Water Resources Yield

WATER RESOURCES YIELD by Thomas A, McMahon and Adebayo J. Adeloye ‘Water Resources Publications, LLCle 2.5, are 25 for the series, the kea int year L-day Dw flow is 13 he sstimate) is columns 1 Example 2.1 >ability of uration, In are used in onsecutive N-n+1) nit series in These are ise. led vent of n= ated ure iat. the dependent estimated ces, Gagiee? 7 ‘Normal and Gammna distributed n-year flows Itis known for a Normal distribution thatthe sum of n independent variables (say annual streamflows in this ease), each with mean jt and variaaceg? is a Normally distributed variable with eee (2.24) 52 = no? (2.25) where ity and O,? are the mean and variance of the summed n-consecutive annual Neamfows, Jf the variables ae auto-comelated then Equation (2.25) is modified Kotz & Neumann, 1963) as follows: Oy? = Ry no? (2.26) 1+p, _ 2pi(l-of) atte Ra = Tp "Alp,)? (2.27) Where Rr isan adjustment factor for the lng-one serial corelation between the summed n- seenecutive values and A, is the Iag-one auto-correlation coefficient of the annual streamflows, 2. gimilar set of equations is available for streamflows that are Gamma distributed (see Srikanthan & McMahon, 1986, p. 71) as follows: B= 8 Ry 2.28) 229) Where 5 and B are the shape and scale parameters of the Gamma distributed annual Streamflows, and 8, and By, are the shape and scale parameters of n-consecutive years streamflows. To estimate the average recurreice interval of an ‘n-year flow (Q, ) drawn from a record of N Yeats, the value is adjusted thus Qi =ZnQy (2.30)Chapter computed as (2.31) ution, ie. 2 (2.32) omal vilny re (2.33) Ginger? Py 28 20 020 eta ‘Sample size ‘Sampo aa Figure 2.15: Ratio of non-overlapping sums to overlapping sums for Gamma distribution Adapted from Srikanthan & McMahon, 1986 Figure 6, by permission of American Society of Civil Engineers (1 coefficient of skewness, py=lag-one auto-correlation coefficient) and where f(z) is the probability distribution function for the standard Gamma distribution, which can be obtained from the Normal z using the Wilson-Hilferty or any other suitable transformation. For the standard Normal distribution, the relationship between z and Pr is tabulated in most standard statistical textbooks; alternatively, given z, Pr may be approximated using (Stedinger et al, 1993, Equation 18.2.2): _ (B32+351)z +562 103 165 Pr=1—0.Sexp] 30 pesueor8 @ & 8 Told wonvjodeanxe Hie oan OMA, — ai oT 8 (2712 59 w9¢0 = W204 wonrmiasP HMAIOM et] 68 5 See ol s Be es ES a oT sr 3 een eet] 98 set] at é ‘nu apmns x wore wens eg a 2 eae twee] ve zat] at z eae eer] oo oart| owt 8 es sso] se sro] at s | wet) 1 set] oat 4 paren wt] os eet] ot & Nonmode are sue icles eit) oF é wil es az) og 3 ez] zs vt| og 8 (560-"y "60 ore] set] ok e wt] wt] 9 a E- 09T ve ore . é cet] ec ez] + 3 wt] oe out] og 5 ho 3 wt} oot] 2 Pa ee $ eof Tar 3 (65 30) a1yyoy one 2g Soprano BS! 06-1999BS 6349-1:1944°ensed Copy: Giorgio Cavalier, none, 1-Deo-00, Uncontrolled Copy. © BSI ‘WN Ws {CPF 90 woRvIaeA— pL Om, (1 wotspu2 ‘sp 40 vorjounp u sanon 30 Jeg, ono 00s oz 201 000 00s. 0902 ‘© BST 06-1099 RBS 6849-1:1984 SSS a —— In 20.2 some simple diffraction solutions are given Which ignore the effects of wave refractions and reflection, These solutions can, in general, only be used to obtain a first estimate of storm or swell wave heights inside large harbours in which the boundaries are several wavelengths distant from the harbour entrance and therefore unlikely to be good reflectors, Under these conditions there is insufficient reflected wave energy present in the harbour to alter seriously the wave diffraction pattern obtained ignoring reflection. This situation is not valid for long waves with periods of the order ‘of minutes where even a beach can be expected to produce a noticeable reflection, nor for smaller harbours for fishing boats and pleasure eraft which frequently have reflective boundaries, such a3 vortical faced quay walls, and which consequently ‘may resonate at storm and swell wave periods ‘The diffraction solutions described in 29.2 are ‘modified in 29.2 to include the affect of wave rofraction and, although these solutions can be further extended to include some wave reflection, ‘the type of model describod is not capable of allowing for harbour resonance and is therefore similarly limited to the caleulation of storm and swoll wave heights inside large harbours, but to a ‘greater dogree of accuracy. In29.4 some of the basic resonant modes of harbour oscillation are described and in 29.6 and 29.6 ‘guidance is given on the use of models in harbour analysis, 28.2 Wave diffraction for a flat seabed 29.2.1 General, The solutions prosonted in 29.2.2 to 29.24 are applicable to regions sheltered by breakwaters within which negligible refraction and reflection takes place, 29.2.2 Single breakwater case, The solution for Waves incident on a single breakwater arm is similar to the solution found for the diffraction of light around a somi-infinite screen. Figure 17(a) shows a typical diffraction diagram in which wave fronts and contours of the diffraction coeflicient, Ky, the ratio of the diffracted wave height over the incident wave height, on the sheltered side of a single, vertical faced, impervious broakwater arm are plotted for the case of uniform waves approaching with their crests parallel to the breakwater. In diffraction diagrams of this type distance is normally expressed in units of wavelongth so that itis necessary to work out (from Figure 11) the value of wavelength appropriate to the water depth and wave period, Diftraction coofficients for a xange of uniform wave directions incident ona single breakwater arm are shown in Table 1in which the coefficient is given for various positions defined by polar coordinates and for various incident wave ditections where angles and distances are as defined in Figure 170). Diagrams ofthis type gonerally show that, along the “shadow” line drawn on the sheltered side from the breakwater tip parallel tothe incident wave direction, the diffracted wave height is approximately half the incident wave height and, on each side ofthis lino, the diffracted wave height is changing rapidly. The relatively rapid variation of ‘wave height in this gion is unlikely to occur to such an extent in real situations due to wave onorgy being spread in direction. Although wave refraction outside the harbour will tend to reduce the directional spread of incident wave energy, the remaining epread will in many situations, be sufficient to reduce the rato of variation in the thooretial diffraction pattorn obtained for uniform waves given in Table 1 29.2.8 Double breakwater case. Where a harbour entrance is formed by two breakwater arm, iis acceptable to apply the solution given above for a single breakwater arm to each breakwater in turn provided the width is 6 wavelengths or greater, For entrance widths oflos than 5 wavelengths a ‘more exact solution, which makes use ofa series of Mathieu functions, can be given, Results for seven angles of wave approach to entrance widths of 05, 1, 2and 3 wavelengths are summarized in Figure 18 where an intensity factor Ip (6 together with its relevant scale is given for each case. The ‘wave diffraction coeflicient, Ki, may be obtained at « position with polar coordinates (7,6) by the following procedure (eee Figure 19) On the intensity factor plot appropriate to the entrance width and incident wave ditection, a line is drawn at the required angle 9 extending from the centre ofthe entrance. The distance along this line to the point where it outa the intensity plot gives the valuo of the intensity factor Ip @) when measured at ‘the relevant scale. The diffraction coefficient Ky at ‘the position (6) is then given by: Ky = Mo@Lir) IfKg is written in the form of the wave height Hg over the incident height, Hi. then the above relationship ean be rowitten to giv Ho = M(H tp(O)Lin) from which it can be seen that the difracted wave Iheight fora given value of@ decreases with radius r just as if sourco of wave energy (Hig! ly) L) existed atthe harbour entrance. ™ ‘© BSL 05-1099 'S€.