equa-
tion:
𝐷𝑂2 = 4.58𝐸 − 11𝑇 + 1.2𝐸 − 9 (4-120)
where:
DO2 = molecular diffusivity of oxygen, m2 sec-1
Using Equation 1 in Table B-2, the variation of KL as a function of temperature using Equation 43 (assuming
D is a constant) and 44 yield similar results (Figure 94).
No temperature correction was made to the calculated value of K L in earlier versions of the model. The
latest version includes the correction with theta set to 1.024.
159 Water Quality

Figure 94. Variation of KLT/KL20 as a function of temperature.
Dam Reaeration
Many rivers and reservoirs have spillways or weirs over which water entrains oxygen into a pool below the
dam. This section outlines the approach for including oxygen and nitrogen entrainment at dams and weirs.
Small Dams or Weirs
Many rivers have small spillways or weirs over which water flows that entrains oxygen. Table 17 lists for-
mulae for predicting the entrainment of dissolved oxygen based on empirical data.
Table 17. Formulae for small dam or weir reaeration effects.

Equation Comments Reference
Da = DO deficit above dam, g m3
Db = DO deficit below dam, g m3
1 T = temperature, oC
𝐷𝑎 − 𝐷𝑏 = [1 − ] 𝐷𝑎 or
1+0.11𝑎𝑏(1+0.046𝑇)𝐻 H = height of water fall, ft Barrett et al.
𝐷𝑎 a = 1.25 in clear to slightly polluted water to (1960)
= 1 + 0.11𝑎𝑏(1 + 0.046𝑇)𝐻
𝐷𝑏 1.00 in polluted water
b = 1.00 for weir with free flow
b = 1.3 for step weirs or cascades
Da = DO deficit above dam, g m-3
Db = DO deficit below dam, g m-3
H = height of water fall, ft Mastropietro
𝐷𝑎 − 𝐷𝑏 = 0.037𝐻𝐷𝑎
(1968)
only valid for dams less than 15 ft and T between
20 and 25oC
𝐷𝑎 Db = DO deficit below dam, g m-3 Butts and Ev-
= 1 + 0.38𝑎𝑏(1 − 0.11𝐻)(1 + 0.046𝑇)𝐻
𝐷𝑏 T = temperature, oC ans (1983)
H = height of water fall, m
Water Quality 160

Equation Comments Reference

a = 1.8 for clean water to 0.65 for gross pol-
luted water
b = 0.05 for sluice gates
b = 1.0 for sharp crested straight faced weir
b = 0.45 for flat broad crested curved face
weir
b = 0.7 for flat broad crested weir with regular
step
b = 0.8 for sharp crested vertical face weir
b = 0.6 for flat broad crested weir vertical face
Most of these equations have been used for dams or weirs with heights of fall between 3 and 10 m. These
equations do not generally predict supersaturation.
Large Dam Spillways/Gates
The USACE has been involved in gas abatement studies on the Columbia and Snake River system for many
years (WES, 1996, 1997). Some of their research efforts have been focused on development of models of
gas generation from spillways. These empirical models have been called CriSP 1.6 (Columbia Basin Research,
2000). The gas production equations used in CriSP are empirical correlations between total dissolved gas
(TDG), usually measured a mile downstream of the dam after turbulence from the spillway had subsided,
and discharge, usually measured in kcfs. The form of these equations is shown in Table 18.
Table 18. Equations used in CRiSP model for gas production.

Equation type Equation Description of empirical coefficients
Qs = total spill, kcfs
Linear function of total spill %𝑇𝐷𝐺 = 𝑚𝑄𝑠 + 𝑏 m = empirical coefficient
b = empirical coefficient
Qs = total spill, kcfs
𝑐𝑄𝑠 a = empirical coefficient
Bounded exponential of total spill %𝑇𝐷𝐺 = 𝑎 + 𝑏𝑒
b = empirical coefficient
c = empirical coefficient
qs = spill through an individual spillway, kcfs

Bounded exponential of the spill on 𝑐𝑞𝑠 a = empirical coefficient
%𝑇𝐷𝐺 = 𝑎 + 𝑏𝑒
a per spillway basis b = empirical coefficient
Examples of some of these correlations are shown in Table 19. In many cases, the %TDG in these correla-
tions was constrained to a maximum of 145% and when the flow reached only a few kcfs, there was as-
sumed to be no change in TDG from the forebay to the tailrace. Also, the correlations in Table 19 sometimes
changed from year to year based on changes in operating conditions or structural changes in the spillway
or deflectors.
161 Water Quality

Table 19. Equations used in CRiSP model for gas production for Columbia and Snake Dams.
Dam Equation Coefficients
Bonneville %𝑇𝐷𝐺 = 𝑚𝑄𝑠 + 𝑏 m = 0.12, b = 105.61
Lower Granite %𝑇𝐷𝐺 = 𝑎 + 𝑏𝑒 𝑐𝑞𝑠 a = 138.0; b = -35.8; c = -0.10
Dworshak %𝑇𝐷𝐺 = 𝑎 + 𝑏𝑒 𝑐𝑄𝑠 a = 135.9; b = -71.1; c = -0.4787
a = 136.8; b = -42.0; c = -0.0340 1995

a = 138.7; b = -79.0; c = -0.0591 1996
Ice Harbor %𝑇𝐷𝐺 = 𝑎 + 𝑏𝑒 𝑐𝑄𝑠
a = 130.9; b = -26.5; c = -0.0220 1997
a = 120.9; b = -20.5; c = -0.0230 1998
a = 138; b = -36; c = -0.02

Hell's Canyon %𝑇𝐷𝐺 = 𝑎 + 𝑏𝑒 𝑐𝑄𝑠
[Assumed relationship - no data]
DO Impacts of Spillways
For each spillway, weir, or gate, the user now has the choice of equation to use for computing the effects
of hydraulic structures on downstream dissolved oxygen. The equations chosen are shown in Table 20.
These equations are based on equations from Table 17 and Table 18.
Table 20. Spillways and weirs reaeration

# Equation type Equation Description of empirical coefficients
%TDG = % total dissolved gas saturation
%𝑇𝐷𝐺 = 𝑎𝑞𝑠 + 𝑏
Linear function of spill qs = spill through an individual spillway,
DO concentration, O2, is determined from
1 on a per spillway basis kcfs
𝛷𝑂2 = %𝑇𝐷𝐺𝛷𝑠
a = empirical coefficient
S = dissolved oxygen saturation
qs = spill through an individual spill-
%𝑇𝐷𝐺 = 𝑎 + 𝑏𝑒 𝑞𝑠
way,kcfs
empirical coefficients a = empirical coefficient
2 O2, is then determined from
a and b b = empirical coefficient
𝛷𝑂2 = %𝑇𝐷𝐺𝛷𝑠𝑎𝑡
sat = dissolved oxygen saturation
Db = DO deficit below dam
T = temperature in oC
H = height of water fall, m
a = 1.8 for clean water to 0.65 for gross
polluted water
Reaeration effect for a 𝐷𝑎 b = 0.05 for sluice gates
= 1 + 0.38𝑎𝑏(1 − 0.11𝑐)(1 + 0.046𝑇)𝑐 1.0 for sharp crested, straight faced
small height weir or 𝐷𝑏
dam (<10 m) weir
3
O2 below the dam is then computed from:
0.45 for flat, broad crested, curved face
empirical coefficients
weir
a, b, and c 𝛷𝑂2 = 𝛷𝑠𝑎𝑡 − 𝐷𝑏
0.7 for flat, broad crested weir with
regular step
0.8 for sharp crested, vertical face weir
0.6 for flat, broad crested weir with
vertical face
sat = dissolved oxygen saturation, g m-3
Note that for equations 1 and 2, the maximum TDG allowed is 145%. If TDG is computed to be less than
100%, there is no effect of the spillway or gate on reaeration. For each spillway or gate defined in the
Water Quality 162

model, there is a section to define whether gas effects for nitrogen and dissolved oxygen are computed
and, if so, by which formula.
For each spillway or weir, the user turns on the computations and then selects an equation number from
Table 20. This algorithm only computes gas effects for flow from upstream to downstream. There is no
adjustment of nitrogen or dissolved oxygen for reverse flow.
SYSTDG Algorithm
An algorithm for computing TDG from spillways was developed by Scheider and Hamilton (2015a, 2015b)
to compute real-time TDG saturation levels in the Columbia River basin, USA. SYSTDG is an empirical re-
gression equation for each spillway at a hydropower facility based on flow rate, tailwater depth, and baro-
metric pressure. This algorithm also incorporates entrainment of water from powerhouse flows and specific
TDG relations for each bay. The original development was an Excel-based spreadsheet model. The Corps of
Engineers (Zhong and Threadgill, 2019) integrated this into CE-QUAL-W2 for the Columbia and Snake River
hydropower facilities.
This review of the SYSTDG model is taken from Zhong and Threadgill (2019) who reviewed and summarized
the SYSTDG equations and approach. Table 21 summarizes the SYSTDG TDG regression equations.
Table 21. List of spillway flow TDG production equations in SYSTDG from Scheider and Hamilton
(2015b).
No TDG Production Equation*
1 𝑇𝐷𝐺𝑠𝑝 = 𝑃1 ∗ (1 − 𝑒 𝑃3∗ 𝑄𝑠𝑝 ) + 𝑏𝑝
∆𝑇𝐷𝐺𝑠𝑝 = 𝑃1 ∗ (1 − 𝑒 𝑃3∗ 𝑄𝑠𝑝 )

2 𝑇𝐷𝐺𝑠𝑝 = 𝑃1 ∗ (𝑡𝑤𝑒 − 𝑡𝑤𝑐𝑒)𝑃2 ∗ (1 − 𝑒 𝑃3∗ 𝑞𝑠 ) + 𝑃4 + 𝑏𝑝
∆𝑇𝐺𝑃𝑠𝑝 = 𝑃1 ∗ (𝑡𝑤𝑒 − 𝑡𝑤𝑐𝑒)𝑃2 ∗ (1 − 𝑒 𝑃3∗ 𝑞𝑠 ) + 𝑃4

3 𝑇𝐷𝐺𝑠𝑝 = 𝑃1 ∗ (𝑡𝑤𝑒 − 𝑡𝑤𝑐𝑒)𝑃2 ∗ 𝑞𝑠 𝑃3 + 𝑃4 + 𝑏𝑝
∆𝑇𝐺𝑃𝑠𝑝 = 𝑃1 ∗ (𝑡𝑤𝑒 − 𝑡𝑤𝑐𝑒)𝑃2 ∗ 𝑞𝑠 𝑃3 + 𝑃4

4 𝑇𝐷𝐺𝑠𝑝 = 𝑃1 ∗ (𝑡𝑤𝑒 − 𝑡𝑤𝑐𝑒) + 𝑃2 ∗ 𝑞𝑠 𝑃3 + 𝑃4 + 𝑏𝑝
∆𝑇𝐺𝑃𝑠𝑝 = 𝑃1 ∗ (𝑡𝑤𝑒 − 𝑡𝑤𝑐𝑒) + 𝑃2 ∗ 𝑞𝑠 𝑃3 + 𝑃4

5 𝑇𝐷𝐺𝑠𝑝 = 𝑃1 ∗ (1 − 𝑒 𝑃2∗ 𝑞𝑠 ) + 𝑃3 ∗ (𝑇𝑒𝑚𝑝𝑡𝑤 − 𝑃4) + 𝑏𝑝
∆𝑇𝐷𝐺𝑠𝑝 = 𝑃1 ∗ (1 − 𝑒 𝑃2∗ 𝑞𝑠 ) + 𝑃3 ∗ (𝑇𝑒𝑚𝑝𝑡𝑤 − 𝑃4)

*TDGsp = spillway discharge total gas pressure (mmHg), ∆TGPsp = spillway discharge gas pressure (mmHg),
bp = observed barometric pressure (mmHg), twe = observed project tailwater elevation (feet), twce = pro-
ject specific tailwater channel elevation (feet), twe - twce = tailwater channel depth (feet), Temptw =
tailwater temperature (oC), Qsp = total project spillway discharge (kcfs), qs = flow weighted specific spillbay
discharge (kcfs), P1 - P4 = project specific coefficients (unitless) and are determined from nonlinear regres-
sion analyses.
Scheider and Hamilton (2015b) provided coefficient values for 11 federal dams on the Columbia and
Snake River system as shown in Table 22.
163 Water Quality

Table 22. Coefficients for Spillway TDG Production in SYSTDG from Scheider and Hamilton (2015b).
Dam E1 E2 P1 P2 P3 P4 TWCE SPAT
Dworshak Dam (DWR) (# of
0.0 0.0 410.00 0.0 -0.210 -26.70 0.0 -
bays known)
Dworshak Dam (DWR) (de-
0.0 0.0 288.20 0.0 -0.510 0.0 0.0 -
fault)
Dworshak Dam (DWR) (RO) 0.0 0.0 240.00 0.0 -0.32 0.0 0.0 -
Lower Granite Dam (LWG) 1.000 0.0 5.300 1.000 -0.123 0.0 585.0 8
Little Goose Dam (LGS) 1.200 0.0 5.566 1.000 -0.080 0.0 480.0 6
Lower Monumental Dam
0.800 0.0 1.000 1.100 0.500 0.0 405.0* 8
(LMN)
Ice Harbor Dam (IHR) 1.000 0.0 0.156 1.345 0.7956 60.970 320.0 8
McNary Dam (MCN) 0.650 0.0 5.620 1.000 -0.119 0.0 228.0 21
John Day Dam (JDA) 0.150 0.0 1.000 1.000 0.671 23.500 135.0 20
Dalles Dam (TDA) 0.0 0.0 4.870 1.000 0.0 -207.51 0.0* 22
Bonneville Dam (BON) 0.0 0.0 1.560 13.660 1.000 23.030 0.0 20
Chief Joseph Dam (CHJ) 0.0 0.0 1.000 1.160 0.260 0.000 743.0 19
Grand Coulee Dam (GCL)
1.380 22.600 0.0 0.0 0.0 312.00 0.0 0
(fbe < 1266)
0.0 0.0 400.00 0.0 -0.020 0.0 0.0 0
(fbe > =1266)
1.0 0.0 316.96 -0.098 1.085 11.4 0.0 0
(fbe > =1266)
*For LMN, the value for TWCE in the input tab was 400.00, but the value used in the TDG equation was
405.0. For TDA, the TWCE is specified as 68.0, but in order to fit the equation properly, a value of 0 had to
be used (Scheider and Hamilton, 2015a, 2015b).
The unit spillway discharge was a surrogate measure for the velocity, momentum, and exposure time of
aerated flow associated with spillway discharge. The higher the unit spillway discharge, the greater the TDG
exchange during spillway flows. Flow weighted specific spillbay discharge can be actual measured dis-
charges through each spillbay or computed as a function of the spill pattern:
∑𝑛𝑏 𝐶
𝑖=1 𝑄𝑖
𝑞𝑠 = (𝐶−1) (4-121)
∑𝑛𝑏
𝑖=1 𝑄𝑖
Where Qi = discharge through spillbay i, nb = the number of project spillbays, C = Project and spill pattern
specific constant.
Table 23 lists C values for 11 federal dams on the Columbia and Snake River system (Scheider and Hamilton,
2015b) ranging from 1 to 4.01.
Water Quality 164

Table 23. Spill Pattern Identification with Exponential Coefficient (C) specified in SYSTDG*
Spill Pattern Name in Spill Pattern Name
Dam When to Use this Spill Pattern C
FPP in SYSTDG
LWG Spring SP with RSW Spring From April 3-June 20 1
LWG Summer SP with RSW Summer From June 21-August 31 2
LWG No RSW No RSW Use when flows <30 kcfs 2
LGS Low Crest with SW Low Crest** Use when flows >85 kcfs 2
Start of spill season and when
LGS High Crest with SW High Crest** 2
flows 35-85 kcfs
LGS No SW No RSW Use when flows < 35 kcfs 2
Used during the periods that
the spillway weir is being
LGS SW change or closure Change N/A1
changed from Hi to Lo or closed
completely.
Start of spill season and when
LMN Bulk with RSW Bulk flows <140 kcfs or full PH does 4
not exist
Use when flows >140 kcfs or
LMN Uniform with RSW Uniform 2
when full PH exist
LMN No RSW No RSW Use when flows < 30 kcfs 2
When spill operation =
IHR 45 kcfs/TDG Spill Cap 45 kcfs/SC 1.2
45kcfs/SC
IHR 30% Spill with RSW 30% When spill operation = 30% 2
IHR No RSW No RSW Use when flows < 30 kcfs 2
Use when flows < 259 kcfs and
MCN With TSWs in 19 & 20 With TSW 3.01
April 10 - June 7
Use when flows > 259 kcfs; and
MCN No TSW No TSW 4.01
June 8 – Aug 31
Used when flows are <165 kcfs,
MCN Navigation with TSW Navigation and there is a navigational lock- 2.01
age & the barge want it
MCN TSW Removal Removal When TSWs are being removed 2.01
Use when spill <305 kcfs and
JDA With TSWs in 18 &19 With TSW 2.01
April 10 –August 31
Used before, during and after
TSW removal or installation;
JDA No TSW No TSW Also when spill >305 kcfs or 1.01
flows > 685 kcfs (this is the flood
pattern)
TDA 40% Spill 40% April 10-August 31 N/A2
BON Vertical gate openings Vertical Openings April 10-August 31 2
CHJ --- Center First All times 2.01
DWR --- Spillway1 Use when 1 spillbay is used N/A3
DWR --- Spillway2 Use when 2 spillbays are used N/A3
DWR --- RO Use with Regulating Outlet spill N/A3
DWR --- No Spill Use when there is NO spill N/A3
GCL --- --- No spill pattern is used -
* Spill pattern name: FPP, SW, RSW, TSW
1. Not currently implemented in SYSTDG.
165 Water Quality

2. Spillway TDG production equation is a function of tailwater elevation only. No C value is used.
3. C values are not used in DWR TDG production equations.
The interaction of powerhouse flows and the highly aerated spillway flows can be considerable at many
dams. The powerhouse discharge can be entrained with the spillway discharge, becoming “gassed” to the
same level as the spillway. In SYSTDG, the entrainment of powerhouse flows is computed as a simple linear
function of spillway flows. Without spillway discharge, there is no mechanism to attract powerhouse flows
into the spillway region and no air-entrainment mechanism to drive local TDG supersaturation. As spillway
flows increase relative to powerhouse flows, both bubble production and water entrainment ramp up to a
point at which nearly all powerhouse flows are exposed to bubbles entrained at the spillway. Table 24 lists
three equations used to calculate the entrainment of powerhouse flow in SYSTDG (Scheider and Hamilton,
2015a).
Table 24. List of powerhouse flow entrainment equations in SYSTDG

Equation # Powerhouse Flow Entrainment Equation*
1 Qent = E1 * Qsp + E2
2 Qent = min[ (Qtot / 60) , 1] * E1 * Qsp + E2
3 Qent = min[ (Qsp / 20) , 1] * E1 * Qsp + E2
*Qent = Total powerhouse flow that is entrained in spillway flow (kcfs), Qtot = total project discharge (pow-
erhouse and spillway) (kcfs), E1 - E2 = project specific coefficients (unitless).
E1 and E2 are model coefficients that represent the site-specific geometric configuration of the spillway in
relation to the outflow direction of the powerhouse. Their values for 11 dams on the Columbia and Snake
River system are listed in Table 2. Qent calculated from equations listed in Table 24 must be less than or
equal to the total powerhouse discharge:
Qent = min(Qent, Qtot - Qsp ) (4-122)
In SYSTDG, the average tailrace mixing TDG pressures generated from a project are computed from the flow
weighted average TDG pressures of the spillway and the powerhouse. TDG pressures of flows released from
the powerhouse are assumed equivalent to the TDG pressure in the forebay.
𝑇𝐷𝐺𝑠𝑝 (𝑄𝑠𝑝 +𝑄𝑒𝑛𝑡 )+𝑇𝐷𝐺𝑝ℎ (𝑄𝑝ℎ −𝑄𝑒𝑛𝑡 )

𝑇𝐷𝐺𝑟𝑒𝑙 = 𝑄𝑝ℎ +𝑄𝑠𝑝
(4-123)
Where TDGrel = project release TGP after mixing (mmHg), TDGsp = spillway TGP (mmHg), TDGph = release
TGP through the powerhouse turbines (mmHg), Qsp = Total project spill (kcfs), Qph = Total flow through
powerhouse turbines (kcfs).
The equation above for tailrace TDG during spill events. Qent, Qsp, Qph are known (gauged) or assumed (fore-
cast). The entrained powerhouse flow was assumed to be exposed to the same conditions that spillway
flows encounter and, hence, achieve the same pressures. The TDG associated with the powerhouse release
is generally assumed to be equivalent to the TDG observed in the forebay. TDGph is either from a gauge
Water Quality 166

(boundary condition) or routed from upstream of the project (DWR is an exception). TDGsp is computed
from the spillway TDG production equations listed in Table 21.
Dissolved Oxygen Saturation Computations
The dissolved oxygen saturation is computed from Mortimer’s (1981) formulation:
𝛷𝑂2 𝑠𝑎𝑡 = 𝑃𝑎𝑙𝑡 𝑒 (7.7117−1.31403[𝑙𝑛{𝑇+45.93}]) (4-124)
where:
T = water temperature, C
𝐻 5.25
Palt = altitude correction factor=(1 − ) ,
44.3
H = elevation of the waterbody, km above sea level
The saturation value is also a function of the chlorinity in saline environments. The following equation is
used to compute DO saturation for a saltwater waterbody type that includes salinity effects (APHA, 1985;
EPA 1985; Thomann and Mueller, 1987).
1.0754𝑋101 2.1407𝑋103
(𝑙𝑛(𝛷𝑂2𝑠𝑎𝑡 )−𝑆[1.7674𝑋10−2 − + ])
𝛷∗ 𝑂2𝑠𝑎𝑡 = 𝑒 𝑇 𝑇2 (4-125)
where:
S = salinity, kg m-3
T = temperature, oK
The section on N2 gas under Generic Constituents shows the computation procedure for N2 gas saturation.
Sediments
Organic sediment contributions to nutrients and dissolved oxygen demand are simulated using two meth-
ods. The first method uses a constant, or zero-order, release and demand. This method has been frequent-
ly used to model sediment demands and nutrient release rates. It does not depend on sediment concen-
trations or require a separate sediment compartment. However, the formulation is not predictive as the
rates do not vary over time except as a result of temperature dependence of the decay rate. As a conse-
quence, results should be interpreted cautiously when evaluating effects of different nutrient loadings on
dissolved oxygen in a waterbody.
The second method uses a sediment compartment to accumulate organic sediments and allow their decay.
Nutrient releases and oxygen demand are thus dependent upon sediment accumulation – a 1st-order pro-
cess. However, there is no release of phosphorus or other diagenesis products when overlying water is
anoxic since this sediment compartment is labile, oxic decay of organics on the sediment surface. A later
version will include a fully predictive sediment diagenesis model. Either of these methods, or a combina-
tion, may be used to simulate effects of organic sediments upon water quality. However, caution must be
exercised to avoid errors if both methods are used.
The 0-order process uses a specified sediment oxygen demand and anoxic release rates for phosphorus,
ammonium, inorganic carbon, and iron that are temperature dependent. Nutrient releases do not occur
when dissolved oxygen concentrations are above a minimum value [O2LIM] nor do they occur if the SOD is
167 Water Quality

SEDIMENT KINETICS
set to zero. The sediment contribution to inorganic carbon is computed as a fraction of the sediment oxygen
demand.
Inorganic C
Phosphate
uptake anoxic
Dissolved oxygen Sediment
release Ammonium
Silica
Figure 95. Internal flux between 0-order sediment compartment and other compartments.
CBOD Dissolved oxygen