@ ‘Ado potlemucoun “00-26¢-+ ‘suoU "uoyeReD o1610I9 foo posuED!][PROOF COPY (HE!2001/022448) 0023050HE * nparametric Approach for Estimating Return Periods of Droughts in Arid Regions Tae-Woong Kim’; Juan B. Valdés*, and Chulsang Yoo* Abstract: % damage in terms of both natural environments and human lives, and hydrologists and water resources managers are conc ‘may not reveal significant ‘the univariate and mltiv emo estimator is presented| Hh estimating the relative frequencies of these events. Univariate parametric methods for equeney analysis ships among drought characteristics. Alternatively, nonparametric methods prove local estimates of sity function by using weighted moving averages of the data in a small neighborhood around the poat fo cxamine the univariate as well as the bivariate behavior of droughts. Aer evaluating end of estimation and opposed?ta ce: methods. A methodology for estimating the retum period of droughts using @ nonparametric validating a nonparametric ke Conchos River Basin in Mexico. ‘than 120 years for the conditional dist 9, & dough frequency analysis is conducted to estimate the rtum periods of droughts forthe 0 that, for the univariate analysis, the return periods of the severe drought occuring in ‘of severity and duration ‘h the be a 10 yes rie For yar ony teen pea we pointy 30 yn fo jot tttos ed re ‘Introduction ee Rope teeter planets Ce eee are an Scares aes See neon eee Spee eee ee oe ree ee paneer Cpe eres Fee tere etn tery Sere err ded mena ore. ex toned hs ee ee ee "Grane Reseach Asian Dep of Gv Engierng and Bagh teeing Mecha, nd Cone for Ssailiy of Sena Hy gy an Ripran Areas (SATA), The Unio Aza, Toso, AZ 521-0072 Bal: ee@enallatocn et “profesor and Head, Dept of Cl Enierig snd Eegnesng Me- chanics, and Center for Saray of Seti Aré Hylony and Ri rian Ares (SAHRA, Te Uni of Arizona, Taso, AZ 65721-0072 Erma eau arnnncts “Associate Profs, Dep of Civil and Envioomeaal Egiecig Kore Uni, Sos, Ker 136-701, Eau yooghenac kr ‘Note. Dison ope ut Feary 1, 2004 Sept discuss sat be bite fo lil paper.) exzad he clsng dt by ‘ve mont, witen reqs mus be fed withthe ASCE Managing ioc The wasp orth paper was submited for evi and poe stblepblcnton on October 18,200, syproved on ftir 21,203 ‘This pone is par of he Journal of Hydloie Engineering, Vol 8, No. 5, Sepember I, 203. ASCE, ISSN 1084065920515" 10S1800, % DOE 10 1061Asc=y1094.069(200)8 we CE Databaso subject headings: Dru CA Froqueney; Estimation v, pacts of droughts (Smakbtin 2001), Lee et al. (1986) developed a practical approach forthe frequency analysis of multiyear drought Aurations of annual streamflow series. A technigue that smooths the frequeney-curve iegularity of drought durations was used to duce the statistical uncertainties associated with sample-sze tations. Nathan and MeMahon (1990) evaluated the epplice- ‘of the Weibull distribution to low-flow frequency analysis. ysyctavestigated the differences between low-flow frequency e=- eso on calendar and hydrologic yeers, and they made hung and Salas 2000). Femindez and Salas the definitions of the retum period and lure of hydraulic structures, especially ‘ows, annual maximum floods, and either dependent or indepen th drought occurrence prob- ight events for dependent hydeoogicel drought? whi det, Chung and Salas 200 ables, eu perods, and i hnyelogic processes. Rather ‘wich wadioally have been w cesses, lowarder discs a (ARMA) models were used for model they are adequate for process exh dence, They coclnded tat low oder AB for representing the occurence of wet end dR sequently, ar capable of modeling and salty tions that are observed historically. ‘Aenea! approach used in drought selted feg is to derive the distibutions of dought durations separately (Sen 1976; Mair etl 1982; Stedinger et ing hydrologic pro- moving average sy ry yours, since JOURNAL OF HYDROLOGIC ENGINEERING © ASCE / SEPTEMERIOCTOBER 2008/1 PROOF COPY [WE:2001/072449) 002305QHEPROOF COPY (HE/2001/072449) on230S0HE ‘mutually corelated. Separate analysis of drought characteristics canaot reveal the significant corelation relationships among drought