Inorganic C
Algae decay
Phosphate
settling decay
LPOM Sediment
mortality Ammonium
RPOM burial
Epiphyton Silica
settling
Figure 96. Internal flux between 1st-order sediment compartment and other compartments.
The 1st-order sediment compartment requires specifying inclusion of this compartment in the simulation, a
decay rate, and initial conditions. The sediment compartment is not transported, except for focusing of
sediments to the bottom by a user-defined settling velocity. Instead, a compartmental equation is written
for the sediment compartment that is solved in the subroutine SEDIMENT.
Referring to Figure 96, the equation for the 1st-order sediment compartment is:
S sed =  POMR bottom  POMR +  POML bottom  POML +   a bottom  a -  om K s  s + K epom K eb e −
A A A
Volcell Volcell Volcell    
        sediment decay epiphyton burial
POMR sedimentat ion POML sedimentat ion algae sedimentat ion
 SED Abottom  CBOD Abottom

s +  CBOD - K
burial  s
 
Volcell Volcell
      sediment burial
sediment sedimentat ion CBOD sedimentat ion
(4-126)
where:
OM = rate multiplier for organic matter
POM = POM settling velocity, m sec-1
a = algal settling velocity, m sec-1
CBOD = particulate CBOD settling velocity, m sec-1
SED = sediment settling velocity, m sec-1
Water Quality 168

KINETICS

LPOM = POM labile concentration, g m-3
RPOM = POM refractory concentration, g m-3
s = organic sediment concentration, g m-3
Volcell = volume of computational cell, m3
Abottom=Area of bottom, m2
e = ephiphyton concentration, g m-3
Kepom=fraction of ephiphyton that go to particulate fraction and settle into sediment at death
Kem = ephiphyton mortality rate
Kburial= sediment burial rate, sec-1
Pauer and Auer (2000) claim that nitrification is generally a sediment-based phenomena rather than one
existing in the water column. Their argument is based on the study of a samples obtained from a hyper-
eutrophic lake-river system which had extremely high ammonia concentrations (2-10 mg/l NH3-N) and
where cell counts indicated much higher nitrifier populations in the sediments rather than in the water
column. They question if nitrification is being modeled correctly in many water quality models, where ni-
trification is treated as occurring in the water column rather than the sediments.
Nitrification rates of water column samples were measured in 2.5 L bottles kept in the dark at 20° Celsius.
Rates of nitrification were to be determined by tracking concentrations of nitrogen species in the bottles.
Since no change in ammonia or nitrate-nitrite concentrations were noted, the authors assumed that the
lack of observable nitrification was due to low nitrifier populations occurring in the water column of the
lake and river. Sediment nitrification rates were measured using sediment core samples under similar con-
ditions. Changes in the concentrations of nitrogen species were apparent and an average sediment nitrifi-
cation rate of 0.34 g N/m2/day was determined. Table 25 summarizes the nitrification rates.
Table 25. Areal nitrification rates of sediment nitrifiers of Onondaga Lake and the Seneca River, New
York (Pauer and Auer, 2000).
Reaction Source Growth Temperature Nitrification rate (g N/m2/day)

(Celsius)
Nitrification Onondaga Lake, Seneca River, 20° 0.21 to 0.67 (mean=0.34)
New York
Sediment Variable Stoichiometry and Kinetics

Variable stoichiometry of sediments is done automatically within the W2 model. In addition to sediment
phosphorus and sediment nitrogen, there is also a sediment carbon compartment. The sediment carbon
stoichiometry is variable because organic matter, algae and periphyton/epiphyton may have differing car-
bon stoichiometry. The initial sediment stoichiometry for P, N, and C is based on the given initial stoichi-
ometry for organic matter. The decay rate of sediment in a model cell is the mass averaged decay rate of
the LPOM, RPOM, and the CBOD groups.
169 Water Quality

KINETICS
Sediment Phosphorus
 POMR Abottom  POML Abottom  a Abottom 
S Psed =  RPOM − P +  LPOM − P +  Pa  a - K
 s − P +  K epom K em Pe  e −

s 
Volcell Volcell Volcell  
    sediment decay epiphyton mortality
RPOM- P sedimentation LPOM - P sedimentation algae sedimentation
 SED Abottom
s−P
Volcell
 
sediment sedimentation
where:
Pe = epiphyton/periphyton stoichiometric coefficient for phosphorus
LPOM-P = labile POM labile concentration, g m-3
RPOM-P = refractory POM concentration, g m-3
s-P = sediment phosphorus concentration, g m-3
Abottom= Area of bottom, m2
Kepom= fraction of epiphyton that go to particulate fraction and settle into sediment at death
Kem = epiphyton mortality rate
Sediment Nitrogen
 POMR Abottom  POML Abottom  a Abottom
S Nsed =  RPOM − N +  LPOM − N +   Na  a - K

 s  s − N +  K epom K em  Ne  e −
  
Vol Vol Vol cell 
cell
 cell    sediment decay epiphyton mortality
RPOM- N sedimentation LPOM - N sedimentation algae sedimentation
 SED Abottom  NO3 SED Abottom

s−N +  NO3− N f NO3− SED
Vol cell Vol cell
 
sediment sedimentation diffusion of NO3 into sediments
where:
Pe = epiphyton stoichiometric coefficient for phosphorus
Water Quality 170

KINETICS

NO3 = NO3-N diffusion rate into sediments, m sec-1
LPOM-N = labile POM concentration, g m-3
RPOM-N = refractory POM concentration, g m-3
NO3-N = NO3-N concentration, g m-3
s-N = sediment nitrogen concentration, g m-3
fNO3-SED= fraction of NO3-N diffused into sediments that is incorporated into organic matter in the sedi-
ments (the rest, 1- fNO3-SED, is denitrified into N2)
Sediment Carbon
 POMR Abottom   POML Abottom   a Abottom 
SCsed = COM  RPOM + COM  LPOM +  Ca  a - K
 s − C +  K epom K em Ce e −

s 
Volcell Volcell Volcell  
      sediment decay epiphyton mortality
RPOM sedimentation LPOM sedimentation algae sedimentation
 SED Abottom
 s −C
Volcell
 
sediment sedimentation
where:
POML=Labile POM settling velocity, m sec-1
POMR = Refractory POM settling velocity, m sec-1
Ce = epiphyton stoichiometric coefficient for carbon
Ca = algal stoichiometric coefficient for carbon
COM =organic matter stoichiometric coefficient for carbon
LPOM = labile POM carbon concentration, g m-3
RPOM = refractory POM carbon concentration, g m-3
s-C = sediment carbon concentration, g m-3
171 Water Quality

KINETICS
Sediment Diagenesis Model

The sediment diagenesis model in CE-QUAL-W2 Version 4 and later is based on work from ERM and Golder
(2011), Prakash et al. (2015), and Berger and Wells (2014) that were supported by CEMA (Cumulative Envi-
ronmental Management Association, Canada). The ERM and Golder (2011) report is included in the model
download package and describes those model processes. Processes described in the Berger and Wells
(2014) report are included below.
Sediment Phosphorus
A schematic of the sediment phosphate model is shown in Figure 97.
Figure 98 and Figure 99 show the phosphate sources and sinks in the aerobic and anaerobic compart-
ments, respectively, in the CE-QUAL-W2 model framework.
Figure 97. Schematic of sediment phosphate model (DiToro, 2001).
Water Quality 172

KINETICS
Water Layer
Above Sediment
Diffusion
Phosphate
Diffusion and
Particle Mixing
Sediment Layer 2
Figure 98. Internal flux between phosphate within the aerobic sediment layer 1 and other compart-
ments.
Sediment Layer 1
Diffusion
and Particle
Mixing
POP Phosphate
Diagenesis
Figure 99. Internal flux between phosphate within the anaerobic sediment layer 2 and other compart-
ments.
Because the aerobic layer is assumed to be very thin, oxidation of particulate organic phosphorus is as-
sumed to be negligible. The source/sink rate equation for phosphate in sediment layer 1 is assumed to be
zero:
𝑆𝑃𝑂41 = 0
The rate equation for phosphate in sediment layer 2 is:
173 Water Quality

KINETICS
𝑆𝑃𝑂42 = ∑ 𝑘𝑃𝑂𝑃𝑖 𝛾𝑃𝑂𝑃 Φ𝑃𝑂𝑃𝑖

⏟
𝑖=1
Diagenesis
where:
𝛾𝑃𝑂𝑃 = particulate organic phosphorus temperature rate multiplier
𝑘𝑃𝑂𝑃𝑖 = particulate organic phosphorus class 𝑖 mineralization rate, sec-1

Φ𝑃𝑂𝑃𝑖 = particulate organic carbon class 𝑖 concentration, g m-3
The three classes correspond to labile, refractory, and inert/slow refractory particulate organic phospho-
rus for i=1, i=2, and i=3, respectively. Phosphate can exist in dissolved and particulate forms. Dissolved
oxygen concentrations determine the extent to which dissolved phosphate sorbs to iron oxyhydroxide
particulates. Within the aerobic layer phosphate sorbs to iron oxyhydroxide. At low dissolved oxygen
concentrations, iron oxyhydroxide is reduced and the sorbed phosphate is released.
Mass balance equations for layer 1 (aerobic) and layer 2 (anaerobic) are:
𝑑Φ𝑃𝑂41
𝐻1 = −𝐾𝐿01 (𝑓𝑑 Φ𝑃𝑂41 − Φ𝑃𝑂40 ) + 𝐾𝐿12 (𝑓𝑑 Φ𝑃𝑂42 − 𝑓𝑑 Φ𝑃𝑂41 )
𝑑𝑡
+ 𝑤12 (𝑓𝑝 Φ𝑃𝑂42 − 𝑓𝑝 Φ𝑃𝑂41 ) − 𝑤2 Φ𝑃𝑂41
𝑑Φ𝑃𝑂42
𝐻2 = −𝐾𝐿12 (Φ𝑃𝑂42 − Φ𝑃𝑂41 ) − 𝑤12 (𝑓𝑝 Φ𝑃𝑂42 − 𝑓𝑝 Φ𝑃𝑂41 ) + 𝑤2 (Φ𝑃𝑂41 − Φ𝑃𝑂42 )
𝑑𝑡
+ 𝐻2 𝑆𝑃𝑂42
where:
Φ𝑃𝑂41 = total phosphate concentration in layer 1, g m-3
Φ𝑃𝑂42 = total phosphate concentration in layer 2, g m-3
Φ𝑃𝑂40 = phosphate concentration in water column, g m-3
𝐻1 = height of layer 1, m
𝐾𝐿01 = mass transfer coefficient between water column and layer 1, m s-1
𝐾𝐿12 = mass transfer coefficient between layer 1 and layer 2, m s-1
𝑓𝑑 = dissolved fraction of phosphate
𝑓𝑝 = particulate fraction of phosphate
𝑤12 = particle mixing velocity between layer 1 and layer 2, m s-1
𝑤2 = burial velocity, m s-1
The implicit finite difference scheme for layer 1 is

𝑡+Δ𝑡 𝑡
Φ𝑃𝑂41 − Φ𝑃𝑂41
𝐻1
Δ𝑡
𝑡+Δ𝑡 𝑡+Δ𝑡 𝑡+Δ𝑡
= −𝐾𝐿01 (𝑓𝑑 Φ𝑃𝑂41 − Φ𝑃𝑂40 ) + 𝐾𝐿12 (𝑓𝑑 Φ𝑃𝑂42 − 𝑓𝑑 Φ𝑃𝑂41 )
+ 𝑤12 (𝑓𝑝 Φ𝑃𝑂42 − 𝑓𝑝 Φ𝑃𝑂41 ) − 𝑤2 Φ𝑃𝑂41
Rearranging
𝐻1 𝑡+Δ𝑡 𝑡+Δ𝑡
(− − 𝑓𝑑 𝐾𝐿01 − 𝑓𝑑 𝐾𝐿12 − 𝑤12 𝑓𝑝 − 𝑤2 ) Φ𝑃𝑂41 + (𝐾𝐿12 𝑓𝑑 + 𝑤12 𝑓𝑝 )Φ𝑃𝑂42
Δ𝑡
𝐻1 𝑡
= − Φ𝑃𝑂41 − 𝐾𝐿01 Φ𝑃𝑂40
Δ𝑡
Water Quality 174

KINETICS
And the finite difference scheme for layer 2
𝑡+Δ𝑡 𝑡
Φ𝑃𝑂42 − Φ𝑃𝑂42
𝐻2
Δ𝑡
𝑡+Δ𝑡 𝑡+Δ𝑡 𝑡+Δ𝑡 𝑡+Δ𝑡
= −𝐾𝐿12 (𝑓𝑑 Φ𝑃𝑂42 − 𝑓𝑑 Φ𝑃𝑂41 ) − 𝑤12 (𝑓𝑝 Φ𝑃𝑂42 − 𝑓𝑝 Φ𝑃𝑂41 )
𝑡+Δ𝑡 𝑡+Δ𝑡
+ 𝑤2 (Φ𝑃𝑂41 − Φ𝑃𝑂42 ) + 𝐻2 𝑆𝑃𝑂42
Rearranging
𝑡+Δ𝑡
𝐻2 𝑡+Δ𝑡
(𝑓𝑑 𝐾𝐿12 + 𝑓𝑝 𝑤12 + 𝑤2 )Φ𝑃𝑂41 + (− − 𝑓𝑑 𝐾𝐿12 − 𝑓𝑝 𝑤12 − 𝑤2 ) Φ𝑃𝑂42
Δ𝑡
𝐻2 𝑡
= −𝐻2 𝑆𝑃𝑂42 − Φ
Δ𝑡 𝑃𝑂42
The fractions associated with dissolved and particulate forms can be calculated with (Chapra, 1997):
1
𝑓𝑑 = ( )
𝜑 + 𝐾𝑑𝑝 (1 − 𝜑)𝜌
𝐾𝑑𝑝 (1 − 𝜑)𝜌
𝑓𝑝 = ( )
𝜑 + 𝐾𝑑𝑝 (1 − 𝜑)𝜌
where:
𝑓𝑑 = dissolved fraction of phosphate
𝑓𝑝 = particulate fraction of phosphate
𝜑 = sediment porosity
𝜌 = sediment density, g m-3
𝐾𝑑𝑝 = phosphorus sorption coefficient, m3 g-1
Sediment Production of Methane and Sulfide

The modeling of methane (CH4) and hydrogen sulfide (H2S) constituents were added to the water column.
CH4 is modeled as mg/l as C. H2S is modeled as mg/l as S. For both constituents, the anaerobic release
from the sediments, aerobic decay, and reaeration were modeled (Figure 100).
175 Water Quality

KINETICS
Atmosphere
reaeration
decay Inorganic
CH4 Carbon
anaerobic release
Sediment
Atmosphere
reaeration
decay
H2S Sulfate
anaerobic release
Sediment
Figure 100. Sources and sinks for methane and hydrogen sulfide.
The rate equations for methane and hydrogen sulfide were identical (assuming saturation values in the
atmosphere of 0 mg/l for both gases):
A
S = SOD  OM  SODR sed + Asur K L (-  )−T −20 K C
 V     
reaeration decay
0-order sediment release
where:
Asur = surface area of surface computational cell, m2
KL = interfacial exchange rate, m sec-1
OM = organic matter temperature rate multiplier
SODR = sediment release rate of H2S or CH4, fraction of SOD
 = constituent concentration (H2S or CH4), g m-3
Water Quality 176

KINETICS
 = temperature rate multiplier

K = decay rate, sec-1
The basic physics of gas transfer are the same for H2S, CH4 and O2. Using the penetration theory for gas
Df 1
transfer, i.e., Ka = 2 where f is the turbulence frequency of surface renewal, D is the molecular
 h
diffusion coefficient for O2 and h is the depth, once the reaeration coefficient for dissolved oxygen is de-
termined, then the value of the reaeration coefficient 𝑘𝐻2𝑆 for H2S is determined from the following equa-
tion (Thibodeaux, 1996):
𝐷𝐻2𝑆
𝑘𝐻2𝑆 = 𝑘𝑂2 √
𝐷𝑂2
Using
DA MWB
= where MW is the molecular weight of the component
DB MWA
then
𝑀𝑊𝑂2 0.25
𝑘𝐻2𝑆 = 𝑘𝑂2 ( ) .
𝑀𝑊𝐻2𝑆
Given the molecular weights of oxygen and hydrogen sulfide :
𝑀𝑊𝑂2 =32.00 g/mol

𝑀𝑊𝐻2𝑆 =34.08 g/mol
the reaeration coefficient for hydrogen sulfide is:
𝑀𝑊𝑂2 0.25
𝑘𝐻2𝑆 = 𝑘𝑂2 ( ) = 𝑘𝑂2 × 0.984
𝑀𝑊𝐻2𝑆
Likewise for methane

𝑀𝑊𝐶𝐻4 =16.04 g/mol
The reaeration coefficient 𝑘𝐶𝐻4 is:
𝑀𝑊𝑂2 0.25
𝑘𝐶𝐻4 = 𝑘𝑂2 ( ) = 𝑘𝑂2 × 1.188
𝑀𝑊𝐶𝐻4
Consumption of Dissolved Oxygen due to Sediment Re-

suspension
Particulate organic matter in the sediments can be resuspended by turbulence caused by wind or by bot-
tom scour caused by high current velocities. Once particulate organic matter has been re-suspended into
an aerobic water column, it will decay and consume dissolved oxygen.
177 Water Quality

KINETICS
Wind Induced Resuspension
This algorithm for predicting the resuspension of particulate organic matter due to wind is based on work
from Kang et al. (1982) where the bottom shear stress is computed based on wind speed, wind fetch and
depth. This algorithm is shown in Chapter 2 in the section Sediment Resuspension.
Bottom Scour Resuspension
Following the approach of Edinger (2002) in the 3D model GLLVHT, the CE-QUAL-W2 model can compute
the bottom scour of sediment (inorganic or organic) in units of g/m 2/s from Nielson (1992) as
𝑀
𝐵𝑜𝑡𝑡𝑜𝑚 𝑠𝑐𝑜𝑢𝑟 𝑟𝑎𝑡𝑒 [ ] = 𝑉𝑠𝑐𝑜𝑢𝑟 𝐶𝑏𝑜𝑡𝑡𝑜𝑚
𝐿2 𝑇
where the scour velocity 𝑉𝑠𝑐𝑜𝑢𝑟 (m/s) is calculated from
𝜃′ 0.6
0.00033 (𝜃 − 1) (𝑆𝑔 − 1) 𝑔0.6 𝑑0.8
𝑐
𝑉𝑠𝑐𝑜𝑢𝑟 =
𝜈
and
𝑔= acceleration due to gravity, 9.78 m2/s
𝑑= particle diameter, m
𝜏
𝜃′= Shield’s parameter, 𝜃 ′ = 𝑏𝑜𝑡𝑡𝑜𝑚
[𝑔(𝑆𝑔 −1)𝑑]
𝜃𝑐 =critical Shield’s parameter [User input for each particle size]
2
𝑔𝑈𝑏𝑜𝑡𝑡𝑜𝑚
𝜏𝑏𝑜𝑡𝑡𝑜𝑚 = bottom shear stress in m2/s2, when using the Chezy coefficient, 𝜏𝑏𝑜𝑡𝑡𝑜𝑚 =
𝐶2
𝑈𝑏𝑜𝑡𝑡𝑜𝑚 = bottom velocity, m/s
𝐶= Chezy coefficient in units, m0.5/s
𝑠𝑔 = specific gravity of solid,solid/water
𝜈= molecular viscosity of water, 𝜈 = 1.79𝐸 − 6exp (−0.0266𝑇)
𝑉𝑠 Δ𝑧
𝐶𝑏𝑜𝑡𝑡𝑜𝑚 = bottom sediment concentration in mg/l, 𝐶𝑏𝑜𝑡𝑡𝑜𝑚 = 𝐶𝐾𝐵 𝑒𝑥𝑝 ( ) or a minimum bot-
𝐷𝑧
tom concentration defined between 1.0 mg/l (0.02 mm particle diameter and S g=1.2), 3.0 mg/l
(0.2 mm particle diameter and Sg=1.8), and 5.0 mg/l (2.0 mm particle diameter and Sg=2.2) [from
Schubel et al. (1978)]
𝑉𝑠 = settling velocity of particles, m/s
Δ𝑧 = vertical grid spacing at the bottom, m
𝐶𝐾𝐵 =Bottom concentration of suspended solids, mg/l
𝐷𝑧 = vertical mass turbulent diffusion coefficient, m2/s
𝑇= water temperature, oC
The model user can input a value of critical Shields parameter in the control file or one can have the
model estimate it based on the following approach. Cao et al. (2006) used a simple estimate of the criti-
cal Shield’s parameter based on the particle Reynolds number explicitly. The particle Reynolds number, R,
is defined as
𝑅 = 𝑑√𝑆𝑔 𝑔𝑑
The critical Shield’s criterion is then
0.3542
[1+(0.0223𝑅)2.8358 ]
𝜃𝑐 = 3.0946𝑅0.6769
for R between 6.6 and 282.8
𝜃𝑐 = 0.1414𝑅−0.2306 for R < 6.6
𝜃𝑐 = 0.045 for R > 282.8
Water Quality 178

KINETICS
Mass Balance Equations for Particulate Organic Matter

The loss of particulate organic matter to the water column due to resuspension needs to be included in
the mass balance equations for particulate organic carbon (POC), particulate organic nitrogen (PON), and
particulate organic phosphorus (POP). The mass balance equation for POC in the anaerobic layer (layer 2)
is (DiToro, 2001):
𝑑Φ𝑃𝑂𝐶𝑖 (𝑇−20)
𝐻2 = −𝑘𝑃𝑂𝐶𝑖 𝜃𝑃𝑂𝐶𝑖 Φ𝑃𝑂𝐶𝑖 𝐻2 − 𝑤2 Φ𝑃𝑂𝐶𝑖 + 𝑓𝑃𝑂𝐶𝑖 𝐽𝑃𝑂𝐶 − 𝐸𝑃𝑂𝐶𝑖
𝑑𝑡
where:
Φ𝑃𝑂𝐶𝑖 = POC concentration in reactivity class 𝑖 in layer 2, g m-3
𝐽𝑃𝑂𝐶 = depositional flux of POC from water column, g m-2 s-1
𝐸𝑃𝑂𝐶𝑖 = resuspension rate of reactivity class 𝑖, g m-2 s-1
𝑘𝑃𝑂𝐶𝑖 = reaction rate of reactivity class 𝑖, s-1
𝑓𝑃𝑂𝐶𝑖 = fraction of 𝐽𝑃𝑂𝐶 that is in reactivity class 𝑖
𝜃𝑃𝑂𝐶𝑖 = temperature coefficient for reactivity class 𝑖
𝑇2 = temperature of layer 2, °C