characteristics (Shiau and Shen 2001). Recent studies ied bivariate distributions to hydrologic. frequency ‘Ys primarily based on the estimation of the probebility density fuggion (PDF). The parametric. ap- proaches for estimating ty density fanction must as- sume that date are draw distributions. However, there tion for hydrologic variables 1994; Smakhtin 2001). Daring methods for fequeney analysis pirealyacepte ds ‘wll asthe bivariate Behavior of droughts condoned ls Ropers of dong. Aer wit snepns of eb feof iret nari and mbes nonpanncan os fretting te psbeity density fasten po lative density function (CDF) using data in Texas Climstologie Fig. 1. Conehos River Basin, Mexico (Schmandt 2002) Tis. 1 Conshos River Basin, Meni (Sehmandt 2002) the precipitation deficit at a certain location, Because iti a stan- ardized valu, the PDST allows a comparison and assesstnent of, ® regional drought. tn adition, it provides a measurement of the | donomality of weather fora region, as well as an opportunity to place cutent conditions in ahistorical perspective. Regional val les ofthe PDSI were calculated using the areal meat of prep ‘ation and temperature reconstructed across the besin, In this study, «drought is defined, using the theory of runs, as { ‘mn event during which the PDS! is continuously below a cerain level as show in Fig 2. Palmer (1968) categorized a drought condition using the PDSI. In dry spells (PDSI<2), « PDSI of 50.5100 is considered near noma, =1.0 to ~0.5 isan incipient ‘san extreme drought. Fora selected threshold level (eg, ht, ~2.0 to —1.0 is a mild drought, -300 to -20 is a ‘Theo, they are evaluted and validated for estimating the cura ogi drought, ~40 0 ~3.0 is a severe drought, nd “AO ok Region S, which has a Tonger record than the Conchos River Basi, Next, the return periods of droughts are estimated by using honperametic methods proposed in this study. They inelude bot the univariate and the bivariate behavior of droughts. Finally, valuations of historical droughts inthe basin and discussions of the methodology for drought fequency analysis are presented. Basin Information ‘The Conchos River Basin, shown in Fig. 1, lies within 26°N- 30°N and 104°W--108°W in the ardisemiari area of the Mex can state of Chihushua, and has aa are of 71,964 lan!, The wi ters of the Conchos River are used primarily for irigation of nearly 80,000 ha, as wel es fr use in hydroelectric powerplants located at La Boquilla and other minor dams, The Conchos River is the most important tributary ofthe BravolGrande River sine it supplies approximately 70-80% of the mainstream flow of the Lower BravovGrande River ebove the binational reservoir of Amistad. Important Mexican cities, such as Chihuatue, Hidalgo el Paral, ond Delicias, are inthe basin and are growing rapidly due to the increasing industalization. Sustainability ofthe ex. nding water usage for agriculture, in addition to urban pusposes and water rights ofthis region are of concer to the United States and Mexico, ‘AS an indicator of drought severity, the Palmer drought severity index (PDSI) was used in this stady. Palmer (1968) based tis index on balance between moisture supply and demand. It mee. sures the departure ofthe moisture supply by taking into account PEDBIEO) «drought eve athe eof negative run (represied sets in Fig. 2), «drought duration (D,) is the length of 2 nt a drought magnitude (iM) i the area below the thr and a drought intensity (Z) i the ratio of magni tude of (lem Mj1D). In addition, drought peak (P)) is defi ‘maximum deviation (rinizaum PDSD during & drought e 8 shows a time series of drought characters. tics for the Cepns River Basin. Drought characterises ere ine ividully uncort@ag@lgandom variables, since each time series of drug E:jpus the autocorelaton function, which is not significant at onfidence level Kim etal. (2002) the drought characteristics in the Conchos River BasinYking I and standardized precipita tion index as indicators of verity. A drought intensity ey Da Magi, ret ao, ree eten Pos! eee ont oe Tie al Fig. 2, Defision of rug crass sig POST H&S eae 2/ JOURNAL OF HYDROLOGIC ENGINEERING © ASCE / SEPTEMERIOCTOBER 2000 PROOF COPY [HErzo01I0z2449) oo2305qHE[PROOF COPY [HE:2001/022440) 002505QHE Fig. 3. Time series of drought characteristics in Conchos Rive Basin with PDSI threshold level=0, 1935-1998: (2) drought aration; (b) drought intensity () drought peak areal extent-frequency curve to characterize the spatial pattems of drought was also developed. The PDSI was shown tobe useful foe both defining and essessing the spatial extents of droughts in that basin Return Period of Droughts ‘The objective ofa frequency analysis is to relate the magnitude of| ‘extreme evens to ther frequencies of occurrence through the use of probability distributions, It is usually assumed that the var ables being analyzed are independent and identically distributed (Chow etal. 1988). The retum period of an event in any obser- vation is the inverse of its exceedance probability, p= P(X Dah ie 1 1 PUOSs) =P R -magniude of the event having @retum period of 7. In this study, the random varible X in Ea, (1) is drought uration. Because droughts sometimes last more than | year, the rought characteristics, shown in Fig. 3, can be analyzed a8 a ‘partal duration series of independent events, Eagleson (1972) and Willems (2000) suggested thet the distribution for partial duration sees of independent events should be converted to a equivalent istibusion for an annual exceedance series using the peak-over r o threshold or pacta-duation series methods. Ifthe marginal ca- mulative disebution of drought draon fra given tresbold level is denoted by F(a), the retum peviod af drought duration, Ty, is defined in Eq) eg) ve Tra) Tra’ ere On; 1=ttal leg ofthe observed PDI (jer) and ‘total numberof drought event, dang N ‘The bivriae retin period of droughts canbe estinated by substintng the univariate CDF into the bivariate CDF. For ex. mpl, the tvaite resin periods of drought curation (d) and doug inna (9 ae given by 1 Fo a] 1 Fo) o where Ty joitzetum period of drought duration and intensity; Ta;=condisoeal return period of drought dation given inte Sif} Fo,(d.i) (= P(D=d,l=i)) joint cumulative distribution ‘of drought duration and intensity; and Foy (di) [=P(D=)], where Op is « nonparametic quantile sstimator and Tis the parametsc estimator, is greater than 1 (Garrell and Davis 1982) Fig. 5 shows the marginal PDF and exceodance probability for ‘tought durations in the Conchos River Basin, The marginal PDE was escinated by the univariate kere density estimator with op- ‘mized bandwidth forthe sample size. The exceedance probabil ity was calculated fiom the CDF, whick was estimated by int ating the PDF, muliplicd by 8 feg, O[1 ~Fo(d)]}. Fig. 5@) sisplys the PDs constructed from four daa sets corresponding to the threshold levels, and Fig. S(O) compares the exceedance probabilities estimated from the PDF, in whieh the underlying shapes contrast more effectively. The PDF for the Conchos Rivet ‘Basin has « more apparent bimodal shape, which causes he ex. ceedance probabilities tobe fat in the tal, as shown in Fig, 5(), Classical parametric estimation procedures ere weighted toward fitting the main body of the probability density and the weights are negligible in the til of the distibution, Considerable uncer. ‘ainties for the magnitude of extteme events exist even if the Retum Psi Yea} Fig. 6. Retum periods of drought duration as faction of threshold levels (TH), Conchos River Basin, Mexico, 1935-1998 prametrc PDF fits well (Moon and Lall 1994). ‘The return per- (ds of drought durations forgiven threshold levels ae shown in Fig. 6. Bivariate Analysis ‘Shiau nd Shen (2001) presented a joint PDF of drought duration i and severity, expressed as the product of the conditional distribu. tion of drought severity given duration and the maeginal distribu. tion of duration. The derivation ofthe joint PDF of drought char. acteristics is not mathematically tractable, because no. known nalytial solutions ere avilable Jn this study, a bivariate kernel estimator is also employed to tthe bivariate behavior