𝑡+Δ𝑡 𝑡
Φ𝑃𝑂𝐶𝑖 − Φ𝑃𝑂𝐶𝑖 (𝑇−20) 𝑡+Δ𝑡 𝑡+Δ𝑡
𝐻2 = −𝑘𝑃𝑂𝐶𝑖 𝜃𝑃𝑂𝐶𝑖 Φ𝑃𝑂𝐶𝑖 𝐻2 − 𝑤2 Φ𝑃𝑂𝐶𝑖 + 𝑓𝑃𝑂𝐶𝑖 𝐽𝑃𝑂𝐶 − 𝐸𝑃𝑂𝐶𝑖
Δ𝑡
Rearranging
𝑡+Δ𝑡 𝑡
Φ𝑃𝑂𝐶𝑖 (𝑇−20) 𝑡+Δ𝑡 𝑡+Δ𝑡 Φ𝑃𝑂𝐶𝑖
𝐻2 + 𝑘𝑃𝑂𝐶𝑖 𝜃𝑃𝑂𝐶𝑖 Φ𝑃𝑂𝐶𝑖 𝐻2 + 𝑤2 Φ𝑃𝑂𝐶𝑖 = 𝐻2 + 𝑓𝑃𝑂𝐶𝑖 𝐽𝑃𝑂𝐶 − 𝐸𝑃𝑂𝐶𝑖
Δ𝑡 Δ𝑡
𝑡+Δ𝑡
And solving for Φ𝑃𝑂𝐶𝑖 :
−1
𝑡+Δ𝑡 𝑡 Δ𝑡 Δ𝑡 𝑤2 Δ𝑡 (𝑇−20)
Φ𝑃𝑂𝐶𝑖 = (Φ𝑃𝑂𝐶𝑖 + 𝑓 𝐽 − 𝐸 ) (1 + + 𝑘𝑃𝑂𝐶𝑖 𝜃𝑃𝑂𝐶𝑖 Δ𝑡)
𝐻2 𝑃𝑂𝐶𝑖 𝑃𝑂𝐶 𝐻2 𝑃𝑂𝐶𝑖 𝐻2
The mass balance equations for PON and POP would be solved for in analogous fashion.
Dynamic Calculation of Sediment pH and Temperature

pH
The dynamic calculation of pH was implemented using a similar approach as a pH model developed by the
United States Geological Survey (Sullivan et al, 2013). In order to calculate pH, sediment organic carbon,
sediment alkalinity, and sediment temperature also had to be modeled.
Sediment Total Inorganic Carbon
A schematic of the sediment total inorganic carbon model is shown in Figure 101. Figure 102 and Figure
103 show the total inorganic carbon sources and sinks in the aerobic and anaerobic compartments, respec-
tively, in the CE-QUAL-W2 model framework.
179 Water Quality

KINETICS
Figure 101. Schematic of sediment inorganic carbon model.
Water Layer
Above Sediment
Diffusion
decay CH4
Inorganic
Carbon
POC
e
Diffusion
Sediment Layer 2
Figure 102 Internal flux between inorganic carbon within the aerobic sediment layer 1 and other com-
partments.
Water Quality 180

KINETICS
Sediment Layer 1
Diffusion
Inorganic
POC
Methanogenesis Carbon
Sulfide Diagenesis
Figure 103 Internal flux between inorganic carbon within the anaerobic sediment layer 2 and other
compartments.
The rate equation for total inorganic carbon (TIC) in the aerobic sediment layer 1 is:
𝑆𝑇𝐼𝐶1 = ∑ 𝑘𝑃𝑂𝐶1 𝛾𝑃𝑂𝐶 Φ𝑃𝑂𝐶 + ⏟

𝑘𝐶𝐻41 𝛾𝐶𝐻4 Φ𝐶𝐻4
⏟
CH4 decay
POC decay
where:
𝛾𝑃𝑂𝐶 = particulate organic carbon temperature rate multiplier
𝛾𝐶𝐻4 = methane temperature rate multiplier
𝑘𝑃𝑂𝐶1 = POC decay rate in layer 1, sec-1
𝑘𝐶𝐻41 = methane decay rate in layer 1, sec-1
𝑘𝑃𝑂𝐶𝑚2 = methanogenesis rate in layer 2, sec-1
𝑘𝑃𝑂𝐶𝑠2 = sulfide diagenesis rate in layer 2, sec-1
Φ𝑃𝑂𝐶 = particulate organic carbon concentration, g m-3
Φ𝐶𝐻4 = methane concentration, g m-3
The rate equation for TIC in the anaerobic sediment layer 2 is:
𝑆𝑇𝐼𝐶2 = ∑ 𝑘𝑃𝑂𝐶𝑚2 𝛾𝑃𝑂𝐶 Φ𝑃𝑂𝐶 + ⏟

𝑘𝑃𝑂𝐶𝑠2 𝛾𝑃𝑂𝐶 Φ𝑃𝑂𝐶
⏟
Sulfide Diagenesis
Methanogenesis
The mass balances for layer 1 and layer 2 are:
𝑑Φ 𝑇𝐼𝐶1
𝐻1 = −𝐾𝐿01 (Φ 𝑇𝐼𝐶1 − Φ 𝑇𝐼𝐶0 ) + 𝐾𝐿12 (Φ 𝑇𝐼𝐶2 − Φ 𝑇𝐼𝐶1 ) + 𝐻1 𝑆𝑇𝐼𝐶1
𝑑𝑡
𝑑Φ 𝑇𝐼𝐶2
𝐻2 = −𝐾𝐿12 (Φ 𝑇𝐼𝐶2 − Φ 𝑇𝐼𝐶1 ) + 𝐻2 𝑆𝑇𝐼𝐶2
𝑑𝑡
181 Water Quality

KINETICS
where:
Φ 𝑇𝐼𝐶1 = total inorganic carbon concentration in layer 1, g m-3
Φ 𝑇𝐼𝐶2 = total inorganic carbon concentration in layer 2, g m-3
Φ 𝑇𝐼𝐶0 = total inorganic carbon concentration in water column, g m-3
The mass transfer coefficient 𝐾𝐿12 is related to the diffusion coefficient 𝐷 using
𝐷 𝐷
𝐾𝐿12 = =
Δ𝑧 1 (𝐻 + 𝐻 )
2 1 2
Φ𝑡+Δ𝑡 𝑡
𝑇𝐼𝐶1 − Φ 𝑇𝐼𝐶1
𝐻1 = −𝐾𝐿01 (Φ𝑡+Δ𝑡 𝑡+Δ𝑡 𝑡+Δ𝑡
𝑇𝐼𝐶1 − Φ 𝑇𝐼𝐶0 ) + 𝐾𝐿12 (Φ 𝑇𝐼𝐶2 − Φ 𝑇𝐼𝐶1 ) + 𝐻1 𝑆𝑇𝐼𝐶1
Δ𝑡
Rearranging
𝐻1 𝐻1 𝑡
(− − 𝐾𝐿01 − 𝐾𝐿12 ) Φ𝑡+Δ𝑡 𝑡+Δ𝑡
𝑇𝐼𝐶1 + 𝐾𝐿12 Φ 𝑇𝐼𝐶2 = −𝐻1 𝑆𝑇𝐼𝐶1 − Φ − 𝐾𝐿01 Φ 𝑇𝐼𝐶0
Δ𝑡 Δ𝑡 𝑇𝐼𝐶1
Φ𝑡+Δ𝑡 𝑡
𝑇𝐼𝐶2 − Φ 𝑇𝐼𝐶2
𝐻2 = −𝐾𝐿12 (Φ𝑡+Δ𝑡 𝑡+Δ𝑡
𝑇𝐼𝐶2 − Φ 𝑇𝐼𝐶1 ) + 𝐻2 𝑆𝑇𝐼𝐶2
Δ𝑡
Rearranging
𝐻2 𝐻2 𝑡
𝐾𝐿12 Φ𝑡+Δ𝑡
𝑇𝐼𝐶1 + (− − 𝐾𝐿12 ) Φ𝑡+Δ𝑡
𝑇𝐼𝐶2 = −𝐻2 𝑆𝑇𝐼𝐶2 − Φ
Δ𝑡 Δ𝑡 𝑇𝐼𝐶2
Sediment Alkalinity
Sources and sinks for non-conservative sediment alkalinity model is shown in Figure 104. Alkalinity is in
units of mg/l as CaCO3.
Water Quality 182

KINETICS
Figure 104. Sources and sinks of the sediment alkalinity model.
The rate equation for alkalinity in the aerobic sediment layer 1 is:
2 equivalents alkalinity
𝑆𝐴𝐿𝐾1 = − ( )𝑘 𝛾 Φ
⏟ 1 mole ammonium nitrified 𝑁𝐻4 𝑁𝐻4 𝑁𝐻4
nitrification
1 equivalent alkalinity
+( )𝑘 𝛾 Φ
⏟1 mole nitrate denitrified 𝑁𝑂𝑥 𝑁𝑂𝑥1 𝑁𝑂𝑥
denitrification
where:
𝛾𝑁𝑂𝑥 = denitrification temperature rate multiplier
𝛾𝑁𝐻4 = nitrification temperature rate multiplier
𝑘𝑁𝑂𝑥1 = denitrification rate in aerobic layer, sec-1
𝑘𝑁𝐻4 = nitrification rate, sec-1
Φ𝑁𝑂𝑥 = nitrate-nitrite concentration, g m-3
Φ𝑁𝐻4 = ammonium concentration, g m-3
The rate equation for alkalinity in the anaerobic sediment layer 2 is:
1 equivalent alkalinity
𝑆𝐴𝐿𝐾2 = ( )𝑘 𝛾 Φ
⏟1 mole nitrate denitrified 𝑁𝑂𝑥 𝑁𝑂𝑥 𝑁𝑂𝑥
denitrification
183 Water Quality

KINETICS
where:
𝑘𝑁𝑂𝑥2 = denitrification rate in anaerobic layer, sec-1
The mass balances for layer 1 and layer 2 are:
𝑑Φ𝐴𝐿𝐾1
𝐻1 = −𝐾𝐿01 (Φ𝐴𝐿𝐾1 − Φ𝐴𝐿𝐾0 ) + 𝐾𝐿12 (Φ𝐴𝐿𝐾2 − Φ𝐴𝐿𝐾1 ) + 𝐻1 𝑆𝐴𝐿𝐾1
𝑑𝑡
𝑑Φ𝐴𝐿𝐾2
𝐻2 = −𝐾𝐿12 (Φ𝐴𝐿𝐾2 − Φ𝐴𝐿𝐾1 ) + 𝐻2 𝑆𝐴𝐿𝐾2
𝑑𝑡
where:
Φ𝐴𝐿𝐾1 = alkalinity concentration in layer 1, g m-3
Φ𝐴𝐿𝐾2 = alkalinity concentration in layer 2, g m-3
Φ𝐴𝐿𝐾0 = alkalinity concentration in water column, g m-3
The implicit finite difference scheme of alkalinity is equivalent to that used for inorganic carbon.
Sediment Temperature
A schematic of the sediment temperature model is shown in Figure 105.
Water Quality 184

KINETICS
Figure 105. Schematic of sediment temperature model.
The heat balances for the aerobic layer 1 and anaerobic layer 2 are:
𝑑T1
𝜌𝐶𝑝 𝐻1 = −𝐾𝐿01 𝜌𝐶𝑝 (T1 − T0 ) + 𝐾𝐿12 𝜌𝐶𝑝 (T2 − T1 )
𝑑𝑡
𝑑T2
𝜌𝐶𝑝 𝐻2 = −𝐾𝐿12 𝜌𝐶𝑝 (T2 − T1 ) + 𝐾𝑠𝑤 (T𝑠𝑒𝑑 − T2 )
𝑑𝑡
where:
T1 = temperature in layer 1, °C
T2 = temperature in layer 2, °C
T0 = temperature in water column, °C
T𝑠𝑒𝑑 = temperature of sediments below anaerobic layer, °C
𝐾𝐿01 = heat transfer coefficient between water column and layer 1, m s-1
𝐾𝐿12 = heat transfer coefficient between layer 1 and layer 2, m s-1
𝐾𝑠𝑤 = coefficient of sediment heat exchange between anaerobic layer and sediments below anaerobic
layer, W m-2 °C-1
𝜌 = water density, g m-3
𝐶𝑝 = water density, J g-1 °C-1
185 Water Quality

KINETICS
T1𝑡+Δ𝑡 − T1𝑡
𝜌𝐶𝑝 𝐻1 = −𝐾𝐿01 𝜌𝐶𝑝 (T1𝑡+Δ𝑡 − T0 ) + 𝐾𝐿12 𝜌𝐶𝑝 (T2𝑡+Δ𝑡 − T1𝑡+Δ𝑡 )
Δ𝑡
Rearranging and cancelling out 𝜌𝐶𝑝
𝐻1 𝐻1
(− − 𝐾𝐿01 − 𝐾𝐿12 ) T1𝑡+Δ𝑡 + 𝐾𝐿12 T2𝑡+Δ𝑡 = − T1𝑡 − 𝐾𝐿01 T0
Δ𝑡 Δ𝑡
T2𝑡+Δ𝑡 − T2𝑡
𝜌𝐶𝑝 𝐻2 = −𝐾𝐿12 𝜌𝐶𝑝 (T2𝑡+Δ𝑡 − T1𝑡+Δ𝑡 ) + 𝐾𝑠𝑤 (T𝑠𝑒𝑑 − T2𝑡+Δ𝑡 )
Δ𝑡
Rearranging
𝜌𝐶𝑝 𝐻2 𝜌𝐶𝑝 𝐻2 𝑡
𝐾𝐿12 𝜌𝐶𝑝 T1𝑡+Δ𝑡 + (− − 𝜌𝐶𝑝 𝐾𝐿12 + 𝐾𝑠𝑤 ) T2𝑡+Δ𝑡 = − T2 − 𝐾𝑠𝑤 T𝑠𝑒𝑑
Δ𝑡 Δ𝑡
Water Quality 186

KINETICS
Metal Complexation and Diagenesis

Ferrous Iron Fe(II)
Iron is modeled using two constituents: iron oxyhydroxide, FeOOH(s), and ferrous iron Fe(II). FeOOH(s) is
very insoluble and is modeled as a particulate whereas Fe(II) is more soluble.
Figure 106 illustrates the iron model within the water column and sediments.
187 Water Quality

KINETICS
Figure 106. Schematic of iron model (Ditoro, 2001).
The modeling of ferrous iron in the water column was added to CE-QUAL-W2. The rate equation in the
water column is:
2(𝑝𝐻−7)
𝐾𝐹𝑒𝑂𝑂𝐻
𝑆𝐹𝑒(𝐼𝐼) = −𝑘
⏟ 𝐹𝑒(𝐼𝐼)0 Φ𝐷𝑂 10 Φ𝐹𝑒(𝐼𝐼) + 𝑘𝐹𝑒𝑂𝑂𝐻 ( )Φ
⏟ 𝐾𝐹𝑒𝑂𝑂𝐻 + Φ𝐷𝑂 𝐹𝑒𝑂𝑂𝐻
Oxidation
𝑅𝑒𝑑𝑢𝑐𝑡𝑖𝑜𝑛
where:
𝑘𝐹𝑒𝑂𝑂𝐻 = reduction rate in water column, sec-1
𝑘𝐹𝑒(𝐼𝐼)0 = oxidation rate in water column, sec-1(g m-3)-1
𝐾𝐹𝑒𝑂𝑂𝐻 =half saturation constant for oxygen for this reaction, g m-3
Φ𝐷𝑂 = dissolved oxygen concentration, g m-3
Φ𝐹𝑒(𝐼𝐼) = ferrous iron concentration, g m-3
Φ𝐹𝑒𝑂𝑂𝐻 = iron oxyhydroxide concentration, g m-3
𝑝𝐻 =pH in water column
The rate equation for ferrous iron in sediment layer 1 is:
Water Quality 188

KINETICS
2(𝑝𝐻1−7)
𝑆𝐹𝑒(𝐼𝐼)1 = −𝑘
⏟ 𝐹𝑒(𝐼𝐼)1 Φ𝐷𝑂 10 𝑓𝑑 Φ𝐹𝑒(𝐼𝐼)1
Oxidation
where:
𝑓𝑑 = dissolved fraction of ferrous iron
𝑘𝐹𝑒(𝐼𝐼)1 = ferrous iron oxidation rate in layer 1, sec-1(g m-3)-1
𝑝𝐻1 = pH in layer 1
The rate equation for ferrous iron in sediment layer 2 is:
𝑆𝐹𝑒(𝐼𝐼)2 = ⏟
𝑘𝐹𝑒𝑂𝑂𝐻2 Φ𝐹𝑒𝑂𝑂𝐻
Reduction
where:
𝑘𝐹𝑒𝑂𝑂𝐻2 = ferrous iron reduction rate in layer 2, sec-1
Mass balances for the aerobic layer 1 and anaerobic layer 2 are:
𝑑Φ𝐹𝑒(𝐼𝐼)1
𝐻1 = −𝐾𝐿01 (𝑓𝑑 Φ𝐹𝑒(𝐼𝐼)1 − Φ𝐹𝑒(𝐼𝐼)0 ) + 𝐾𝐿12 (𝑓𝑑 Φ𝐹𝑒(𝐼𝐼)2 − 𝑓𝑑 Φ𝐹𝑒(𝐼𝐼)1 )
𝑑𝑡
+ 𝑤12 (𝑓𝑝 Φ𝐹𝑒(𝐼𝐼)2 − 𝑓𝑝 Φ𝐹𝑒(𝐼𝐼)1 ) − 𝑤2 Φ𝐹𝑒(𝐼𝐼)1 + 𝐻1 𝑆𝐹𝑒(𝐼𝐼)1
𝑑Φ𝐹𝑒(𝐼𝐼)2
𝐻2 = −𝐾𝐿12 (𝑓𝑑 Φ𝐹𝑒(𝐼𝐼)2 − 𝑓𝑑 Φ𝐹𝑒(𝐼𝐼)1 ) − 𝑤12 (𝑓𝑝 Φ𝐹𝑒(𝐼𝐼)2 − 𝑓𝑝 Φ𝐹𝑒(𝐼𝐼)1 )
𝑑𝑡
+ 𝑤2 (Φ𝐹𝑒(𝐼𝐼)1 − Φ𝐹𝑒(𝐼𝐼)2 ) + 𝐻2 𝑆𝐹𝑒(𝐼𝐼)2
where:
Φ𝐹𝑒(𝐼𝐼)0 =ferrous iron concentration in water column, g m-3
𝑓𝑝 = particulate fraction of ferrous iron
189 Water Quality

KINETICS
𝑡+Δ𝑡 𝑡
Φ𝐹𝑒(𝐼𝐼)1 − Φ𝐹𝑒(𝐼𝐼)1
𝐻1
Δ𝑡
= −𝐾𝐿01 (𝑓𝑑 Φ𝐹𝑒(𝐼𝐼)1 − Φ𝐹𝑒(𝐼𝐼)0 ) + 𝐾𝐿12 (𝑓𝑑 Φ𝐹𝑒(𝐼𝐼)2 − 𝑓𝑑 Φ𝐹𝑒(𝐼𝐼)1 )
+ 𝑤12 (𝑓𝑝 Φ𝐹𝑒(𝐼𝐼)2 − 𝑓𝑝 Φ𝐹𝑒(𝐼𝐼)1 ) − 𝑤2 Φ𝐹𝑒(𝐼𝐼)1
𝑡+Δ𝑡
− 𝐻1 𝑘𝐹𝑒(𝐼𝐼)1 Φ𝐷𝑂 102(𝑝𝐻1−7) 𝑓𝑑 Φ𝐹𝑒(𝐼𝐼)1
Rearranging
𝐻1 𝑡+Δ𝑡
(− − 𝑓𝑑 𝐾𝐿01 − 𝑓𝑑 𝐾𝐿12 − 𝑤12 𝑓𝑝 − 𝑤2 − 𝐻1 𝑘𝐹𝑒(𝐼𝐼)1 Φ𝐷𝑂 102(𝑝𝐻1−7) 𝑓𝑑 ) Φ𝐹𝑒(𝐼𝐼)1
Δ𝑡
𝑡+Δ𝑡 𝐻1 𝑡
+ (𝐾𝐿12 𝑓𝑑 + 𝑤12 𝑓𝑝 )Φ𝐹𝑒(𝐼𝐼)2 = − Φ𝐹𝑒(𝐼𝐼)1 − 𝐾𝐿01 Φ𝐹𝑒(𝐼𝐼)0
Δ𝑡
𝑡+Δ𝑡 𝑡
Φ𝐹𝑒(𝐼𝐼)2 − Φ𝐹𝑒(𝐼𝐼)2
𝐻2
Δ𝑡
= −𝐾𝐿12 (𝑓𝑑 Φ𝐹𝑒(𝐼𝐼)2 − 𝑓𝑑 Φ𝐹𝑒(𝐼𝐼)1 ) − 𝑤12 (𝑓𝑝 Φ𝐹𝑒(𝐼𝐼)2 − 𝑓𝑝 Φ𝐹𝑒(𝐼𝐼)1 )
+ 𝑤2 (Φ𝐹𝑒(𝐼𝐼)1 − Φ𝐹𝑒(𝐼𝐼)2 ) + 𝐻2 𝑆𝐹𝑒(𝐼𝐼)2
Rearranging
𝑡+Δ𝑡 𝐻2 𝑡+Δ𝑡
(𝑓𝑑 𝐾𝐿12 + 𝑓𝑝 𝑤12 + 𝑤2 )Φ𝐹𝑒(𝐼𝐼)1 + (− − 𝑓𝑑 𝐾𝐿12 − 𝑓𝑝 𝑤12 − 𝑤2 ) Φ𝐹𝑒(𝐼𝐼)2
Δ𝑡
𝐻2 𝑡
= −𝐻2 𝑆𝐹𝑒(𝐼𝐼)2 − Φ
Δ𝑡 𝐹𝑒(𝐼𝐼)2
1
𝑓𝑑 = ( )
𝜑 + 𝐾𝑑𝐹𝑒 (1 − 𝜑)𝜌
𝐾𝑑𝐹𝑒 (1 − 𝜑)𝜌
𝑓𝑝 = ( )
𝜑 + 𝐾𝑑𝐹𝑒 (1 − 𝜑)𝜌
where:
𝑓𝑝 = particulate fraction of ferrous iron
𝐾𝑑𝐹𝑒 = ferrous sorption coefficient, m3 g-1
Iron Oxyhydroxide FeOOH(s)
The rate equation in the water column for iron oxyhydroxide is:
𝐾𝐹𝑒𝑂𝑂𝐻
𝑘𝐹𝑒(𝐼𝐼)0 Φ𝐷𝑂 102(𝑝𝐻−7) Φ𝐹𝑒(𝐼𝐼) − 𝑘𝐹𝑒𝑂𝑂𝐻 (
𝑆𝐹𝑒𝑂𝑂𝐻 = ⏟ ) Φ𝐹𝑒𝑂𝑂𝐻
⏟ 𝐾𝐹𝑒𝑂𝑂𝐻 + Φ𝐷𝑂
Oxidation
Water Quality 190

KINETICS
where:
𝑘𝐹𝑒𝑂𝑂𝐻 = reduction rate in water column, sec-1
𝑘𝐹𝑒(𝐼𝐼)0 = oxidation rate in water column, sec-1(g m-3)-1
𝐾𝐹𝑒𝑂𝑂𝐻 =half saturation constant for oxygen for this reaction, g m-3
𝑝𝐻=pH in water column
The rate equation for iron oxyhydroxide in the aerobic sediment layer 1 is:
2(𝑝𝐻1−7)
𝑆𝐹𝑒𝑂𝑂𝐻1 = 𝑘
⏟𝐹𝑒(𝐼𝐼)1 Φ𝐷𝑂 10 𝑓𝑑 Φ𝐹𝑒(𝐼𝐼)1
Oxidation
where:
𝑘𝐹𝑒(𝐼𝐼)1 = ferrous iron oxidation rate in layer 1, sec-1(g m-3)-1
The rate equation for iron oxyhydroxide in the anaerobic sediment layer 2 is:
𝑆𝐹𝑒𝑂𝑂𝐻2 = ⏟
−𝑘𝐹𝑒𝑂𝑂𝐻2 Φ𝐹𝑒𝑂𝑂𝐻
Reduction
where:
𝑘𝐹𝑒𝑂𝑂𝐻2 = ferrous iron reduction rate in layer 2, sec-1
Mass balances for layer 1 and layer 2 are:
𝑑Φ𝐹𝑒𝑂𝑂𝐻1
𝐻1 = 𝑤12 (Φ𝐹𝑒𝑂𝑂𝐻2 − Φ𝐹𝑒𝑂𝑂𝐻1 ) − 𝑤2 Φ𝐹𝑒𝑂𝑂𝐻1 + 𝑤1 Φ𝐹𝑒𝑂𝑂𝐻0 + 𝐻1 𝑆𝐹𝑒𝑂𝑂𝐻1
𝑑𝑡
𝑑Φ𝐹𝑒𝑂𝑂𝐻2
𝐻2 = −𝑤12 (Φ𝐹𝑒𝑂𝑂𝐻2 − Φ𝐹𝑒𝑂𝑂𝐻1 ) + 𝑤2 (Φ𝐹𝑒𝑂𝑂𝐻1 − Φ𝐹𝑒𝑂𝑂𝐻2 ) + 𝐻2 𝑆𝐹𝑒𝑂𝑂𝐻2
𝑑𝑡
where:
Φ𝐹𝑒𝑂𝑂𝐻0 =iron oxyhydroxide concentration in water column, g m-3
𝑤1 = particle settling velocity, m s-1
191 Water Quality

KINETICS
𝑡+Δ𝑡 𝑡
Φ𝐹𝑒𝑂𝑂𝐻1 − Φ𝐹𝑒𝑂𝑂𝐻1
𝐻1
Δ𝑡
= 𝑤12 (Φ𝐹𝑒𝑂𝑂𝐻2 − Φ𝐹𝑒𝑂𝑂𝐻1 ) − 𝑤2 Φ𝐹𝑒𝑂𝑂𝐻1 + 𝑤1 Φ𝐹𝑒𝑂𝑂𝐻0
2(𝑝𝐻1−7) 𝑡
+ 𝐻1 𝑘𝐹𝑒(𝐼𝐼)1 Φ𝐷𝑂 10 𝑓𝑑 Φ𝐹𝑒(𝐼𝐼)1
Rearranging
(− − 𝑤12 − 𝑤2 ) Φ𝐹𝑒𝑂𝑂𝐻1 + 𝑤12 Φ𝐹𝑒𝑂𝑂𝐻2
Δ𝑡
𝐻1 𝑡 𝑡
= − Φ𝐹𝑒𝑂𝑂𝐻1 − 𝑤1 Φ𝐹𝑒𝑂𝑂𝐻0 − 𝐻1 𝑘𝐹𝑒(𝐼𝐼)1 Φ𝐷𝑂 102(𝑝𝐻1−7) 𝑓𝑑 Φ𝐹𝑒(𝐼𝐼)1
Δ𝑡
𝑡+Δ𝑡 𝑡
Φ𝐹𝑒𝑂𝑂𝐻2 − Φ𝐹𝑒𝑂𝑂𝐻2
𝐻2
Δ𝑡
= −𝑤12 (Φ𝐹𝑒𝑂𝑂𝐻2 − Φ𝐹𝑒𝑂𝑂𝐻1 ) + 𝑤2 (Φ𝐹𝑒𝑂𝑂𝐻1 − Φ𝐹𝑒𝑂𝑂𝐻2 )
𝑡+Δ𝑡
− 𝐻2 𝑘𝐹𝑒𝑂𝑂𝐻 Φ𝐹𝑒𝑂𝑂𝐻2
Rearranging
𝑡+Δ𝑡
𝐻2 𝑡+Δ𝑡
𝐻2 𝑡
(𝑤12 + 𝑤2 )Φ𝐹𝑒𝑂𝑂𝐻1 + (− − 𝑤12 − 𝑤2 − 𝐻2 𝑘𝐹𝑒𝑂𝑂𝐻 ) Φ𝐹𝑒𝑂𝑂𝐻2 = − Φ𝐹𝑒𝑂𝑂𝐻2
Δ𝑡 Δ𝑡
Water Quality 192

KINETICS
Manganese Mn(II)
The manganese model within the water column and sediments is illustrated in
Figure 106.
193 Water Quality

KINETICS
Water layer above sedi-

ment Oxida-
Mn(II tion MnO2(s
) )
Reduc-
Diffusion MnO2(s tion Mn(II
of ) )
Mn(II)
Partition- Particle
Mn(II)ing Mn(II) Settling
of
Oxida- MnO2(s)
Mn(IItion MnO2(s
) ) Diffusion
Aerobic sediment of
layer Mn(II)
Particle
Mixing
Partition-
of
Mn(II)
ing Mn(II)
MnO2(s)
Reduc-
MnO2(s tion Mn(II
MnO2(s ) )
Burial ) Mn(II Anaerobic sediment
) layer
Model Segment
i
Figure 107. Schematic of manganese model (DiToro, 2001).
The following rate equation is used to model ionic divalent manganese Mn(II) in the water column is:
𝐾𝑀𝑛𝑂2
−𝑘𝑀𝑛(𝐼𝐼)0 Φ𝐷𝑂 102(𝑝𝐻−7) Φ𝑀𝑛(𝐼𝐼) + 𝑘𝑀𝑛𝑂2 (
𝑆𝑀𝑛(𝐼𝐼) = ⏟ ) Φ𝑀𝑛𝑂2
⏟ 𝐾𝑀𝑛𝑂2 + Φ𝐷𝑂
Oxidation
where:
𝑘𝑀𝑛𝑂2 =reduction rate in water column, sec-1
𝑘𝑀𝑛(𝐼𝐼)0 = oxidation rate in water column, sec-1(g m-3)-1
𝐾𝑀𝑛𝑂2 =half saturation constant for oxygen for this reaction, g m-3
Φ𝑀𝑛(𝐼𝐼) = Mn(II) concentration, g m-3
Φ𝑀𝑛𝑂2 = manganese dioxide concentration, g m-3
𝑝𝐻 =pH in water column
Water Quality 194

KINETICS
The rate equation for Mn(II) in the aerobic sediment layer 1 is:
2(𝑝𝐻1−7)
𝑆𝑀𝑛(𝐼𝐼)1 = −𝑘
⏟ 𝑀𝑛(𝐼𝐼)1 Φ𝐷𝑂 10 𝑓𝑑 Φ𝑀𝑛(𝐼𝐼)1
Oxidation
where:
𝑓𝑑 = dissolved fraction of Mn(II)
𝑘𝑀𝑛(𝐼𝐼)1 = Mn(II) oxidation rate in layer 1, sec-1(g m-3)-1
Φ𝑀𝑛(𝐼𝐼)1 =Mn(II) concentration, g m-3
The rate equation for manganese dioxide in the anaerobic sediment layer 2 is:
𝑆𝑀𝑛(𝐼𝐼)2 = ⏟
𝑘𝑀𝑛𝑂2 Φ𝑀𝑛𝑂2
Reduction
where:
𝑘𝑀𝑛𝑂2 =manganese dioxide reduction rate in layer 2, sec-1
Mass balances for layer 1 and layer 2 are then:
𝑑Φ𝑀𝑛(𝐼𝐼)1
𝐻1 = −𝐾𝐿01 (𝑓𝑑 Φ𝑀𝑛(𝐼𝐼)1 − Φ𝑀𝑛(𝐼𝐼)0 ) + 𝐾𝐿12 (𝑓𝑑 Φ𝑀𝑛(𝐼𝐼)2 − 𝑓𝑑 Φ𝑀𝑛(𝐼𝐼)1 )
𝑑𝑡
+ 𝑤12 (𝑓𝑝 Φ𝑀𝑛(𝐼𝐼)2 − 𝑓𝑝 Φ𝑀𝑛(𝐼𝐼)1 ) − 𝑤2 Φ𝑀𝑛(𝐼𝐼)1 + 𝐻1 𝑆𝑀𝑛(𝐼𝐼)1
𝑑Φ𝑀𝑛(𝐼𝐼)2
𝐻2 = −𝐾𝐿12 (𝑓𝑑 Φ𝑀𝑛(𝐼𝐼)2 − 𝑓𝑑 Φ𝑀𝑛(𝐼𝐼)1 ) − 𝑤12 (𝑓𝑝 Φ𝑀𝑛(𝐼𝐼)2 − 𝑓𝑝 Φ𝑀𝑛(𝐼𝐼)1 )
𝑑𝑡
+ 𝑤2 (Φ𝑀𝑛(𝐼𝐼)1 − Φ𝑀𝑛(𝐼𝐼)2 ) + 𝐻2 𝑆𝑀𝑛(𝐼𝐼)2
where:
Φ𝑀𝑛(𝐼𝐼)0 =Mn(II) concentration in water column, g m-3
𝑓𝑝 = particulate fraction of Mn(II)
195 Water Quality

KINETICS
𝑡+Δ𝑡 𝑡
Φ𝑀𝑛(𝐼𝐼)1 − Φ𝑀𝑛(𝐼𝐼)1
𝐻1
Δ𝑡
= −𝐾𝐿01 (𝑓𝑑 Φ𝑀𝑛(𝐼𝐼)1 − Φ𝑀𝑛(𝐼𝐼)0 ) + 𝐾𝐿12 (𝑓𝑑 Φ𝑀𝑛(𝐼𝐼)2 − 𝑓𝑑 Φ𝑀𝑛(𝐼𝐼)1 )
+ 𝑤12 (𝑓𝑝 Φ𝑀𝑛(𝐼𝐼)2 − 𝑓𝑝 Φ𝑀𝑛(𝐼𝐼)1 ) − 𝑤2 Φ𝑀𝑛(𝐼𝐼)1
𝑡+Δ𝑡
− 𝐻1 𝑘𝑀𝑛(𝐼𝐼)1 Φ𝐷𝑂 102(𝑝𝐻1−7) 𝑓𝑑 Φ𝑀𝑛(𝐼𝐼)1
Rearranging
𝐻1 𝑡+Δ𝑡
(− − 𝑓𝑑 𝐾𝐿01 − 𝑓𝑑 𝐾𝐿12 − 𝑤12 𝑓𝑝 − 𝑤2 − 𝐻1 𝑘𝑀𝑛(𝐼𝐼)1 Φ𝐷𝑂 102(𝑝𝐻1−7) 𝑓𝑑 ) Φ𝑀𝑛(𝐼𝐼)1
Δ𝑡
𝑡+Δ𝑡 𝐻1 𝑡
+ (𝐾𝐿12 𝑓𝑑 + 𝑤12 𝑓𝑝 )Φ𝑀𝑛(𝐼𝐼)2 = − Φ𝑀𝑛(𝐼𝐼)1 − 𝐾𝐿01 Φ𝑀𝑛(𝐼𝐼)0
Δ𝑡
𝑡+Δ𝑡 𝑡
Φ𝑀𝑛(𝐼𝐼)2 − Φ𝑀𝑛(𝐼𝐼)2
𝐻2
Δ𝑡
= −𝐾𝐿12 (𝑓𝑑 Φ𝑀𝑛(𝐼𝐼)2 − 𝑓𝑑 Φ𝑀𝑛(𝐼𝐼)1 ) − 𝑤12 (𝑓𝑝 Φ𝑀𝑛(𝐼𝐼)2 − 𝑓𝑝 Φ𝑀𝑛(𝐼𝐼)1 )
+ 𝑤2 (Φ𝑀𝑛(𝐼𝐼)1 − Φ𝑀𝑛(𝐼𝐼)2 ) + 𝐻2 𝑆𝑀𝑛(𝐼𝐼)2
Rearranging
𝑡+Δ𝑡 𝐻2 𝑡+Δ𝑡
(𝑓𝑑 𝐾𝐿12 + 𝑓𝑝 𝑤12 + 𝑤2 )Φ𝑀𝑛(𝐼𝐼)1 + (− − 𝑓𝑑 𝐾𝐿12 − 𝑓𝑝 𝑤12 − 𝑤2 ) Φ𝑀𝑛(𝐼𝐼)2
Δ𝑡
𝐻2 𝑡
= −𝐻2 𝑆𝑀𝑛(𝐼𝐼)2 − Φ
Δ𝑡 𝑀𝑛(𝐼𝐼)2
1
𝑓𝑑 = ( )
𝜑 + 𝐾𝑑𝑀𝑛 (1 − 𝜑)𝜌
𝐾𝑑𝑀𝑛 (1 − 𝜑)𝜌
𝑓𝑝 = ( )
𝜑 + 𝐾𝑑𝑀𝑛 (1 − 𝜑)𝜌
where
𝑓𝑝 = particulate fraction of Mn(II)
𝐾𝑑𝑝 = manganese sorption coefficient, m3 g-1
Manganese Dioxide MnO2
The rate equation in the water column for manganese dioxide is:
𝐾𝑀𝑛𝑂2
𝑘𝑀𝑛(𝐼𝐼)0 Φ𝐷𝑂 102(𝑝𝐻−7) Φ𝑀𝑛(𝐼𝐼) − 𝑘𝑀𝑛𝑂2 (
𝑆𝑀𝑛𝑂2 = ⏟ ) Φ𝑀𝑛𝑂2
⏟ 𝐾𝑀𝑛𝑂2 + Φ𝐷𝑂
Oxidation
where:
𝑘𝑀𝑛𝑂2 =reduction rate in water column, sec-1
Water Quality 196

KINETICS
𝑘𝑀𝑛(𝐼𝐼)0 = oxidation rate in water column, sec-1(g m-3)-1

𝐾𝑀𝑛𝑂2 =half saturation constant for oxygen for this reaction, g m-3
𝑝𝐻=pH in water column
The rate equation for manganese dioxide in the aerobic sediment layer 1 is:
2(𝑝𝐻1−7)
𝑆𝑀𝑛𝑂2 1 = 𝑘
⏟𝑀𝑛(𝐼𝐼)1 Φ𝐷𝑂 10 𝑓𝑑 Φ𝑀𝑛(𝐼𝐼)1
Oxidation
where:
𝑘𝑀𝑛(𝐼𝐼)1 = Mn(II) oxidation rate in layer 1, sec-1(g m-3)-1
The rate equation for manganese dioxide in the anaerobic sediment layer 2 is:
𝑆𝑀𝑛𝑂2 2 = −𝑘
⏟ 𝑀𝑛𝑂2 2 Φ𝑀𝑛𝑂2
Reduction
𝑘𝑀𝑛𝑂2 2 = manganese dioxide reduction rate in layer 2, sec-1

Mass balances for layer 1 and layer 2 are:
𝑑Φ𝑀𝑛𝑂2 1
𝐻1 = 𝑤12 (Φ𝑀𝑛𝑂2 2 − Φ𝑀𝑛𝑂2 1 ) − 𝑤2 Φ𝑀𝑛𝑂2 1 + 𝑤1 Φ𝑀𝑛𝑂2 0 + 𝐻1 𝑆𝑀𝑛𝑂2 1
𝑑𝑡
𝑑Φ𝑀𝑛𝑂2 2
𝐻2 = −𝑤12 (Φ𝑀𝑛𝑂2 2 − Φ𝑀𝑛𝑂2 1 ) + 𝑤2 (Φ𝑀𝑛𝑂2 1 − Φ𝑀𝑛𝑂2 2 ) + 𝐻2 𝑆𝑀𝑛𝑂2 2
𝑑𝑡
where:
Φ𝑀𝑛𝑂2 0 =manganese concentration in water column, g m-3
𝑤1 = particle settling velocity, m s-1

𝑡+Δ𝑡 𝑡
Φ𝑀𝑛𝑂 21
− Φ𝑀𝑛𝑂21
𝐻1
Δ𝑡
= 𝑤12 (Φ𝑀𝑛𝑂 22
− Φ𝑀𝑛𝑂 21
) − 𝑤2 Φ𝑀𝑛𝑂 21
+ 𝑤1 Φ𝑀𝑛𝑂2 0
2(𝑝𝐻1−7) 𝑡
+ 𝐻1 𝑘𝑀𝑛(𝐼𝐼)1 Φ𝐷𝑂 10 𝑓𝑑 Φ𝑀𝑛(𝐼𝐼)1
197 Water Quality

KINETICS
Rearranging
(− − 𝑤12 − 𝑤2 ) Φ𝑀𝑛𝑂 21
+ 𝑤12 Φ𝑀𝑛𝑂 22
Δ𝑡
𝐻1 𝑡 𝑡
= − Φ𝑀𝑛𝑂 21
− 𝑤1 Φ𝑀𝑛𝑂2 0 − 𝐻1 𝑘𝑀𝑛(𝐼𝐼)1 Φ𝐷𝑂 102(𝑝𝐻1−7) 𝑓𝑑 Φ𝑀𝑛(𝐼𝐼)1
Δ𝑡
𝑡+Δ𝑡 𝑡
Φ𝑀𝑛𝑂 22
− Φ𝑀𝑛𝑂22
𝐻2
Δ𝑡
= −𝑤12 (Φ𝑀𝑛𝑂 22
) + 𝑤2 (Φ𝑀𝑛𝑂 21
)
𝑡+Δ𝑡
− 𝐻2 𝑘𝑀𝑛𝑂2 Φ𝑀𝑛𝑂 22
Rearranging
𝑡+Δ𝑡
𝐻2 𝑡+Δ𝑡
𝐻2 𝑡
(𝑤12 + 𝑤2 )Φ𝑀𝑛𝑂 21
+ (− − 𝑤12 − 𝑤2 − 𝐻2 𝑘𝑀𝑛𝑂2 ) Φ𝑀𝑛𝑂 2 =− Φ
Δ𝑡 2
Δ𝑡 𝑀𝑛𝑂2 2
Total Inorganic Carbon

Carbon, hydrogen, and oxygen are the most abundant elements in living matter and form the essential
backbone of organic material. Inorganic carbon directly influences pH.
Macrophytes Atmosphere
Sediment
photosynthesis
reaeration
RDOM
Algae
decay LDOM
Inorganic
respiration
Carbon
LPOM
Epiphyton
respiration
RPOM
photosynthesis Zooplankton
CBOD
Figure 108. Internal flux between inorganic carbon and other compartments.
Referring to Figure 108, the rate equation for total inorganic carbon is:
Water Quality 198

KINETICS
STIC =  (K ar - K ag )  C a  a +  (K e r - K e g )  C e  e + K LPOM  COM  OM  LPOM

    
net algal production net epiphyton production labile POM decay
+ K LDOM  OM  COM  LCOM + K RDOM  OM  COM  RDOM + K RPOM  COM  OM  RPOM

        
labile DOMdecay refractory DOM decay refractory POM decay
A A
+ K s ed  OM  COM  s ed + SOD  OM  COM sed + sur K Lc ( s CO 2 -  CO 2)
   V Vsur
1st - order sediment release    
0 - order sediment release reaeration
+  K CBODRCBOD  C − CBOD 
 
T − 20
 + (K m r - K m g )  C m  macro
CBOD  
CBOD decay net macrophyte production
+ K zr  Czoo  zoo  zoo


zooplankton respiration
(4-127)
where:
 = temperature rate multiplier for BOD decay
OM = organic matter temperature rate multiplier
Ce = epiphyton stoichiometric coefficient for carbon
Ca = algal stoichiometric coefficient for carbon
COM = organic matter stoichiometric coefficient for carbon
C-CBOD = CBOD stoichiometric coefficient for carbon
RBOD = 5-day CBOD to ultimate CBOD ratio
Asur = surface area of surface computational cell, m2
KLc = inorganic carbon interfacial exchange rate, m sec-1
TIC = inorganic carbon concentration, g m-3
CO2 = carbon dioxide concentration, g m-3
199 Water Quality

KINETICS
sCO2 = carbon dioxide saturation concentration, g m-3
and the rate terms are evaluated in subroutine INORGANIC_CARBON.
The basic physics of gas transfer are the same for CO 2 and O2. Using Higbie penetration theory, the gas
transfer coefficient for CO2 is related to that of oxygen by:
𝐷𝐶𝑂2
𝐾𝐿𝐶𝑂2 = 𝐾𝐿𝑂2 √ (4-128)
𝐷𝑂2
where:
KLCO2 = reaeration coefficient for CO2
KLO2 = reaeration coefficient for oxygen
DCO2 = molecular diffusion coefficient for CO2
DO2 = molecular diffusion coefficient for oxygen
Using the fact that the ratio of molecular diffusion coefficients of two gases A and B are related to their
molecular weights, MW (Thibedoux, 1996)
𝐷𝐴 𝑀𝑊𝐵
=√ (4-129)
𝐷𝐵 𝑀𝑊𝐴
then
𝑀𝑊𝑂2 0.25
𝐾𝐿𝐶𝑂2 = 𝐾𝐿𝑂2 ( ) = 0.923𝐾𝐿𝑂2 (4-130)
𝑀𝑊𝐶𝑂2
Hence, the reaeration rate equations presented for oxygen transfer are applicable to CO2 transfer using a
factor of 0.923.
Carbon dioxide concentration for the interfacial exchange rate is determined from total inorganic carbon
and alkalinity in subroutine PH_CO2. The saturation concentration of carbon dioxide is determined from:
′
Φ𝐶𝑂2 = 0.286exp (−0.0314𝑇𝑠 𝑃𝑎 ) (4-131)
where:
CO2 = carbon dioxide saturation concentration, g m-3
Pa = altitude correction factor
Ts = surface cell water temperature, oC
Altitude correction is from Mortimer (1981):
5.25
 H 
Pa =  1 -  (4-132)
 44.3 
where:
H = reservoir elevation from sea level, km
TIC is in units of mg/l as C.
Water Quality 200

KINETICS
Alkalinity
Alkalinity is in units of mg/l as CaCO3 and provides an indication of the buffering capacity of aquatic systems
and their resistance to pH changes from acidic or alkaline loadings. Alkalinity, along with total inorganic
carbon, is used to determine pH and concentrations of carbonate species in subroutine PH_CO2.
Alkalinity can be treated as conservative in the model and its internal rate term is set to zero. Alkalinity
variations though are common in aquatic systems. Whiting, or large precipitations of carbonates, may oc-
cur. Sediment release of carbonates may increase alkalinity in the anoxic zones of many reservoirs. If these
effects are important, a non-conservative model of alkalinity should be used.
An algorithm for non-conservative alkalinity was added to the CE-QUAL-W2 model based on work of Sulli-
van, et al. 2013 based on Stumm and Morgan (1996). This algorithm allows for determining alkalinity vari-
ations over time based on utilization of ammonia and nitrate during photosynthesis, production of ammo-
nia during respiration, nitrification, and denitrification.
pH and Carbonate Species

The pH and carbonate species are computed using water temperature, TDS or salinity, alkalinity, and total
inorganic carbon concentrations using basic carbonate relationships. Computed carbonate species are car-
bon dioxide, bicarbonates, and carbonates. The pH and carbonate species are computed in subroutine
PH_CO2. These materials are not subject to transport and are computed at each water quality update
interval.
Calculations performed by subroutine PH_CO2 are based on the carbonate-bicarbonate equilibrium reac-
tion (Stumm and Morgan 1996):
HCO3  CO3 + H +
- =
(4-133)
H 2 O  H+ + OH- (4-134)
These equilibria express the source of bicarbonate and carbonate ions, alkaline constituents, and dissolu-
tion of atmospheric CO2 in water. Contribution of calcium and magnesium carbonate to alkalinity is not
included. The equilibrium state in terms of the equilibrium constants Ki is:
[ H + ] [ HCO-3]
K1 = (4-135)
[ H 2 CO3]
[ H + ] [CO3= ]
K2 = (4-136)
[ HCO-3]
[ H + ] [OH-]
Kw = (4-137)
[ H 2 O]
201 Water Quality

KINETICS
where:
[X] = molar concentration, moles liter-1
[H2O] = unity (by definition)
Alkalinity is defined as:
ALK = HCO-3 + 2 CO3= + OH- - H+ (4-138)
The following condition prevails for the dissolution of carbonic acid:
H2 CO3 + HCO-3 + CO3= = CONSTANT = CT (4-139)
By combining equations 53, 54, and 57, the quantities [HCO3-] and [CO3=] can be expressed in terms of [H+]
and the constant CT. In addition, equation 55 allows for [OH-] to be expressed in terms of [H+]. When
these expressions are included in equation 56, the result is:
+
[ H + ] + 2 K2 K w
ALK  = CT [ H ] K1 + + - [H+] (4-140)
[ H + ] K1 + K1 K2 + [ H + ] 2 [H+] [H ]
The model interprets the constant CT as moles per liter of total inorganic carbon and assumes [ALK] is
known. Equation 58 is iteratively solved in subroutine PH_CO2 until the value of [H+] converges. The neg-
ative logarithm of [H+] is, by definition, pH.
Once equation 58 has been solved for [H+], then [H2CO3] is given by:
[ H 2 CO3] = CT
(4-141)
1 + K+1 + K1 K22
[H ] [H+]
and is the same as [CO2]. Bicarbonate concentration is computed from:
[ HCO-3] = CT
+ (4-142)
[H ]
1+ + K+2
K1 [H ]
and carbonate from:
[CO3= ] = C T
+ 2 (4-143)
[ ] [ +]
1+ H + H
K1 K2 K2
which are then converted to grams per cubic meter.
Equilibrium constants in the preceding equations are obtained by first expressing a thermodynamic tem-
perature dependence for a related constant, Ki*:
202
KINETICS
b
log K*i = a + + cT + d log T (4-144)
T
The constants a, b, c, and d are:
a b c d
Kw* 35.3944 -5242.39 -0.00835 11.8261
K1* 14.8435 -3404.71 -0.03278 0
K2* 6.4980 -2902.39 -0.02379 0
The relation between Ki and Ki* is obtained from the definition of the activity of a chemical species:
{𝑋} = 𝛾 [ X ] (4-145)
where:
{X} = activity of species X, moles liter-1
 = dimensionless activity coefficient
[X] = concentration, moles liter-1
For the reaction:
A + B ⇔ C + D (4-146)
the equilibrium constant K* is:
* {C} { D}
K = (4-147)
{ A } { B}
thus:
*  C [C]  D [D]  C  D
K = = K (4-148)
 A [A]  B [B]  A  B
 A B *
K= K (4-149)
 C D
Activity coefficients are obtained from an extension of Debye-Huckel theory as:
2
- AZ I
Log  = + k1 + k 2 I + k 3 I 2 (4-150)
1 + 0.33 a I
203 Water Quality

KINETICS
where:
I = ionic strength
Z = ionic charge
A = approximately 0.5 for water at 25ºC
a = ionic size parameter
ki = empirical coefficients
Ionic strength based on the concentration of Total Dissolved Solids in mg/l is approximated as (Sawyer and
McCarty 1967):
𝐼 = 2.5𝑋105 Φ 𝑇𝐷𝑆 (4-151)
or for salinity in units of kg/m3:
𝐼 = 0.00147 + 0.019885Φ𝑠𝑎𝑙𝑖𝑛𝑖𝑡𝑦 + 0.000038Φ𝑠𝑎𝑙𝑖𝑛𝑖𝑡𝑦 2 (4-152)
where:
TDS = total dissolved solids, g m-3
salinity = salinity, kg m-3
Values of the other parameters are:
Z a k1 k2 k3
HCO-3 1 4 0.0047 0.042 -0.0093
CO=3 2 4.5 0.0121 0.0972 -0.0207
Activity coefficients for [H+], [OH-], [H2CO3], and [H2O] are treated as special cases:
[𝐻 + ]𝛾 = [𝐻2 𝑂]𝛾 = 1 (4-153)
[𝐻2 CO3 ]𝛾 = [OH − ]𝛾=0.0755I (4-154)
Temperature Rate Multipliers

Most biological and chemical rates are temperature dependent. Subroutine RATE_MULTIPLIERS calculates
the temperature dependence for all rates. It is called after the temperature solution so the temperature of
the current computational interval is used.
A representative rate multiplier function is shown in Figure 109 with its K and T parameters. The curve
represents how biological process rates exhibit an optimum range and diminish asymmetrically at higher
and lower temperatures (Thornton and Lessem, 1978).
204
KINETICS
T = 0 whe re T  T 1
 (T- )  ( T -T)
K1 e 1 T1 K4 e 2 4
T = whe re T 1 < T < T 4
1 + K 1 e 1 (T-T 1 ) - K 1 1 + K 4 e 2 ( T 4 -T) - K 4 (4-155)
   
 
ar af
T = 0 whe re T  T 4
where:
1 (1 - K 1 )
 ar = ln K 2
T 2 - T1 K 1 (1 - K 2 )
1 (1 - K 4 )
 af = ln K 3
T4 - T3 K 4 (1 - K 3 )
ar and af are the rising and falling limb temperature multipliers.
The user supplies temperatures T1 to T4 and multiplier factors K1 to K4. Temperatures T1 and T4 represent
mortality limits, and T2 and T3 are used to define the optimum range. Maximum reaction rates supplied by
the user are multiplied by λT to determine rates corresponding to the water temperature of a model cell.
For non-algae temperature rate multipliers, only the K1 and K2 corresponding to T1 and T2 are used.
Note that for generic constituents and CBOD, the model user can choose Arrehnius temperature coeffi-
cients which are of the form
𝑘 𝑇1 = 𝑘 𝑇2 𝜃 𝑇1−𝑇2
where kT is the kinetic first order rate constant in day-1 at a temperature T in oC and is an empirical coef-
ficient dependent on the reaction.
205 Water Quality

KINETICS
1 25, 0.99 35, 0.99
0.9
0.8
Temperature multiplier
0.7
0.6
0.5
0.4
0.3
0.2 5, 0.1 40, 0.1
0.1
0
0 10 20 30 40 50
Temperature, oC
Figure 109. Temperature rate multiplier function.
206
NUMERICAL SOLUTION
5. Numerical Solution
Characteristics of the Finite Difference
Scheme
The solution of the governing equations in CE-QUAL-W2 is based on a finite difference numerical ap-
proach using variables defined in Figure 110 for no channel slope or Figure 111 for arbitrary channel
slope. The finite difference solution has the following characteristics:
• space-staggered grid
• conservative formulation
• regular, structured, Cartesian grid with variable cell length x and cell thickness z
• t for stability requirement computed internally and is updated for each time step
This section outlines the numerical approach for the hydrodynamic equations (solution of U, W, and )
and advective-diffusion equations (T and C or ).
CE-QUAL-W2 coordinate system ,,P,B

U,Ax,Dx, xx
Segment
W,Dz Az
Layer i-1 i i+1
x
z=0 
kt k-1
z
kt+1 k
kb k+1 Hk=zk
xi
z=h at bottom
Figure 110. Computational grid variable definitions for no channel slope.
207
NUMERICAL SOLUTION
CE-QUAL-W2 coordinate system,  > 0 ,, P,B

U, Ax, Dx, BR, xx
i-1 segment
z=0 Az
W, BB, Dz
i
i+1 g

z=zsurface=
kt
k-1
kt+1
Hk=zk
k
h-
k+1
kb
x
z
x z=h

Figure 111. Computational grid variable definitions for arbitrary channel slope.
Hydrodynamic Equations Numerical Solution
Characteristics of the Hydrodynamics Numerical Solu-

tion
• horizontal velocities and water levels are not solved simultaneously; water levels are solved first fol-
lowed by velocities
• branch hydrodynamics are solved sequentially rather than simultaneously (this can cause instabilities
if t is too large, may have to be manually reduced)
• water surface solution is solved implicitly using central spatial differences and the resulting simulta-
neous equations are solved using the Thomas algorithm
• the horizontal momentum equation is solved explicitly using upwind (or backward) spatial differenc-
ing for the advection terms and central spatial differences for all other terms
• the continuity equation is used to solve for vertical velocities using a central difference explicit solu-
tion
• solution order:
208
NUMERICAL SOLUTION
• solve water surface equation for  at time level n+1 (n is the current time level)
• compute U at n+1 using the momentum equation
• compute W at n+1 using the continuity equation
Free-Water Surface Numerical Solution for 

The free surface equation:
𝜕𝜂 𝜕 ℎ ℎ
𝐵𝜂 𝜕𝑡 = 𝜕𝑥 ∫𝜂 𝑈𝐵𝑑𝑧 − ∫𝜂 𝑞𝐵𝑑𝑧 (5-1)
is solved by substituting the momentum equation:
t
+
x
+
z
x
−
  x dz
(5-2)
1 B xx 1  B  x z
+ + + qBU x
 x  z
in finite difference form and then simplifying. The finite difference form of the momentum equation is:
  UUB  WUB 
UBin + 1 = UBin + t  − − + gB sin  + g cos B
 x z x
n
(5-3)
g cos B  1 B xx 1  B x z 
z
−
 
 x
dz +
 x
+
 z
+ qBU x 
i
Defining for simplicity the term F as:
 UUB  WUB 1 B xx

F =− − + (5-4)
x z  x
and substituting in for xx, F becomes:
 U 
  BAx 
 UUB  WUB  x 
F =− − + (5-5)
x z x
209
NUMERICAL SOLUTION
Substituting in the term UBin + 1 in the free surface equation for UB, the free surface equation becomes:
  h n  h n  h
t x  i
B = UB dz + t  F dz + t  gB sin dz
x  x 
 h   h g cos B z 
n n
+ t  g cos B dz − t   x dzdz
x  x x   
(5-6)
 1  B x z 
h n h
+ t 
x    z
dz + t  qBU xn dz
x 
h
−  q n Bdz

Some of these terms can be simplified as follows:
𝜕 ℎ 𝜕 ℎ
∫ 𝑔𝐵 𝑠𝑖𝑛 𝛼 𝑑𝑧
𝜕𝑥 𝜂
= 𝑔 𝑠𝑖𝑛 𝛼 ∫ 𝐵 𝑑𝑧
𝜕𝑥 𝜂
(5-7)
𝜕 ℎ 𝜕𝜂 𝜕 𝜕𝜂 ℎ
∫ 𝑔 𝑐𝑜𝑠 𝛼 𝐵 𝜕𝑥 𝑑𝑧
𝜕𝑥 𝜂
= 𝑔 𝑐𝑜𝑠 𝛼 ( ∫ 𝐵 𝑑𝑧)
𝜕𝑥 𝜕𝑥 𝜂
(5-8)
𝜕 ℎ 𝑔 𝑐𝑜𝑠 𝛼𝐵 𝑧 𝜕𝜌 𝑔 𝑐𝑜𝑠 𝛼 𝜕 ℎ 𝑧 𝜕𝜌
∫
𝜕𝑥 𝜂 𝜌
∫𝜂 𝜕𝑥 𝑑𝑧𝑑𝑧 = 𝜌
∫ 𝐵 ∫𝜂 𝜕𝑥 𝑑𝑧𝑑𝑧
𝜕𝑥 𝜂
(5-9)
 h 1  B x z

x    z
dz =
1 
 x
(
B x z h − B x z  ) (5-10)
Then substituting these into the above equation:
  h  h  h
B =  UBin dz + t  F n dz + tg sin   Bdz
t x  x  x 
    g cos   
n h h z n
+ tg cos    
 x   x
Bdz − t B dzdz (5-11)
x  x 


 1
( )  h h
+ t B x z h − B x z  + t  qBU x dz −  q n Bdz
n n
x  x  
210
NUMERICAL SOLUTION
All terms with  are grouped on the LHS such that:
    h   h
  h n
x  x   x 
B − tg cos  Bdz = UBi dz + t  F dz
n
t  
x 

h
x 
+ tg sin  Bdz
g cos   
h z n
− t  B
 x    x
dzdz
(5-12)
+ t
 1
x 
(
B x z h − B x z  )
n
 h
+ t 
x 
qBU xn dz
h
−  q n Bdz

The first term on the LHS can be put into a backward finite difference form as:
𝜕𝜂 𝜂𝑖𝑛 −𝜂𝑖𝑛−1
𝐵𝜂 𝜕𝑡 ≈ 𝐵𝜂 𝛥𝑡
(5-13)
𝜕 𝜕𝜂 ℎ 𝑛
The second term, −𝛥𝑡𝑔 𝑐𝑜𝑠 𝛼 ( ∫ 𝐵 𝑑𝑧) , can be simplified using the chain rule for partial differential
𝜕𝑥 𝜕𝑥 𝜂
equations to:
𝑛
𝜕𝜂 𝑛 𝜕 ℎ ℎ 𝜕2 𝜂
−𝛥𝑡𝑔 𝑐𝑜𝑠 𝛼 𝜕𝑥 | ∫ 𝐵 𝑑𝑧
𝜕𝑥 𝜂
− 𝛥𝑡𝑔 𝑐𝑜𝑠 𝛼 ∫𝜂 𝐵 𝑑𝑧 𝜕𝑥 2 | (5-14)
Then using a second-order central difference for the second derivative and a first order backward difference
for the first derivative such that:
𝑛
𝜕𝜂 𝑛 𝜕 ℎ ℎ 𝜕2 𝜂
−𝛥𝑡𝑔 𝑐𝑜𝑠 𝛼 𝜕𝑥 | ∫
𝜕𝑥 𝜂
𝐵𝑑𝑧 − 𝛥𝑡𝑔 𝑐𝑜𝑠 𝛼 ∫𝜂
𝐵𝑑𝑧 𝜕𝑥 2 |
𝜂𝑖𝑛 −𝜂𝑖−1
𝑛
𝜕 ℎ ℎ 𝑛
𝜂𝑖+1 −2𝜂𝑖𝑛 +𝜂𝑖−1
𝑛
≈ −𝛥𝑡𝑔 𝑐𝑜𝑠 𝛼 𝛥𝑥
∫ 𝐵 𝑑𝑧
𝜕𝑥 𝜂
− 𝛥𝑡𝑔 𝑐𝑜𝑠 𝛼 ∫𝜂 𝐵𝑑𝑧 𝛥𝑥 2
(5-15)
Also using a backward difference:

𝑛
𝜕 ℎ 1 ℎ ℎ
∫ 𝐵 𝑑𝑧
𝜕𝑥 𝜂
= 𝛥𝑥 (∫𝜂 𝐵𝑑𝑧| − ∫𝜂 𝐵𝑑𝑧| ) . (5-16)
𝑖 𝑖−1
Grouping and collecting terms and multiplying through by tx, the LHS becomes after simplification:
211
NUMERICAL SOLUTION
𝑛 −𝑔 𝑐𝑜𝑠 𝛼𝛥𝑡 2 ℎ 𝑔 𝑐𝑜𝑠 𝛼𝛥𝑡 2 ℎ ℎ

𝜂𝑖−1 [ 𝛥𝑥
∫𝜂 𝐵𝑑𝑧| ] + 𝜂𝑖𝑛 [𝐵𝜂 𝛥𝑥 + 𝛥𝑥
{∫𝜂 𝐵𝑑𝑧| + ∫𝜂 𝐵𝑑𝑧| }] +
𝑖−1 𝑖 𝑖−1
𝑛 −𝑔 𝑐𝑜𝑠 𝛼𝛥𝑡 2 ℎ
𝜂𝑖+1 [ 𝛥𝑥
∫𝜂 𝐵𝑑𝑧| ] = (𝑅𝐻𝑆)𝑛𝑖 𝛥𝑥𝛥𝑡 + 𝐵𝜂 𝜂𝑖𝑛−1 𝛥𝑥 (5-17)
𝑖
where the RHS is defined as:
  
h h h
RHS = 
x 
UBin dz + t  F dz + tg sin 
x  x 
Bdz
g cos     1
( )
h z
− t  B
  x   x
dzdz + t
x 
B x z h − B x z  (5-18)
 h h
+ t  qBU x dz −  qBdz
x  
and is evaluated at time level n.
The integral of the cell widths can be put into a summation over the vertical layers as:
ℎ
∫𝜂 𝐵𝑑𝑧| = ∑𝑘𝑡
𝑘𝑏 𝐵𝐻𝑟𝑖 (5-19)
𝑖
ℎ
∫𝜂 𝐵𝑑𝑧| = ∑𝑘𝑡
𝑘𝑏 𝐵𝐻𝑟𝑖−1 (5-20)
𝑖−1
where BHr is the value of the width times the layer depth for the right-hand side of a cell. In the code, this
is the variable BR(I,K) times H(K), or the derived variable BHR(I,K).
Some of the right-hand side terms can be put into a format compatible with the model schematization such
as:
𝑘𝑏
𝜕 ℎ 𝜕
∫ (𝑈𝐵)𝑛𝑖 𝑑𝑧 ≈ ∑ 𝑈𝐵𝐻𝑟
𝜕𝑥 𝜂 𝜕𝑥
𝑘𝑡
1 1 𝑛
≈ 𝛥𝑥
(∑𝑘𝑏
𝑘𝑡 𝑈𝐵𝐻𝑟 |𝑖 − ∑𝑘𝑏
𝑘𝑡 𝑈𝐵𝐻𝑟 |𝑖−1 ) = 𝑘𝑏
∑ (𝑈𝐵𝐻𝑟 |𝑖 −
𝛥𝑥 𝑘𝑡
𝑈𝐵𝐻𝑟 |𝑖−1 ) (5-21)
𝑘𝑏
𝜕 ℎ 𝜕
𝛥𝑡 ∫ 𝐹 𝑛 𝑑𝑧 ≈ 𝛥𝑡 ∑ 𝐹𝐻𝑟
𝑘𝑡
𝛥𝑡 𝛥𝑡 𝑛
≈ 𝛥𝑥 (∑𝑘𝑏 𝑘𝑏 𝑘𝑏
𝑘𝑡 𝐹𝐻𝑟 |𝑖 − ∑𝑘𝑡 𝐹𝐻𝑟 |𝑖−1 ) = 𝛥𝑥 ∑𝑘𝑡 (𝐹𝐻𝑟 |𝑖 − 𝐹𝐻𝑟 |𝑖−1 ) (5-22)
𝑘𝑏
𝜕 ℎ 𝜕
𝛥𝑡𝑔 𝑠𝑖𝑛 𝛼 ∫ 𝐵𝑑𝑧 ≈ 𝛥𝑡𝑔 𝑠𝑖𝑛 𝛼 ∑ 𝐵𝐻𝑟
𝑘𝑡
𝛥𝑡𝑔 𝑠𝑖𝑛 𝛼 𝛥𝑡𝑔 𝑠𝑖𝑛 𝛼 𝑘𝑏
≈ 𝛥𝑥
(∑𝑘𝑏
𝑘𝑡 𝐵𝐻𝑟 |𝑖 − ∑𝑘𝑏
𝑘𝑡 𝐵𝐻𝑟 |𝑖−1 ) = 𝛥𝑥
∑𝑘𝑡 (𝐵𝐻𝑟 |𝑖 − 𝐵𝐻𝑟 |𝑖−1 ) (5-23)
212
NUMERICAL SOLUTION
𝑔 𝑐𝑜𝑠 𝛼 𝜕 ℎ 𝑧 𝜕𝜌 𝑔 𝑐𝑜𝑠 𝛼 𝜕 ℎ 𝜕𝜌
𝛥𝑡
𝜌
∫ 𝐵 ∫𝜂 𝜕𝑥 𝑑𝑧𝑑𝑧
𝜕𝑥 𝜂
≈ 𝛥𝑡 𝜌
∫ 𝐵 ∑𝑘𝑏
𝜕𝑥 𝜂 𝑘𝑡 𝜕𝑥 𝐻𝑟 𝑑𝑧 ≈
𝑔 𝑐𝑜𝑠 𝛼 𝜕𝜌
𝛥𝑡 𝜌𝛥𝑥 ∑𝑘𝑏 𝑘𝑏
𝑘𝑡 𝜕𝑥 𝐻𝑟 ∑𝑘𝑡 (𝐵𝐻𝑟 |𝑖 − 𝐵𝐻𝑟 |𝑖−1 ) (5-24)
t
 1
x 
(
B x z h − B x z  
t
x
) (
B x z h − B x z  − B x z h − B x z 
i
) ( ) 
i −1
(5-25)
The lateral inflow of momentum term represents the gradient over x of the inflow momentum:
𝜕 ℎ 𝜕
𝛥𝑡 𝜕𝑥 ∫𝜂 𝑞 𝐵𝑈𝑥 𝑑𝑧 ≈ 𝛥𝑡 𝜕𝑥 ∑𝑘𝑏
𝑘𝑡 𝑞𝑈𝑥 𝐵𝐻𝑟 (5-26)
ℎ
∫𝜂 𝑞𝐵𝑑𝑧 ≈ ∑𝑘𝑏
𝑘𝑡 𝑞𝐵𝐻𝑟 (5-27)
Combining these terms into one equation:

𝑛
𝐴𝜂𝑖−1 + 𝑋𝜂𝑖𝑛 + 𝐶𝜂𝑖+1
𝑛
=𝐷 (5-28)
where:
𝑘𝑏
−𝑔 𝑐𝑜𝑠 𝛼 𝛥𝑡 2
𝐴=[ ∑𝐵𝐻𝑟 | ]
𝛥𝑥
𝑘𝑡 𝑖−1
2 𝑘𝑏 𝑘𝑏
𝑔 𝑐𝑜𝑠 𝛼 𝛥𝑡
𝑋 = [𝐵𝜂 𝛥𝑥 + {∑ 𝐵𝐻𝑟 | + ∑ 𝐵𝐻𝑟 | }]
𝛥𝑥
𝑘𝑡 𝑖 𝑘𝑡 𝑖−1
𝑘𝑏
−𝑔 𝑐𝑜𝑠 𝛼 𝛥𝑡 2
𝐶=[ ∑𝐵𝐻𝑟 | ]
𝛥𝑥
𝑘𝑡 𝑖
𝑘𝑏 𝑘𝑏
𝐷 = 𝛥𝑡 ∑(𝑈𝐵𝐻𝑟 |𝑖 − 𝑈𝐵𝐻𝑟 |𝑖−1 ) + 𝐵𝜂 𝜂𝑖𝑛−1 𝛥𝑥 2

+ 𝛥𝑡 ∑(𝐹𝐻𝑟 |𝑖 − 𝐹𝐻𝑟 |𝑖−1 )
𝑘𝑡 𝑘𝑡
𝑘𝑏 𝑘𝑏 𝑘𝑏
𝑔 𝑐𝑜𝑠 𝛼 𝜕𝜌
+𝛥𝑡 2 𝑔 𝑠𝑖𝑛 𝛼 ∑(𝐵𝐻𝑟 |𝑖 − 𝐵𝐻𝑟 |𝑖−1 ) + 𝛥𝑡 2 ∑(𝐵𝐻𝑟 |𝑖 − 𝐵𝐻𝑟 |𝑖−1 ) ∑ 𝐻𝑟
𝜌 𝜕𝑥
𝑘𝑡 𝑘𝑡 𝑘𝑡
𝑘𝑏 𝑘𝑏 2
𝜕 𝛥𝑡
+𝛥𝑥𝛥𝑡 ∑ 𝑞𝐵𝐻𝑟 + 𝛥𝑥𝛥𝑡 2 ∑ 𝑞𝑈𝑥 𝐵𝐻𝑟 + [(𝐵𝜏𝑥𝑧 |ℎ − 𝐵𝜏𝑥𝑧 |𝜂 )𝑖 − (𝐵𝜏𝑥𝑧 |ℎ − 𝐵𝜏𝑥𝑧 |𝜂 )𝑖−1 ]
𝜕𝑥 𝜌
𝑘𝑡 𝑘𝑡
This equation is applied at all cells in the branch generating a system of equations that are solved for the
water surface elevation at the n+1 time level using the Thomas algorithm. The resulting FDA is implicit in
 because of the dependence of  in the momentum equation (see Figure 112).
213
NUMERICAL SOLUTION
The boundary conditions of either flow or head boundary are accounted for in applying the above equa-
tion to the boundary cells as shown in Figure 113 and Figure 114, respectively. The boundary condition
implementation is the same as described in Cole and Buchak (1995).
Time level
n+1
n+1
n Un
n
Figure 112. Solution of the water level, .
Figure 113. Grid for computation of a flow boundary in x. U, the longitudinal velocity, at the
flow boundary or at the right-hand side of inactive segment 1 is computed from the flow rate.
214
NUMERICAL SOLUTION
Figure 114. Grid for computation of a head boundary in x. U, the longitudinal velocity, at the
flow boundary or at the right-hand side of inactive segment 1 is computed from the head dif-
ference between the boundary head (at center of inactive segment 1) and the head at the first
grid cell (segment 2).
Horizontal Momentum Numerical Solution for Horizontal Veloc-

ity U
The x-momentum equation:
t
+
x
+
z
x
−
  x dz

(5-29)
1 B xx 1  B  x z
+ + + qBU x
 x  z
is solved using either a fully explicit or an explicit/implicit finite difference solution technique specified by
the user after the water surface equation is solved.
Explicit Solution
This scheme is based on solving the partial differential terms using an explicit finite difference technique
where:
215
NUMERICAL SOLUTION
 UUB  WUB 
U in +1 Bin +1 = U in Bin + t{ − − + gB sin  + g cos B
x z x
(5-30)
g cos B  1 B xx 1  B  x z
z
−
 

dz
x   x
+
 z
+ qBU x }in
The various terms are put into finite difference form as follows. The longitudinal advection of momentum
is an upwind difference scheme where the order of differencing is dependent on the sign of U, e.g., for U >
0
𝜕𝑈𝑈𝐵 1 𝑛 𝑛 𝑛 𝑛 𝑛 𝑛
≅ [𝐵𝑖,𝑘 𝑈𝑖,𝑘 𝑈𝑖,𝑘 − 𝐵𝑖−1,𝑘 𝑈𝑖−1,𝑘 𝑈𝑖−1,𝑘 ]| (5-31)
𝜕𝑥 𝑖,𝑘 𝛥𝑥𝑖
The vertical advection of momentum is also an upwind scheme based on the velocity of W. For W > 0 or
downward flow
𝜕𝑊𝑈𝐵 1 𝑛 𝑛 𝑛 𝑛 𝑛 𝑛
𝜕𝑧 𝑖,𝑘
≅ 𝛥𝑧 [(𝑊𝑖,𝑘 𝑈𝑖,𝑘 𝐵𝑖,𝑘 ) − (𝑊𝑖,𝑘−1 𝑈𝑖,𝑘−1 𝐵𝑖,𝑘−1 )]| (5-32)
𝑘
The gravity force is:
𝑔𝐵 𝑠𝑖𝑛 𝛼 = 𝑔 𝑠𝑖𝑛 𝛼 𝐵𝑖𝑛 (5-33)
The pressure gradient is:
𝜕𝜂 𝑔 𝑐𝑜𝑠 𝛼𝐵 𝑧 𝜕𝜌 𝑔 𝑐𝑜𝑠 𝛼𝐵𝑖𝑛

𝑔 𝑐𝑜𝑠 𝛼 𝐵 𝜕𝑥 − ∫𝜂 𝜕𝑥 𝑑𝑧 = (𝜂𝑖+1 − 𝜂𝑖 )𝑛
𝜌 𝛥𝑥
𝑔 𝑐𝑜𝑠 𝛼𝐵𝑖𝑛 𝑛
− ∑ (𝜌𝑖+1,𝑘 − 𝜌𝑖,𝑘 ) 𝛥𝑧𝑘 (5-34)
𝜌𝛥𝑥
The horizontal advection of turbulent momentum is:
U
BAx
1 B xx x =  Bi +1 / 2 Ax (U n − U n )
n
=  x x  i +1,k
 x x
i ,k
 i i +1 / 2  (5-35)
 B Ax  n
(U i ,k − U in−1,k )
n
−  i −1 / 2
 
 i i −1 / 2 
x x
The contribution to longitudinal momentum by lateral branch inflows is:
𝑞𝐵𝑈𝑥 = 𝑞𝐵𝑈𝑥 |𝑛𝑖,𝑘 (5-36)
Using the definition of the shear stress:
 U 
 x z =  w +  b + Az
z 
(5-37)

216
NUMERICAL SOLUTION
where Az is the turbulent kinematic viscosity, the vertical transport of momentum is:
1  B x z  B  U   Bi ,k +1/ 2 
n
=  w +  b + Az =
 z  z   z   z k z k +1/ 2  
 n Az i ,k +1/ 2 n 
 w i ,k +1/ 2 +  b i ,k +1/ 2 +
n
z k +1/ 2
(
U i ,k +1 − U in,k  ) (5-38)
 
 Bin,k −1/ 2   n Az i ,k −1/ 2 n 
−    w
 + 
n
+ ( )
U i ,k − U in,k −1 
 z k z k −1/ 2    z k −1/ 2
i , k −1 / 2 b i , k −1 / 2

Implicit Solution
The implicit technique was utilized to reduce the time step limitation for numerical stability when values of
Az were large, as for an estuary or a river system. This occurs because the time step limitation is a function
of Az. Only the vertical transport of momentum term was solved implicitly. All other terms for the solution
of the horizontal momentum equation were the same as the explicit scheme.
The horizontal momentum equation can be separated into the following two equations:
t
+
x
+
z
x
−
  x dz
(5-39)
1 B xx 1  B(  b +  w )
+ + + qBU x
 x  z
𝜕𝑈𝐵 1 𝜕 𝜕𝑈
𝜕𝑡
= 𝜌 𝜕𝑧 (𝐵𝐴𝑧 𝜕𝑧 ) (5-40)
Equation 93 is written as:
 UUB  WUB 
U i* Bin +1 = U in Bin + t{ − − + gB sin  + g cos B
x z x
(5-41)
g cos B  1 B xx 1  B ( b +  w )
z
−
 

x
dz +
 x
+
 z
+ qBU x }in
where U* is the velocity at the new time level before the application of equation 94. Equation 92 is solved
similarly to the solution of the fully explicit technique outlined above.
Equation 94 is then solved using a fully implicit technique as:
𝜕𝑈𝐵 (𝑈𝑖𝑛+1 𝐵𝑖𝑛+1 −𝑈𝑖∗ 𝐵𝑖𝑛+1 ) 1 𝜕 𝜕𝑈 𝑛+1

= = (𝜌𝐵𝑛+1 𝐴𝑧 ) (5-42)
𝜕𝑡 𝛥𝑡 𝜌 𝜕𝑧 𝜕𝑧
This can be rewritten as:
217
NUMERICAL SOLUTION
 tBin,k++11/ 2   Az i ,k +1/ 2 n+1 

U n +1 n +1
B =U B * n +1
+  
 ( )
U i ,k +1 − U in,k+1 
 zk   zk +1/ 2
i i i i

(5-43)
 tBin,k+−11/ 2   Az i ,k −1/ 2 n+1 
−  
 (
U i ,k − U in,k+−11 )
 zk   zk −1/ 2 
Regrouping terms at n+1 time level on the LHS, the equation can be written as
AU in, k+−11 + VU in, k+1 + CU in, k++11 = DU i*, k (5-44)
where:
 − tBin,k+−11/ 2   Az i ,k −1/ 2 
A =  n+1 
 
B 
 i ,k k   zk −1/ 2 
z
 tBin,k++11/ 2   Az i ,k +1/ 2   tBin,k+−11/ 2   Az i ,k −1/ 2 
V = 1 +  n+1 
 z  +  n+1 
 
B  z B 
 i ,k k   k +1/ 2   i ,k k   zk −1/ 2  z
 − tB n+1   A 
C =  n+1i ,k +1/ 2   z i ,k +1 / 2 
 Bi ,k zk   zk +1 / 2 
D =1
The resulting simultaneous equations are solved for Un+1 using the Thomas algorithm.
The resulting horizontal velocities are then corrected to preserve branch continuity (or more correctly to
“force” continuity). The lack of a continuity balance is a result of numerical errors in the solution of the
FDAs for n and Un+1. Cole and Buchak (1995) discuss the algorithm to correct horizontal velocities from
continuity. The correction is calculated segment by segment: the change in the water layers over time
should equal the net flow into the segment by horizontal velocities. The change or correction in velocity is
apportioned equally to the horizontal velocity of all vertical layers. Other numerical models iterate by re-
solving the water surface equation after updated horizontal and vertical velocities are computed. This re-
sults in significant computational time but a more correct solution.
218
NUMERICAL SOLUTION
Solution of Vertical Velocity W

The continuity equation
 UB  WB
+ = qB
x z
is used to solve for W after discretizing into finite differences as
WBi ,k − WBi ,k −1  UBi ,k − UBi −1,k 

= −  + qi ,k Bi ,k
z k  xi 
These variables are illustrated in Figure 115.
Figure 115. Solution of vertical velocity, w, from continuity equation.
The algorithm starts at the bottom cell using corrected horizontal velocities and where WBk=0 such that:
 UBi ,k − UBi −1,k 

WBi ,k −1 = −WBi ,k + z k   − z k qi ,k Bi ,k
 xi 
Numerical Stability
The numerical stability criterion is based on the following equation that only considers momentum equa-
tion stability
219
NUMERICAL SOLUTION
1
t 
 gH
 Ax Az  Q  2
2  + + +
  x 2  z2  V x
t = timestep, sec
Ax = longitudinal eddy viscosity, m2 sec-1
Az = vertical eddy viscosity, m2 sec-1
Q = total flow into or out of a cell, m3 sec-1
V = cell volume, m3
g = gravitational acceleration, m sec-2
H = maximum waterbody depth, m
 = water density, kg m-3
 = surface to bottom water density difference, kg m-3
This criterion does NOT guarantee numerical stability since it is based on a linearized form of the PDEs.
There is no check on stability for the advective-diffusion equation since it is usually not limiting. But the
advective-diffusion equation can become unstable also.
220
NUMERICAL SOLUTION
Advective-Diffusion Equation Numerical So-

lution
Characteristics of the Constituent Numerical Solution
• the advective diffusion equation is solved by splitting the solution between horizontal and vertical
transport
• the horizontal transport is solved explicitly where QUICKEST (a third order numerical scheme, Leon-
ard, 1979) numerical scheme is used for horizontal explicit transport, central differences are used for
diffusion
• vertical transport is solved using an implicit/explicit general time weighted scheme for advection and
a fully implicit scheme for diffusion, simultaneous equations are solved using the Thomas algorithm
Solution of Advective-Diffusion Equation

The advective-diffusion transport equation
     
 BDx   BDz 
 B  UB   WB   x  z
+ + - - = q  B + S B
t x z x z
is solved by splitting the two-dimensional FDA into 2 one-dimensional parts: (1) horizontal transport and
sources/sinks and lateral inflows/outflows and (2) vertical transport. The horizontal part of the solution is
Figure 116. Horizontal advective and diffusive transport.
221
NUMERICAL SOLUTION
n
    
  UB   BDx  
 B  B i ,k − B i ,k  x
* n
  − + + q  B + S B ,
t t  x x 
 
i ,k
and the variables are illustrated in Figure 116. The time level * is the intermediate concentration (or tem-
perature) only considering horizontal fluxes, sources/sinks, and lateral inflows. In this step the diffusion
term is a central spatial difference and the advection term uses the ULTIMATE/QUICKEST differencing tech-
nique (see discussion below). In Version 2, the QUICKEST scheme was introduced which uses quadratic
upstream differencing rather than linear upstream differencing. Cole and Buchak (1995) discuss this differ-
encing algorithm which eliminates much of the numerical diffusion associated with upstream differencing.
The result from this explicit computation is then used to compute the vertical transport (see Figure 117).
The resulting FDA is put in a general time weighted scheme:
Figure 117. Vertical advective and diffusive transport.
n +1
    
B in,+k1 − B *i ,k n +1   BDz  
 B   WB  
n
  WB    z
  −   +  − (1 − )  +
t t   z  i ,k   z  i ,k  z 
 
  i ,k
where  is the degree of implicitness (=0 explicit; =0.5 Crank-Nicholson; =1 fully implicit). Central differ-
ences are used for the spatial derivatives for both the vertical diffusion and vertical advection terms. This
equation is applied to all vertical layers and solved using the Thomas algorithm. The model user can
choose the degree of implicitness for this equation.
222
NUMERICAL SOLUTION
Advective Transport in x and z

The first step in the numerical solution is to define the computational grid (Figure 118). The grid is space-
staggered since some state variables are defined at the segment centers and some at segment interfaces,
displaced by Δx/2 or Δz/2 from the segment center. The grid discretizes a waterbody into computational
cells whose locations are defined by their segment [I] and layer number [K], i.e., cell (K,I). Variables defined
at the boundary include the horizontal and vertical velocities, U and W, longitudinal eddy viscosity and
diffusivity, Ax and Dx, vertical eddy viscosity and diffusivity, Az and Dz, and internal shear stress τx. The
density, ρ, temperature, T, constituent concentration, Ф, pressure, P, and average cell width, B are defined
at the cell center.
Since the constituent concentration is defined at the center and velocities are defined at the boundaries,
spatial averaging of velocities is not required to determine fluxes into or out of the control volume at a fixed
time level. In addition, the horizontal velocity is surrounded by a cell with water surface elevations and
densities defined on either side. Thus, the horizontal velocity is computed from horizontal gradients of the
surface slope and densities without requiring spatial averaging of these variables.
The geometry is specified by a cell width, B, thickness or vertical grid spacing, H or Δz, and length, Δx. Several
additional geometric variables are used in the calculations. These additional variables include the average
cross-sectional area between two cells (k,i) and (k,i+1):
B k,i H k,i + B k,i H k,i+1 (5-45)

BH rk,i =
2
the average widths between two cells (k,i) and (k+1,i):
B k,i + B k+1,i
Bbk,i = (5-46)
2
and the average layer thickness between layers k and k+1:
H k + H k+1 (5-47)
H k,i =
2
The numerical procedure for solving the six unknowns at each timestep is to first compute water surface
elevations. With the new surface elevations, new horizontal velocities can be computed. With new hori-
zontal velocities, the vertical velocities can be found from continuity. Using new horizontal and vertical
velocities, the water surface elevation equation can be solved for η simultaneously for a model branch. The
solution for η is thus spatially implicit at the same time level and eliminates the surface gravity wave speed
criterion:
∆𝑥
∆𝑡 < (5-48)
√𝑔𝐻𝑚𝑎𝑥
that can seriously limit timesteps in deep waterbodies. Then the temperature and constituent concentra-
tions are computed from the energy and constituent mass balance equations.
223
NUMERICAL SOLUTION
segment I-1 segment I segment I+1
layer KT
layer K-1 z
layer K
layer K+1
x
Legend
U, A x , D x , τ x
P, B, ρ, Ф
W, D z
Az
Figure 118. Variable locations in computational grid.
Version 1.0 used upwind differencing in the constituent transport advective terms in which the cell concen-
tration immediately upstream of the velocity was used to calculate mass fluxes. A major problem with
upwind differencing is the introduction of numerical diffusion, which acts like diffusion but is an artifact of
the numerical scheme. The false longitudinal diffusion coefficient (m2/s) added to the model can be com-
puted from
𝑢∆𝑥
𝛼𝑒 = (1 − 𝐶) (5-49)
2
where:
e = numerical diffusion coefficient, m2/s
u = velocity, m/s
x = longitudinal grid spacing, m
𝑢∆𝑡
C = Courant number= ∆𝑥
224
NUMERICAL SOLUTION
A similar condition holds for vertical advection. In many cases, numerical diffusion can overwhelm physical
diffusion producing inaccurate results when strong gradients are present. The problem is particularly pro-
nounced for stratified reservoirs and estuaries.
Numerical diffusion has been reduced by implementing an explicit, third-order accurate QUICKEST horizon-
tal/vertical transport scheme (Leonard, 1979), and time-weighted, implicit vertical advection. Tests of this
scheme are reported in Chapman and Cole (1992).
QUICKEST uses an additional spatial term to estimate concentrations used in computing horizontal and
vertical fluxes. A nonuniform grid QUICKEST scheme was developed using a three-point Lagrangian inter-
polation function to estimate constituent values at grid cell interfaces. Specifically, advective multipliers
for each of three upstream weighted grid cells are derived in terms of cell lengths and the local cell interface
velocity. Time invariant parts of the interpolation functions are calculated once thus minimizing computa-
tions for additional constituents.
Implicit vertical transport including variable layer heights has also been implemented. Vertical diffusion is
fully implicit and advection employs a time-weighted, central difference, implicit scheme. A unique feature
of vertical advection, in the explicit part of the time-weighted scheme, is QUICKEST which increases overall
accuracy.
As implemented in the code, the new transport scheme is a two-part solution for constituent concentra-
tions at the new timestep. First, horizontal advection is computed using QUICKEST and diffusion is com-
puted using central differencing. This part also includes the explicit vertical advection contribution (which
utilizes QUICKEST) and all sources and sinks.
Next, the implicit part of vertical advection and diffusion are included. Diffusion is always fully implicit. The
user can time-weight advection by specifying a value for [THETA] which varies from 0 to 1. For [THETA]
equal to 0, the solution is explicit in time and vertical advection is accounted for in the first part of the
algorithm. For [THETA] equal to 1, the solution is fully implicit in time and vertical advection is accounted
for in this part of the algorithm. A Crank-Nicholson scheme where vertical advection is time-weighted be-
tween the explicit (using QUICKEST) and implicit parts results if [THETA] is set to 0.5 or greater. The fol-
lowing is a description of QUICKEST, the preferred transport scheme.
Non-Uniform Grid QUICKEST Formulation
In one dimension, the conservative control volume advective transport of a constituent Φ integrated over
a timestep is:
t
i = i - ( U r  nr - U l  ln )
n+1 n
(5-50)
x
where:
i = constituent concentration at a grid point, g m-3
r,l = right and left cell face constituent concentrations, g m-3
Ur,l = right and left cell face velocity, m s-1
t = time, s
225
NUMERICAL SOLUTION
The QUICKEST algorithm was originally derived using an upstream weighted quadratic interpolation func-
tion defined over three uniformly spaced grid points. This interpolation function estimates cell face con-
centrations required by the conservative control volume transport scheme. For example, the right cell face
concentration estimate for a flow positive to the right is:
 r = T i-1 i-1 + T i i + T i+1 i+1 (5-51)
where T are advective multipliers which weight the contribution of three adjacent grid point concentra-
tions.
The advective multipliers are obtained by collecting terms associated with each constituent defined by the
QUICKEST advection operator. For a non-uniform grid, a combination of two and three-point Lagrangian
interpolation functions (Henrici, 1964) are used to compute the QUICKEST estimate for the right cell face
concentration centered about cells i and i+1:
 r = P 1 (x) -
U t 
P 2 (x) +  D x  t -
1
 
2 
 x 2 - (U  t )  P 2 (x) (5-52)
2  6 
where:
x = the local right cell face position

Dx = diffusion coefficient
Defining a local coordinate system of three non-uniformly spaced grid cells denoted by xi-1, xi, and xi+1 with
corresponding constituent values, the interpolation functions required in Equation 118 are:
(x- x i ) ( x i+1 - x)
P1 (x) =  i+1 + i (5-53)
( x i+1 - x i ) ( x i+1 - x i )
(x- xi )(x- xi-1 ) (x- xi+1 )(x- xi-1 )

P 2 (x) = i+1 + i
( xi+1 - xi )( xi+1 - xi-1 ) ( xi - xi+1 )( xi - xi-1 )
(5-54)
(x- xi+1 )(x- xi )
+ i-1
( xi-1 - xi+1 )( x x-1 - xi )
Taking the first derivative of P1(x) and the second derivative of P2(x) and substituting into Equation 118, it
is then possible to group terms and obtain the advective multipliers. For example, the Ti+1 multiplier is:
(x - xi ) U  t (x - xi ) + (x - xi-1 )
T i+1 = -
( xi+1 - xi ) 2 ( xi+1 - xi ) ( xi+1 - xi-1 )
(5-55)

2  Dx  t -
1
 
 xi2 - ( U  t )2  
+  
6
( xi+1 - xi ) ( xi+1 - xi-1 )
226 Numerical Solution

Numerical Solution
Similar functions are obtained for Ti and Ti-1 multipliers that complete the formulation for the QUICKEST
algorithm.
From a computational standpoint, most geometric components of the multipliers are time-invariant and
are computed once and stored in arrays. The time-varying part of the multipliers (U, Δt, Dx) are updated
each timestep during computation of the T arrays. However, when the QUICKEST scheme is applied verti-
cally, the spatial part of the multipliers for layers [KT] and [KT]+1 are updated each timestep to accommo-
date the surface elevation fluctuation.
ULTIMATE/QUICKEST Numerical Transport Solution Scheme
In Version 2, the QUICKEST numerical scheme replaced the upwind numerical scheme used in Version 1 for
solving the advective terms in the advection-diffusion equation. Compared to the upwind scheme,
QUICKEST resulted in improved numerical accuracy in simulating sharp fronts since the upwind transport
scheme adds excessive numerical diffusion.
A problem with the QUICKEST scheme is that it can give rise to spurious oscillations at the leading and
trailing edge of a sharp front or gradient. This can occur where there are fresh/salt water interfaces, point
source discharges, or cases of strong temperature stratification. Even though the upwind scheme always
gives physically realistic solutions, it introduces numerical diffusion that artificially reduces sharp gradients.
An improvement was introduced by Leonard (1991) that eliminated spurious oscillations but preserved the
higher-order solution scheme of QUICKEST. This technique is a universal (in the sense that it can be applied
to numerical schemes other than QUICKEST) limiter for maintaining monotonic profiles near a gradient and
is called the ULTIMATE solution scheme.
In order to illustrate the scheme, consider the solution of the unsteady advective equation:
𝜕𝐵𝛷 𝜕𝐵𝑈𝛷
𝜕𝑡
+ 𝜕𝑥
=0 (5-56)
where:
 = concentration, g m-3
B = width, m
U = velocity, m s-1
x = longitudinal coordinate, m
t = time, s
The finite difference scheme for this based on a positive flow (U > 0) is:
1 𝑈𝐵𝛥𝑡 𝑈𝐵𝛥𝑡
𝛷𝑖𝑛+1 = 𝐵𝑛+1 (𝐵𝑖𝑛 𝛷𝑖𝑛 − {[ ] 𝛷
𝛥𝑥 𝑅 𝑅
−[ 𝛥𝑥 𝐿
] 𝛷𝐿 }) (5-57)
𝑖
The value i refers to the center grid point, R is the right-hand face value, and L is the left hand face value
(Figure 119).
227
Figure 119. ULTIMATE schematization for positive flow.
Where (in Figure 119):

Фi-1 = upstream concentration, g m-3
Фi = center concentration, g m-3
Фi+1 = downstream concentration, g m-3
ФL = left face concentration, g m-3
ФR = right face concentration, g m-3
Ur = right face horizontal velocity, m s-1
Figure 120 shows a sketch of variables used for a negative flow. The term 𝑈𝛥𝑡/𝛥𝑥is called the Courant
number. The problem to resolve is how to choose the concentrations at the “face” values since concentra-
tions are defined at the center of a cell. An upwinding scheme would say that the concentration at the left
face is Φi-1 and the concentration at the right face is Φi for positive flow. In order to improve numerical
accuracy, there are other higher-order numerical techniques, such as QUICKEST (Leonard, 1979), to esti-
mate these face values. In CE-QUAL-W2, ΦR and ΦL are initially computed based on the QUICKEST method.
However, if the criteria for a monotonic solution are violated, the values for Φ R and ΦL are revised to assure
a monotonic solution. This is the essence of the ULTIMATE algorithm that eliminates over/undershoots in
the numerical transport scheme.
Figure 120. ULTIMATE schematization for negative flow.
228
Numerical Solution

Фi-1 = downstream concentration, g m-3
Фi = center concentration, g m-3
Фi+1 = upstream concentration, g m-3
ФL = left face concentration, g m-3
ФR = right face concentration, g m-3
UL = left face horizontal velocity, m s-1
To outline the procedure for simple conditions, the velocity, segment spacing and segment widths are as-
sumed constant, so that Equation 123 can be written as:
𝑈𝛥𝑡 𝑈𝛥𝑡
𝛷𝑖𝑛+1 = (𝛷𝑖𝑛 − {[ ] 𝛷𝑅 − [ ] 𝛷𝐿 }) (5-58)
𝛥𝑥 𝛥𝑥
Defining normalized variables based on:

𝑛
̃ = 𝛷−𝛷
𝛷 𝑖−1
, (5-59)
𝛷 𝑛 −𝛷𝑛
𝑖+1 𝑖−1
for location i (the center location) and the right face value for positive flow as in Figure 119, we have the
following:
𝑛 𝑛
̃𝑖𝑛 = 𝛷𝑛𝑖 −𝛷𝑖−1
𝛷 (5-60)
𝛷 −𝛷 𝑛 𝑖+1 𝑖−1
𝑛
𝛷𝑅 −𝛷𝑖−1
̃𝑅 =
𝛷 𝑛 𝑛 (5-61)
𝛷𝑖+1 −𝛷𝑖−1
̃ 𝑖𝑛 ~n
for 0   i  1
𝛷
̃𝑅 ≤
If 𝛷 𝑈𝛥𝑡 (5-62)
[ ⁄𝛥𝑥 ]
𝑅
̃𝑖𝑛 ≤ 𝛷
𝛷 ̃𝑖𝑛 ≤ 1
̃𝑅 ≤ 1 for 0 < 𝛷 (5-63)
̃𝑖𝑛 = 𝛷
𝛷 ̃𝑖𝑛 < 0 or 𝛷
̃𝑅 for 𝛷 ̃𝑖𝑛 > 1 (5-64)
The face value is unadjusted from that computed by the numerical scheme and the QUICKEST value of R
is used without alteration. These conditions are shown as the shaded region in Figure 121.
If these conditions are not met, then R is adjusted to force a monotonic solution. The value of the face is
~
replaced with the nearest allowable value of  R based on the above criteria that will ensure these criteria
are met. The face value is determined by using:
𝑛
𝛷𝑅 = 𝛷𝑖−1 ̃𝑅 (𝛷𝑖+1
+𝛷 𝑛 𝑛
− 𝛷𝑖−1 ) (5-65)
229
This procedure is applied to all the faces and then Equation 124 is solved to update the concentration at
the next time level. This means that the right face concentration will be the left face concentration for the
next segment thus ensuring mass conservation.
~ ~ ~
R  in   R  1
~
~  in
R 

Ut
x R

~ ~
 in =  R
~
 in
0
0 1
Figure 121. Definition sketch for monotonic solution domain.
According to tests performed by Lin and Falconer (1997), the QUICKEST-ULTIMATE scheme with splitting of
the diffusion and source/sink terms conserved mass and eliminated numerical oscillations. Leonard (1991)
also indicated that the QUICKEST scheme coupled with the ULTIMATE scheme was numerically accurate
and cost-effective in terms of computational time.
Figure 122 shows the results of a square pulse of 100 g m-3 moving downstream using the UPWIND,
QUICKEST, and ULTIMATE/QUICKEST numerical transport schemes. The UPWIND scheme has a large
amount of numerical diffusion whereas the QUICKEST scheme has non-physical oscillations about the lead-
ing and trailing edge of the solution. The ULTIMATE/QUICKEST numerical solution greatly reduces numerical
diffusion and eliminates the over and undershoots.
For any case of unequal grid spacing, the order of accuracy diminishes. Leonard (1991) recommends that
if used with non-uniform grid spacing, the formal accuracy of the method is only preserved if the grid does
not expand or contract more than about 125%.
Even though the ULTIMATE scheme can be used with any suitable numerical technique, Leonard (1991)
indicates that when coupled with the 3rd order QUICKEST scheme and a 2nd order central difference diffu-
sion operator, the results are virtually indistinguishable from other higher-order advection solvers.
230
Numerical Solution
Square wave pulse moving downstream

DLX=1750 m, U=0.091 m/s, DT(approx)=100 s
110
100
90
ULTIMATE-QUICKEST
QUICKEST
80
UPWIND
Analytical solution
70
Concentration, mg/l
60
50
40
30
20
10
-10
0 10 20 30 40
Model segment number
Figure 122. Comparison of UPWIND, QUICKEST, and ULTIMATE/QUICKEST schemes for conservative
tracer transport.
The following shows a straightforward procedure by Leonard to compute the right and left face values if
they do not meet the monotonic criteria, as well as a procedure for variable velocities.
Based on the sign of the velocity for each face, compute DEL:
𝑛 𝑛
𝐷𝐸𝐿 = 𝛷𝑖+1 − 𝛷𝑖−1 (5-66)
If |𝐷𝐸𝐿| < 10−5 set 𝛷𝑓𝑎𝑐𝑒 = 𝛷𝑖𝑛 and proceed to the next face (the face value is the L or R
face). Otherwise, compute:
𝑛 𝑛 𝑛 𝑛
̃𝑖𝑛 = 𝛷𝑖𝑛 −𝛷𝑖−1𝑛 = 𝛷𝑖 −𝛷𝑖−1
𝛷 (5-67)
𝛷 −𝛷
𝑖+1 𝑖 𝐷𝐸𝐿
̃𝑖𝑛 < 0 or 𝛷
If 𝛷 ̃𝑖𝑛 > 1, set 𝛷𝑓𝑎𝑐𝑒 = 𝛷𝑖𝑛 and proceed to the next face. If not, compute:
𝑛 𝑛
̃𝑓𝑎𝑐𝑒 = 𝛷𝑓𝑎𝑐𝑒
𝛷 𝑛
−𝛷𝑖−1
𝑛
𝛷 −𝛷
= 𝑓𝑎𝑐𝑒 𝑖−1 (5-68)
𝛷𝑖+1 −𝛷𝑖−1 𝐷𝐸𝐿
231
where 𝛷𝑓𝑎𝑐𝑒 is computed based on the user’s chosen numerical scheme (such as QUICKEST or
another scheme).
̃𝑖𝑛 , 𝛷
̃𝑓𝑎𝑐𝑒 < 𝛷
1. If 𝛷 ̃𝑖𝑛
̃𝑓𝑎𝑐𝑒 = 𝛷
̃ 𝑖𝑛
𝛷 ̃ 𝑖𝑛
𝛷
̃𝑓𝑎𝑐𝑒 >
2. if 𝛷 𝑈𝛥𝑡
̃𝑓𝑎𝑐𝑒 =
,𝛷 𝑈𝛥𝑡
( ) ( )
𝛥𝑥 𝛥𝑥
̃𝑓𝑎𝑐𝑒 > 1, 𝛷
3. if 𝛷 ̃𝑓𝑎𝑐𝑒 = 1
̃𝑓𝑎𝑐𝑒 𝐷𝐸𝐿 + 𝛷𝑖𝑛 .

Then recompute the face value according to 𝛷𝑓𝑎𝑐𝑒 = 𝛷
Once all the face values are determined, use the finite difference form of the solution to de-
termine the solution (Equation 123). Leonard (1991) also showed a numerical technique that
minimized the computational burden when using ULTIMATE with the QUICKEST scheme. This
technique is described as follows.
Based on the sign of the velocity for each face, define C, D, and U nodes (center, downstream,
and upstream) corresponding for positive flow to i, i+1, and i-1, respectively.
𝑛
Compute DEL and CURV, where 𝐷𝐸𝐿 = 𝛷𝑖+1 − 𝛷𝑖𝑛 and 𝐶𝑈𝑅𝑉 = 𝛷𝑖+1
𝑛
− 2𝛷𝑖𝑛 + 𝛷𝑖−1
𝑛
1. If |𝐶𝑈𝑅𝑉| ≤ 0.6|𝐷𝐸𝐿|, then use the QUICKEST computed value for the face value, 𝛷𝑓𝑎𝑐𝑒
(the face value is the R or L face as shown in Figure 123).
2. If not, and if |𝐶𝑈𝑅𝑉| ≥ |𝐷𝐸𝐿|, then set 𝛷𝑓𝑎𝑐𝑒 = 𝛷𝑖𝑛
3. If not, and DEL>0, limit 𝛷𝑓𝑎𝑐𝑒 by 𝛷𝑖𝑛 below, or the smaller of
𝑛 (𝛷𝑖𝑛 −𝛷𝑖−1
𝑛
) 𝑛
𝛷𝑟𝑒𝑓𝑒𝑟𝑒𝑛𝑐𝑒 = 𝛷𝑖−1 + 𝑈𝛥𝑡 and 𝛷𝑖+1
( )
𝛥𝑥
(𝛷𝑖𝑛 −𝛷𝑖−1
𝑛
)
4. If not, and DEL<0, limit 𝛷𝑓𝑎𝑐𝑒 by 𝛷𝑖𝑛 , or the larger of 𝛷𝑟𝑒𝑓𝑒𝑟𝑒𝑛𝑐𝑒 = 𝛷𝑖−1
𝑛
+ 𝑈𝛥𝑡 and
( )
𝛥𝑥
𝑛
𝛷𝑖+1
5. Once all the face values are determined, use Equation 123 to obtain the solution.
The procedure outlined above is based on unidirectional and uniform magnitude velocity and segment
widths. If the velocity regime and widths are variable, then the following procedure is followed in order to
maintain a monotonic solution. Consider the following limitations based on definitions shown in Figure
123.
Figure 123. Definition sketch for variable velocity.
232
Numerical Solution

Фi-1 = upstream concentration
Фi = center concentration
Фi+1 = center concentration
ФL = left face concentration at i
ФR = right face concentration at i
ФRR = right face concentration at I+1
Assuming that UR is positive and the concentration times width (𝛷𝐵) increases monotonically, i.e.,
(𝐵𝛷)𝑛𝑖−1 < (𝐵𝛷)𝑛𝑖 < (𝐵𝛷)𝑛𝑖+1 < (𝐵𝛷)𝑛𝑖+2 , between segment i-1 and i+2, then the goal of the technique is
to estimate 𝛷𝑅 , or really 𝐵𝑅 𝛷𝑅 , and to update the concentration using Equation 123. Equation 123 can be
written as a condition for  R as:
𝑈𝐵𝛥𝑡 𝑈𝐵𝛥𝑡
[ ] 𝛷
𝛥𝑥 𝑅 𝑅
= 𝐵𝑖𝑛 𝛷𝑖𝑛 − 𝛷𝑖𝑛+1 𝐵𝑖𝑛+1 + [ 𝛥𝑥 𝐿
] 𝛷𝐿 (5-69)
If it is assumed that 𝐵𝑖𝑛 𝛷𝑖𝑛 ≤ 𝐵𝑅 𝛷𝑅 ≤ 𝐵𝑖+1𝑛 𝑛

𝛷𝑖+1 , then this is equivalent to saying that 𝐵𝑖𝑛 𝑈𝑖𝑛 𝛷𝑖𝑛 ≤
𝑛 𝑛 𝑛 𝑛 𝑛
𝐵𝑅 𝑈𝑅 𝛷𝑅 ≤ 𝐵𝑖+1 𝑈𝑖+1 𝛷𝑖+1 or that the net flux into a segment is increasing.
Using the monotonic limitation that 𝐵𝑖𝑛+1 𝛷𝑖𝑛+1 ≥ 𝐵𝑖−1

𝑛 𝑛
𝛷𝑖−1 then Equation 135 becomes:
𝑈𝐵𝛥𝑡 𝑈𝐵𝛥𝑡
[ ] 𝛷𝑅 ≤ 𝐵𝑖𝑛 𝛷𝑖𝑛 − 𝛷𝑖−1
𝑛 𝑛
𝐵𝑖−1 +[ ] 𝛷𝐿 (5-70)
𝛥𝑥 𝑅 𝛥𝑥 𝐿
𝑛 𝑛
Then making a conservative assumption that 𝐵𝐿 𝛷𝐿 ≥ 𝐵𝑖−1 𝛷𝑖−1 , Equation 136 becomes:
𝑈𝐵𝛥𝑡 𝑈𝐵𝛥𝑡 𝑛
[ ] 𝛷𝑅 ≤ 𝐵𝑖𝑛 𝛷𝑖𝑛 − 𝛷𝑖−1
𝑛 𝑛
𝐵𝑖−1 +[ ] 𝑛
𝛷𝑖−1 (5-71)
𝛥𝑥 𝑅 𝛥𝑥 𝑖−1
Another condition is also imposed on 𝐵𝑅 𝛷𝑅 by looking at the control volume segment centered at i+1.
Using similar reasoning as above and assuming that URR > 0, the other criterion for 𝐵𝑅 𝛷𝑅 is
𝑈𝐵𝛥𝑡 𝑛 𝑛 𝑛 𝑛 𝑈𝐵𝛥𝑡 𝑛
[ ] 𝛷𝑅 ≤ 𝐵𝑖+2 𝛷𝑖+2 − 𝛷𝑖+1 𝐵𝑖+1 +[ ] 𝛷𝑖+1 (5-72)
𝛥𝑥 𝑅 𝛥𝑥 𝑅𝑅
These criteria would be altered appropriately if the function were monotonically decreasing rather than
increasing.
Vertical Implicit Transport
Focusing on vertical advective and diffusive transport, constituent transport can be written as:
 B   WB     
+ -  BD z  = RHS (5-73)
t z z  z 
where RHS represents horizontal transport and all sources/sinks. Integrating the transport equation verti-
cally and over time gives:
  n+1 
BH  n+1 + H  t  z (WB  n+1) - H t  z  BD z   = BH * (5-74)
 z 
233
where:
* = all n-time level horizontal and explicit vertical transport and sources/sinks
= time-weighting for vertical advection; 0 for fully explicit, 0.55 for Crank-Nicholson, and 1 for fully
implicit
Expanding the differential operators in terms of central differences and collecting terms, the above equa-
tion can be recast as:
At i i-1 +V t i i + C t i i+1 = Dt i
n+1 n+1 n+1
(5-75)
where:
 t   W k,i Bbk,i - W k-1,i Bbk-1,i  B D B D 
Vt i=1 +    + bk,i z k,i + bk-1 z k-1,i 
BH k,i   2  Hk H k-1 
 t Bbk,i  W k,i D zk,i 

At i =  - 
BH k,i  2 Hk 
 t Bbk-1,i  W k-1,i D zk-1,i 

Ct i = -  + 
BH k,i  2 H k-1 
The coefficients are computed once, stored in arrays, and used to update each constituent. This is accom-
plished by loading the explicit part of the solution, Φ*, with each successive constituent and inverting the
resulting matrix via a Thomas tridiagonal solver.
234
6. References
Abbot, M. B. 1982. “A model system for design and operation of storm-sewer networks,” Engineering ap-
plication of computational hydraulics, M. B. Abbot and J. A. Cunge, eds., Vol. 1, Homage to Alexandra
Preissmann, Pitman Advanced Publishing Program, Boston, 11-36.
Adams, E. E., Harleman, D.R.F., Jirka, G., and Stolzenback, K. D. 1981. “Heat Disposal in the Water Environ-
ment”, Ralph M. Parsons Laboratory for Water Resources and Hydrodynamics, Department of Civil En-
gineering, MIT, Cambridge, MA.
Ambrose, R. B.; Wool, T.; Connolly, J. P.; and Schanz, R. W. 1988. “WASP4, A Hydrodynamic and Water
Quality Model: Model Theory, User’s Manual, and Programmer’s Guide,” Environmental Research La-
boratory, Environmental Protection Agency, EPA 600/3-87/039, Athens, Ga.
Anderson, D. A., Tannehill, J. C., Pletcher, R. H. 1984. Computational Fluid Mechanics and Heat Transfer,
Hemisphere, New York.
Annear, R. and Wells, S. A. 2007. A comparison of five models for estimating clear-sky solar radiation, Water
Resources Research, 43, W10415, doi:10.1029/2006WR005055.
APHA 1985. Standard Methods for the Examination of Water and Wastewater, APHA, Washington, DC.
Ashton, G.D. 1979. "Suppression of River Ice by Thermal Effluents", CRREL Rpt. 79-30, US Army Engineer
Cold Regions Research and Engineering Laboratory, Hanover, NH.
Banks, R.B. 1975. “Some Features of Wind Action on Shallow Lakes,” ASCE, J. Env. Engr. Div., 101(EE3), pp.
489-504.
Banks, R. B. and Herrera, F. F. 1977. Effect of Wind and Rain on Surface Reaeration. J. Envir. Engr. Div. ASCE,
101(EE5):813-827.
Barkau, R. 1997. "UNET One-Dimensional Unsteady Flow Through a Full Network of Open Channels, User's
Manual, US Army Corps of Engineers, Hydrologic Engineering Center, Davis, Ca.
Barrett, M.J.; Gameson, A.L.; and Ogden, C.G. 1960. Aeration Studies of Four Weir Systems, Water and
Water Engineering, London.
Batchelor, G. K. 1967. An Introduction to Fluid Dynamics, Cambridge University Press, NY.
Berger, C. J. 2000. “Modeling Macrophytes of the Columbia Slough,” Ph.D. Dissertation, Environmental Sci-
ences and Resources, Portland State University, Portland, OR.
Berger, C. J. and Wells, S. A. 1999. “Hydraulic and Water Quality Modeling of the Columbia Slough, Volume
1: Model Description, Geometry, and Forcing Data”, Technical Report EWR-2-99, Department of Civil
Engineering, Portland State University, 109 pp.
235 Water Quality

References
Berger, C. and Wells, S. A. 2008. “A Macrophyte Water Quality and Hydrodynamic Model,” ASCE, Journal of
Environmental Engineering, Volume 134, Issue 9, pp. 778-788.
Berger, C. and Wells, S. A. 2014. “Updating the CEMA Oil Sands Pit Lake Model,” Report to Cumulative
Environmental Management Association, Alberta, Canada, August 2014.
Brady, D.K., Graves, W.L., and Geyer, J.C. 1969. "Surface Heat Exchange at Power Plant Cooling Lakes",
Cooling Water 2, Discharge Project No. 5, Publication No. 69-901, Edison Electric Institute, New York, NY.
Broecker, H. C., Petermann, J., and Siems, W. 1978. The Influence of Wind on CO2 Exchange in a Wind-
Wave Tunnel, J. Marine Res., 36(4):595-610.
Butts, T.A. and Evans, R.L. 1983. "Small Stream Channel Dam Aeration Characteristics," ASCE J. Env. Engr.,
Vol 109, No 3, pp. 555-573, June 1983.
Carr, G. M., H. C. Duthie and W. D. Taylor (1997). “Models of aquatic plant productivity: a review of the
factors that influence growth,” Aquatic Botany, 59, 195-215.
Celik I., Rodi W. 1984. “Simulation of Free-Surface Effects in Turbulent Channel Flow”, PhysicoChemical
Hydrodynamics, Vol. 5, No 3 / 4, 1984, pp 217-227.
Celik I., Rodi W. 1988. “Modeling Suspended Sediment Transport in Nonequilibrium Situations”, Journal of
Hydraulic Engineering, Vol. 114, No. 10, October, 1988, pp 1157-1191.
Cao, Z., Pender, G., and Meng, J. 2006. Explicit Formulation of the Shields Diagram for Incipient Motion of
Sediment, ASCE, Journal of Hydraulic Engineering, October, 1097-1099,
10.1061/(ASCE)07339429(2006)132:10(1097).
Chapman, R.S. and T.M. Cole. 1992. "Improved Thermal Predictions in CE-QUAL-W2", in Hydraulic Engineer-
ing Saving a Threatened Resource - In Search of Solutions Proceedings of the Hydraulic Engineering ses-
sions at Water Forum '92, M. Jennings and N. G. Bhowmik, eds., August 2-6, 1992, Baltimore, MD. Amer-
ican Society of Civil Engineers. New York, NY.
Chapman, R. S., Gerald, T. K., and Dortch, M. S. 1994. “New York Bight Study. Report 3, Three-Dimensional
Particle Tracking Model for Floatables and Dissolved and Suspended Materials.” Technical Report CERC-
94-4, U.S. Army Engineer Waterways Experiment Station, Vicksburg, MS.
Chapra, S. 1997. Surface Water Quality Modeling, McGraw-Hill, NY.
Chapra, S.C., and Reckhow, K.W. 1983. Engineering Approaches for Lake Management, Vol 12: Mechanistic
Modeling, Butterworth Publishers, Boston, MA.
Chu, C. and Jirka, G. (2003) “Wind and Stream Flow Induced Reaeration,” Journal of Environmental Engi-
neering, ASCE, Vol 129, No 12, 1129-1136.
Churchill, M.A., Elmore, H. L., and Buckingham, R.A. 1962. "Prediction of Stream Reaeration Rates," J. San.
Engr. Div., ASCE, SA4:1, Proc. Paper 3199.
Clay, C.H. 1995. Design of Fishway and Other Fish Facilities, Lewis Publishers, Ann Arbor.
References 236
References
Cole, T. and Buchak, E. 1995. CE-QUAL-W2: A Two-Dimensional, Laterally Averaged, Hydrodynamic and
Water Quality Model, Version 2.0, Technical Report El-95-1, U.S. Army Engineer Waterways Experiment
Station, Vicksburg, MS.
Columbia Basin Research. 2000. “Columbia River Salmon Passage Model, CRISP 1.6, Theory and Calibra-
tion”, Technical Report, School of Aquatic and Fisheries Sciences, University of Washington, 238 pp.
Cotton, M., Reedha, D., Stansby, P. 2005. “Low-Reynolds Number two-equation turbulence modeling for
open channel flow: development and evaluation of free surface boundary conditions on the dissipation
rate equation”, Journal of Hydraulic Research, Vol 43, No 6, pp. 632-643.
Covar, A. P. 1976. "Selecting the Proper Reaeration Coefficient for Use in Water Quality Models," presented
at the US EPA Conference on Environmental Simulation and Modeling, April 19-22, Cincinnati, OH.
Cushman-Roisin, B. 1994. Introduction to Geophysical Fluid Dynamics, Prentice-Hall, Englewood Cliffs, NJ.
DiLaura, D.L. 1984. “IES Calculation Procedures Committee Recommended practice for the calculation of
daylight availability.” Journal of the Illuminating Engineering Society of North America, 13 (4) p. 381-392.
DiToro, D.M. 2001. Sediment Flux Modeling. Wiley-Interscience. New York. 656pp.
Downing, A.L., and G.A. Truesdale. 1955. Some factors affecting the rates of solution of oxygen in water. J.
Applied Chemistry, Vol. 5, pp.570-581.
Dunsbergen, D. W. and Stalling, G. S. 1993. The combination of a random walk method and a hydrodynamic
model for the simulation of dispersion of dissolved matter in water, Transactions on Ecology and the
Environment Vol 2, WIT Press, ISSN 1743-3541.
Edinger, J.E., Brady, D.K., and Geyer, J.C. 1974. "Heat Exchange and Transport in the Environment", Rpt.
No. 14, EPRI Publication No. 74-049-00-34, prepared for Electric Power Research Institute, Cooling Water
Discharge Research Project RP-49), Palo Alto, CA.
Edinger, J. E. 2002. Waterbody Hydrodynamic and Water Quality Modeling, ASCE Press, Reston, Virginia.
Elliot, W. P. 1958. “The growth of the atmospheric internal boundary layer”, EOS, Transactions of the Amer-
ican Geophysical Union, Vol 39, Issue 6, 1048-1054.
Engelund, F. 1978. Effect of Lateral Wind on Uniform Channel Flow. Progress Report 45, Inst. of Hydrody-
namic and Hydraulic Engr., Tech. Univ. of Denmark.
Environmental Laboratory. 1985. "CE-QUAL-R1: A Numerical One-Dimensional Model of Reservoir Water

Quality; User's Manual", Instruction Rpt. E-82-1, US Army Corps of Engineers, Waterways Experiment
Station, Vicksburg, MS.
Environmental Laboratory. 1995. “CE-QUAL-RIV1: A Dynamic, One-Dimensional Longitudinal. Water Quality

Model for Streams: User’s Manual,” Instruction Report EL-95-2, USACE Waterways Experiments Station,
Vicksburg, MS.
237
References
EPA. 1971. “Effect of Geographical Location on Cooling Pond Requirements and Performance,” in Water
Pollution Control Research Series, Report No. 16130 FDQ, Environmental Protection Agency, Water
Quality Office, Washington, District of Columbia, 160 pp.
EPA 1985. Rates, Constants and Kinetics in Surface Water Quality Modeling, Environmental Research La-
boratory, EPA/600/3-85/040, Athens, Ga.
ERM and Golder. 2011. CEMA Oil Sands Pit Lake Model. Technical Report 09-1336-1008, submitted to
CEMA, Canada, May 2011, 86 pp.
Fang, X. and Stefan, H. G. 1994. “Modeling Dissolved Oxygen Stratification Dynamics in Minnesota Lakes
under Different Climate Scenarios,” Project Report 339, St. Anthony Falls Hydraulic Laboratory, Univer-
sity of Minnesota, Minneapolis. [update]
Ferziger J., Peric, M. 2002. Computational Methods for Fluid Dynamics, Springer Verlag.
French, R. H. 1985. Open-Channel Hydraulics, McGraw-Hill, New York.
Ford, D.E., and Johnson, M.C. 1983. "An Assessment of Reservoir Density Currents and Inflow Processes",
Technical Rpt. E-83-7, US Army Engineer Waterways Experiment Station, Vicksburg, MS.
Ford, D. E. and Stefan, H. G. 1980. “Thermal predictions using integral energy model”, J. Hydraul. Div., Am.
Soc. Civ. Eng. 106: 39–55.
Gameson, A.L.H. and Gould, D.J., 1975. In Proc. Int. Symp. On Discharge of Sewage from Sea Outfalls, Per-
gamon Press, London.
Gelda, R. K., Auer, M. T., Effler, S. W., Chapra, S. C., and Storey, M. L. 1996. “Determination of Reaeration
Coefficients: A Whole Lake Approach”, ASCE J. Envir. Engr., Vol 122, Issue 4, 269.
Gill, A.E. 1982. "Appendix 3, Properties of Seawater", Atmosphere-Ocean Dynamics, Academic Press, New
York, NY, pp 599-600.
Giese, A.C. 1968. Cell Physiology, 3d ed., W.B. Saunders Co., Philadelphia, PA.
Gould, S. 2006. “k- Turbulence Model in CE-QUAL-W2”, Research Project Report, Department of Civil and
Environmental Engineering, Portland State University, Portland, OR, 53 pp.
Hamblin and Salmon (1975), "the vertical diffusion of momentum is probably the most important internal
parameter" f
Hargrave, B.T. 1972. "Oxidation-Reduction Potentials, Oxygen Concentration and Oxygen Uptake of Pro-
fundal Sediments in a Eutrophic Lake", Oikos, Vol 23, pp 167-177.
Hathhorn, W. E. 1997. “Simplified Approach to Particle Tracking Methods for Contaminant Transport”,
ASCE, J. Hydraul. Eng., 123(12):1157-1160.
HEC 1997a. "UNET One-Dimensional Unsteady-Flow Through a Full Network of Open Channels, User's Man-
ual," US Army Corps of Engineers, Hydrologic Engineering Center, Davis, CA.
References 238
References
HEC 1997b. "HEC-RAS River Analysis System Hydraulic Reference Manual Version 2.0," US Army Corps of
Engineers, Hydrologic Engineering Center, Davis, CA.
Henderson, H. 1966. Open Channel Flow, The MacMillan Company, NY.
Henrici, P. 1964. Elements of Numerical Analysis, John Wiley & Sons, Inc., New York.
Henry, Michelle. 2008. “Wind-Wave Effects and an Improved Fetch Algorithm For CE-QUAL-W2.” Honors
thesis submitted in partial fulfillment of the requirements for the degree of Bachelor’s of Science, Civil
Engineering, Portland State University, Portland, OR.
Hoerner, S. F. (1965). Fluid-Dynamic Drag. Hoerner Fluid Dynamics.
Hsi, G. and J. H. Nath (1968). “Wind drag within simulated forest canopy field.” Technical Report prepared
for U. S. Army Materials Command, Grant No. DA-AMC-28-043-65-G-20.
Kang, S. W., Sheng, Y. P., and Lick, W. 1982. “Wave Action and Bottom Shear Stresses in Lake Erie”, Journal
Great Lakes Research, 8(3):482-494.
Kanwisher, J. 1963. "On the Exchange of Gases Between the Atmosphere and the Sea", Deep Sea Research
with Oceanography, Vol 10, pp 195-207.
Kraus, E., 1972. Atmosphere-Ocean interaction, Oxford University Press, London, 275 pp, 275 pp.
Krishnappan, B., Lau, Y. 1986. “Turbulence Modeling of Flood Plain Flow”, Journal of Hydraulic Engineering,
Vol. 112, No. 4, April, pp 251-266.
Lai (1986) correction to the x-momentum equation
Langbien, W. B. and Durum, W. H. 1967. "The Aeration Capacity of Streams," USGS, Washington, D.C. Circ.
542.
Leonard, B. P. 1979. "A Stable and Accurate Convective Modelling Procedure Based on Upstream Interpo-
lation", Computer Methods in Applied Mechanics and Engineering, Vol 19, pp. 59-98.
Leonard, B. P. 1991. “The ULTIMATE conservative difference scheme applied to unsteady one-dimensional
advection,” Computer Methods in Applied Mechanics and Engr, 88, 1991), 17-74.
Limerinos, J.T. 1970. “Determination of the Manning coefficient from measured bed roughness in natural
channels”, U.S. Geological Survey Water-Supply Paper 1898-B, 47 pp.
Lin, B., and Falconer, R. A. 1997. “Tidal Flow and Transport Modeling Using ULTIMATE QUICKEST Scheme”,
J. of Hydraulic Engineering, ASCE, 123(4), 303-314.
Liss, P.S. 1973. “Processes of gas exchange across an air-water interface,” Deep Sea Research, Vol. 20, pp
221-238.
Jarrett, R. D., 1984. “Hydraulics of high-gradient streams”, American Society of Civil Engineers, Journal of
Hydraulic Engineering, HY11, v. 110, p. 1519-1539.
239
References
Jarrett, R. D. 1985. “Determination of Roughness Coefficients for Streams in Colorado”, USGS Water Re-
sources Investigations Report 85-4004, Lakewood, CO, 60 pp.
Martin, J. L., and McCutcheon, S. C. 1999. Hydrodynamics and Transport for Water Quality Modeling, CRC
Press, Lewis Publishers.
Mastropietro, M. A. 1968. "Effects of Dam Reaeration on Waste Assimilation Capacities of the Mohawk
River," Proceedings of the 23rd Industrial Waste Conference, Purdue University.
Maykut, G. N. and N. Untersteiner. 1971. "Some results from a time dependent, thermodynamic model of
sea ice", J. Geophys. Res., 83: 1550-1575.
Melching, C. and Flores, H. 1999. “Reaeration Equations Derived from USGS Database,” J. Envir. Engr., ASCE,
125(5), 407-414.
Mills, W.B.; Porcella, D.; Ungs, M; Gherini, S.; Summers, K.; Lingfung, M.; Rupp, G.; Bowie, G. Haith, D. 1985.
"A Screening Procedure for Toxic and Conventional Pollutants in Surface and Ground Water,"
EPA/600/6-85/002a, Environmental Research Laboratory, Athens, GA.
Mitchell, R., and Chamberlain, C. 1978. "Factors affecting the survival of indicator organisms", Indicators of
Enteric Contamination in Natural Waters, G. Berg ( ed.), Ann Arbor Science Publ., Inc., Ann Arbor, MI,
pp 15-38.
Moog, D.B. and Jirka, G.H. 1998. "Analysis of reaeration equations using multiplicative error," J. Envin Engr,
ASCE, 124(2), 104-110.
Mortimer, C.H. 1981. "The Oxygen Content of Air Saturated Fresh Waters over Ranges of Temperature and
Atmospheric Pressure of Limnological Interest", International Vereinigung Theoretische and An-
gewandte Limnologie, Vol 22, pp 2-23.
Munk, W.H., Anderson, E.R. 1948. “Notes on a theory of the thermocline”, J. Mar. Res., 7, 276–295.
Nakagawa H., Nezu I., Ueda H. 1975. “Turbulence of Open Channel Flow Over Smooth and Rough Beds”,
Proceedings of the Japan Society of Civil Engineers, No 241 September, pp. 155-168.
Nautical Almanac. 2001. The Nautical Almanac 2001. TheNatuticalAlmanac.com.
Nielson, P. (1992) Coastal bottom boundary layers and sediment transport, Volume 4, Advanced Series on
Ocean Engineering, World Scientific Publishing Co. Ltd., River Edge, NJ.
O'Connor, D. J. and Dobbins, W.E. 1958. "Mechanism of Reaeration in Natural Streams," ASCE Trans.,
86(SA3):35-55.
O’Connor, D.J. 1983. “Wind Effects on Gas-Liquid Transfer Coefficients” J. Envir Engr ASCE, Vol. 109, pp.
731-752.
Owens, M., Edwards, R., and Gibbs, J. 1964. "Some Reaeration Studies in Streams," Int. J. Air Water Poll.,
8:469-486.
Patterson, J.C. and Hamblin, P.F., "Thermal Simulation of a Lake with Winter Ice Cover", Limnology and
Oceanography, 33(3), 1988, P. 323-338.
References 240
References
Pauer, James J. and Martin T. Auer. 2000. “Nitrification in the water column and sediment of a hyper-
eutrophic lake and adjoining river system,” Water Resources, Volume 34, No. 4, pp. 1247-1254.
Petryk, S. (1969), “Drag on Cylinders in Open Channel Flow,” PhD. Thesis, Colorado State University, Fort
Collins, Colorado.
Petryk, S. and G. Bosmajian III (1975). “Analysis of flow through vegetation,” Journal of the Hydraulics Divi-
sion, Proceedings of the ASCE, 101(HY7), 871-884.
Prakash, S., Vandenberg, J. A., E. M. Buchak. 2015. Sediment Diagenesis Module for CE-QUAL-W2. Part 2:
Numerical Formulation. Environmental Modeling & Assessment. Print ISSN 1420-2026. Online ISSN
1573-2967. http://dx.doi.org/10.1007/s10666-015-9459-1. Springer International Publishing. April.
Press W., Teukolsky S., Vetterling W., Flannery B. 1996. Numerical Recipes in Fortran: The Art of Scientific
Computing, Cambridge University Press.
Rodi, W. 1993. Turbulence Models and Their Application in Hydraulics, 3rd edition, IAHR, A.A. Balkema,
Rotterdam.
Roesner, L. A., J. A. Aldrich, and R. E. Dickinson. 1988. “Storm water management model user’s manual
version 4: EXTRAN addendum 1 EXTRAN,” Cooperative Agreement CR-811607, U.S. Environmental Pro-
tection Agency, Cincinnati, Ohio.
Ryan, Patrick J. and Keith D. Stolzenbach. 1972. Chapter 1: "Environmental Heat Transfer" in Engineering
Aspects of Heat Disposal from Power Generation, D. R. F. Harleman, ed., R. M. Parson Laboratory for
Water Resources and Hydrodynamics, Department of Civil Engineering, Massachusetts Institute of Tech-
nology, Cambridge, MA.
Ryan, P.J. and D.R.F. Harleman. 1974. "Surface Heat Losses from Cooling Ponds," Water Resources Research,
Vol 10, No 5, Oct 1974, pp. 930-938.
Sabersky, R.; A. Acosta, E. Haupmann. 1989. Fluid Flow A First Course in Fluid Mechanics, Macmillan Pub-
lishing Co., NY.
Sand-Jensen, K. and J. Borum (1991). “Interactions among phytoplankton, periphyton, and macrophytes in
temperate freshwaters and estuaries.” Aquatic Botany, 41, 137-175.
Sawyer, C.N., and McCarty, P.L. 1967. Chemistry for Sanitary Engineers, 2d ed., McGraw-Hill, St. Louis, MO.
Schindler, D.W. 1971. "Food Quality and Zooplankton Nutrition", J. of Animal Ecology, Vol 40, pp 598-595.
Schindler, D.W., Kling, H.J., Schmidt, R.V., Prokopowich, J., Frost, V.E., Reid, R. A., Capel, M. 1973. "Eu-
trophication of Lake 227 by Addition of Phosphate and Nitrate: The Second, Third and Fourth Years of
Enrichment, 1970, 1971, 1972", J. of the Fisheries Research Board of Canada, Vol 30, pp 1415-1428.
Schneider, M. and L. Hamilton. 2015a. SYSTDG Developer’s Manual. U.S. Army Corps of Engineers, North-
western Division, Reservoir Control Center.
Schneider, M. and L. Hamilton. 2015b. SYSTDG User’s Manual. U.S. Army Corps of Engineers, Northwestern
Division, Reservoir Control Center.
241
References
Shanahan, P. 1980. “XXXXXXXXXXx” Technical Report 268, R. M. Parson Laboratory, MIT, Cambridge, MA.
– setting vertical eddy viscosity critical
Shanahan, P. and Harleman, D. 1982. “Linked Hydrodynamic and Biogeochemical Models of Water Quality
in Shallow Lakes,” Technical Report 268, R. M. Parson Laboratory, MIT, Cambridge, MA.
Simoes, F. 1998. "An Eddy Viscosity Model for Shallow-Water Flows," Water Resources Engineering 98,
ASCE, NY, 1858-1863.
Sher-Kaul, S., Oertli, B., Castella, E. and J. Lachavanne (1995), “Relationship between biomass and surface
area of six submerged aquatic plant species.” Aquatic Botany, 51, 147-154.
Shubel, J. R., Wilson R. E., and Okubo, A. 1978. Vertical transport of suspended sediment in Upper Chesa-
peake Bay. Estuarine transport processes. Edited by B. Kjerfve, University of South Carolina Press, Co-
lumbia, S.C., pp. 161-176.
Smith, D.J. 1978. “WQRRS, Generalized computer program for River-Reservoir systems,” USACE Hydrologic
Engineering Center HEC, Davis, California, User's Manual 401-100, 100A, 210 pp.
Soong, T. W., DePue, M., Anderson, D. 1995. “The Changes of Manning’s Roughness Coefficient with River
Stages,” Water Resources Engineering, ed. by W. H. Epsey and P. Combs, ASCE, pp.1759-1763.
Spencer, J. W. 1971. “Fourier series representation of the position of the sun.” Search 2(5), 172.
St. John, J. P, T. W, Gallagher, and P. R. Paquin, 1984. "The Sensitivity of the Dissolved Oxygen Balance to
Predictive Reaeration Equations", in Gas Transfer at Water Surfaces, W, Brutsaert and G. H. Jirka (Eds).,
D. Reidel Publishing Co., Boston, 577-589.
Stumm, W., and Morgan, J.J. 1996. Aquatic Chemistry, Wiley Interscience, New York, NY.
Sullivan, A. B., Rounds, S. A., Asbill-Case, J. R., and Deas, M. L. 2013. Marcrophyte and pH Buffering Updates
to the Klamath River Water Quality Model Upstream of Keno Dam, Oregon. U. S. Geological Survey
Scientific Investigations Report 2013-5016, 52 p.
Swean, T., Leighton, R., Handler, R., Swearingen J. 1991. “Turbulence Modeling Near the Free Surface in
an Open Channel Flow”, AIAA 91-0613, 29th Aerospace Sciences Meeting, Reno, Nevada, January
7-10.
Thackston, E. L. and Krenkel, P. A. 1966. "Reaeration Predictions in Natural Streams," J. San. Engr. Div.,
ASCE, 89(SA5):1-30.
Thackston, E., L. and Dawson, J. W. 2001. “Recalibration of a Reaeration Equation,” J. Envir. Engr., ASCE,
127(4), 317-321.
Thibodeaux, L. 1996. Environmental Chemodynamics, Wiley-Interscience, NY.
Thomann, R. V. and Fitzpartrick, J. F. 1982. “Calibration and Verification of a Mathematical Model of the
Eutrophication of the Potomac Estuary,” report by HydroQual, Inc. Mahwah, NJ, to DES, District of Co-
lumbia.
References 242
References
Thomann, R. V. and Mueller, J. A. 1987. Principles of Surface Water Quality Modeling and Control, Harper
and Row, NY.
Thornton, K.W., and Lessem, A.S. 1978. "A Temperature Algorithm for Modifying Biological Rates", Trans-
actions of the American Fisheries Society, Vol 107, No. 2, pp 284-287.
Tsivoglou, E. C. and Wallace, S. R. 1972. "Characterization of Stream Reaeration Capacity," USEPA, Report
No. EPA-R3-72-012.
Ugarte, A. and Madrid, M. 1994. “Roughness Coefficient in Mountain Rivers,” Proceedings National Confer-
ence on Hydraulic Engineering, ASCE, pp. 652-656.
USBR. 2001. Water Measurement Manual, US Department of the Interior, Bureau of Reclamation, Wash-
ington, D.C., 317 pp.
Vollenweider, R.A. 1968. "Scientific Fundamentals of the Eutrophication of Lakes and Flowing Waters, with
Particular Reference to Nitrogen and Phosphorus as Factors in Eutrophication", Tech. Rept. OECD,
DAS/CSI/68.27, Paris, France.
. 1976. "Advances in Defining Critical Loading Levels for Phosphorus in Lake Eutrophication", Mem.
Inst. Ital. Idrobiol., Vol 33, pp 53-83.
Wake, A. 1977. "Development of a Thermodynamic Simulation Model for the Ice Regime of Lake Erie",
Ph.D. thesis, SUNY at Buffalo, Buffalo, NY.
Wanninkhof, R., J. R. Ledwell, and J. Crucius. 1991. “Gas transfer velocities on lakes measured with sulfur
hexafluoride”, in Proceedings of the Second International Symposium on Gas Transfer at Water Surfaces,
edited by S.C. Wilhelms and J. S. Gulliver, pp. 441-455, American Society of Civil Engineers, New York.
Weiler, R.R. 1974. “Carbon dioxide exchange between water and atmosphere, “Journal of Fisheries Re-
search, Board Committee, Vol. 31, pp. 329-332.
Wells, S. A. 1997. “Theoretical Basis for the CE-QUAL-W2 River Basin Model, “ Department of Civil Engi-
neering, Technical Report EWR-6-97, Portland State University, Portland, OR.
Wells, S. 2001. “Turbulence Closure Modeling in CE-QUAL-W2”, Research Report, Department of Civil and
Environmental Engineering, Portland State University, 37 pp.
Wells, S. A. 2009. “Hydrodynamic Compressible and Incompressible Modeling of the Density Structure and
Hydrodynamics of Hypersaline Systems,” Proceedings, IAHR Conference, Vancouver, BC, August 9-14.
WES. 1996. Evaluation and Analysis of Historical Dissolved Gas Data from the Snake and Columbia Rivers,
Waterways Experiments Station, ACOE Dissolved Gas Abatement Study Phase 1 Technical Report, Vicks-
burg, MS.
WES. 1997. "Total Dissolved Gas Production at Spillways on the Snake and Columbia Rivers, Memorandum
for Record," Waterways Experiments Station, ACOE Dissolved Gas Abatement Study Phase 1 Technical
Report, Vicksburg, MS.
243
References
Wuest, A. and Lorke, A. 2003. Small-Scale Hydrodynamics In Lakes. Annu. Rev. Fluid Mech. 2003. 35:373–
412, doi: 10.1146/annurev.fluid.35.101101.161220.
Wunderlich, W. 1972. "Heat and Mass Transfer between a Water Surface and the Atmosphere", Rpt. No.
14, Rpt. Publication No. 0-6803, Water Resources Research Laboratory, Tennessee Valley Authority, Di-
vision of Water Control Planning, Engineering Laboratory, Norris, TN.
Yakhot, V. and Orszag, S. 1986. "Renormalization Group Analysis of Turbulence 1: Basic Theory," J. Sci.
Comput. 1(1), 3-51.
Yu, S.L., T.J. Tuffy, and D.S. Lee. 1977. “Atmosphere Reaeration in a lake,” Office of Water Resources and
Technology, U.S. Department of the Interior.
Zhang, Z. and Treadhill, T. 2019. “CE-QUAL-W2 Version 4.2 for Predicting Total Dissolved Gas,” Technical
Note, ERDC/TL, Corps of Engineers.
Zison, S.W., Mills, W.B., Deimer, B., and Chen, C.W. 1978. "Rates, Constants, and Kinetics Formulations in
Surface Water Quality Modeling", EPA-600/3-68-105, US Environmental Protection Agency, Washing-
ton, DC.
References 244

W2manual42 Part2 Theory Rev1

Uploaded by

Copyright:

Available Formats

W2manual42 Part2 Theory Rev1

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

W2manual42 Part2 Theory Rev1

Uploaded by

Copyright:

Available Formats

CE-QUAL-W2: A Two-Dimensional, Later-

ally Averaged, Hydrodynamic and Water

User Manual Part 2: Hydrodynamic and Wa-

Department of Civil and Environmental Engineering

Bottom Shear Stress ........................................................................................................... 28

Absorbed Solar Radiation by Water Under Ice ........................................................... 97

River Reaeration Equations ............................................................................................. 152

Advective-Diffusion Equation Numerical Solution ............................................................ 221

Figure 1. Definition sketch of coordinate system for governing equations where x is

Table 1. Governing equations with and without channel slope. ......................................... 17

• User Manual Part 4: Model examples

This report should be cited as follows:

Wells, S. A. (2019) "CE-QUAL-W2: A two-dimensional, laterally averaged, hydrodynamic and wa-

or if a specific section with a different primary author:

Berger, C. (2019) “Macrophytes” in "CE-QUAL-W2: A two-dimensional, laterally averaged, hydro-

The manual is divided into the following sections:

• x and y are horizontal coordinates

Turbulent Time-Averaged Equations

2 Hydrodynamics and Transport

Figure 2. Definition sketch of turbulent time averaging for velocity.

Hydrodynamics and Transport 3

Figure 3. Definition sketch of turbulent shear stresses in x-direction.

4 Hydrodynamics and Transport

Note that the turbulent shear stresses are defined as follows:

Adjusting the Coordinate System

Hydrodynamics and Transport 5

The gravity acceleration is a body force that is then represented by a vector:

h = surface normal from the earth’s surface

This term can be written as three vector components:

6 Hydrodynamics and Transport

and also in terms of the gravity components:

Governing Equations for General Coordinate System

Hydrodynamics and Transport 7

Simplification of z-Momentum Equation

8 Hydrodynamics and Transport

𝜕(𝑢̄̄ +𝑢″ ) 𝜕(𝑣̄̄ +𝑣 ″ ) ̄̄ +𝑤 ″ )

The lateral average of a double primed variable is by definition zero:

Hydrodynamics and Transport 9

𝜕(𝑢̄̄ +𝑢″ ) 1 𝑦2 𝜕(𝑢̄̄ +𝑢″ ) 1 𝑦2 𝜕𝑢̄̄ 1 𝑦2 𝜕𝑢″ 1 𝜕 𝑦2 1 𝜕𝐵𝑢̄ ̄

Combining terms, the continuity equation becomes:

The unsteady acceleration term:

The convective acceleration terms:

𝜕(𝑢̄̄ + 𝑢″ )(𝑢̄̄ + 𝑢″ ) 1 𝑦2 𝜕(𝑢̄̄ + 𝑢″ )(𝑢̄̄ + 𝑢″ )

10 Hydrodynamics and Transport

Similarly for the other two convective acceleration terms:

𝜕(𝑢̄̄ +𝑢″ )(𝑤

𝜕(𝑢̄̄ +𝑢″ )(𝑣̄̄ +𝑣 ″ )

The gravity term:

The pressure gradient term:

The shear stress terms:

Hydrodynamics and Transport 11

𝜕𝐵𝑢̄̄ 𝜕𝐵𝑢̄̄ 𝑢̄̄ 𝜕𝐵𝑢̄̄ 𝑤̄ 𝐵 𝜕𝑝̄̄ 1 𝜕𝐵𝜏𝑥𝑥 𝜕𝐵𝜏𝑥𝑧

Summary of Laterally Averaged Equations

Simplification of Pressure Term

12 Hydrodynamics and Transport

Figure 9. Schematization for simplification of pressure term.

The revised form of the x-momentum equation is then:

 UB  UUB  WUB  g cos B  z

Hydrodynamics and Transport 13

CE-QUAL-W2 coordinate system,  = 0 g