of droughts (duration intensity, Peak) inthe Conchos River Basin, Fr the observations 1 (0835-1850, 72 doug evens elo ¢ FDstnt ° presents a drought condition. The saple statistics of rou tes of interest for dhe bivariate study are presented srought duration is highly eomeleted wth intensity Fig. 7 shows the joint PDP (JPDF) for drought uration sand for drought duration and peak est mated diret date using a bivariate kernel estimator ‘The JPDF for peak has an apparent principal axis in the disgonal dire to thei high comelation, Parametric ‘models are dificult ause ofthe high corelation be- intrest. For exemple, in a Gumbel rd,Giibel marginal distributions, the soss-corelation coefficient ve a value between 0.0 and 0.67 (Oliveria 1975; Yue et For both duration-iatensity sand duation-peak, the corel ove the upper limit, Maximum Minimum 3740 310 Meaa =1910 Standard deviation 1030 Stew ~0210 ~ Kurtosis 1000 “1 CCoutelton with drain 724 as {6/ JOURNAL OF HYDROLOGIC ENGINEERING © ASCE / SEPTEMERIOCTOBER 2003, PROOF Copy [HEV200%/072449) on2s05QHEPROOF COPY [HE/20011022449] 02305QHE Fig. 7. Joint probability density fuxetion (PDF) fr: (a) duration intensity; (0) duraion-peak, Conchos River Basin, Mexico, 1935- 1998 Joint Distributions of Drought Duration {and IntensityPeak ‘The estimation of the joint return periods of Ty (duration and intensity) and 7, (Curation and peak) is of interest in this study, Fins, the joint probebilty density finction for durations corresponding io a certain intensity value, P(D,I=i), was estimated fom the JPDF by slicing the IPDF between 0 tnd fp and sum- ming up for each daration, Then, the joint cumulative density fiction for duration corresponding to certain intensity value, P(D=d,!i), was constructed by integrating the joint probability density function, P(D,t=i9) Similarly, the joint probability density function for durations corresponding to certain esl valie, P(D,Psp), and the joint cumulative density fusion for durations corresponding toa certin peak value, P(D vert Erato Npouesetn @ . . » 0 0 @ uration (month) ®) ee nd nonparametic method (NP), Conchos River Basia, Mexico, 1935~1998: (a) histogram and PDFs; (b) CDFS ‘record period, 1895~1998. Both SAB and ARB measure system atc errors inthe esiation of the CDF, which means the consis- tency of estimation of the CDF above or below the actual value (EP) over the entre period and the subperiod of observations, ‘respectively. Table 3 compares the sum of ARB (SARB) and the | average of ARB (ARB). The results in Table 3 show that the CDF estimated by the nonperametric kemel estimator has less bias and gives more stable results fr estimating the CDF than the parametric methods under both eriteria for both validation peti- os Applications ‘The nonparametric kernel estimator for estimating the PDF and CDF of droughis has proven to be more atrative than parametric ‘Table 2. Sum of Absolut Bias of PDF and CDF Estimated by Para- Table 9. Absolute Relative Bias of CDF Estinated by Paramerie and Nonparameaie Method for Texas Cimatologc Region 5 Gtitioa Paved ___NP__Logoomal Gumbel Pearson SARB 1895-1950 09155 61038 25220 2.0005 1935-1998 37728 LLITBL—GaDKS 4318, AARB 1895-1950 0.0381 02563 0.1051 00834 1938-1998 01309 04469 026K amas ‘Note: SAR=aum ofthe absolute ehtve bias; AARB=average of the shoo rlave bias; NP=noeperametis, rethods, using the erieria for this case study under the assump- tion that the PDF and CDF ae alterative smooth curves in order to represent te histogram an the cumulative distribution, respectively, Bowman and Azzalini (1997) provided an introduction to Keemel-based methods with an easy-to-use S-plus library, SM, ‘wich we applied to drought frequency analysis. A Gaussian density function was used fora smooth Kemel fonction with « mean of zero and standard deviation of h. When « Gaussian kere! function is used, / will be a smooth curve having derivatives of| all orders to satsy the condition that fis inherently continuous and dtfreniabe by definition. When K isa noma density, the evaluation ofthe optimal formule for h yields 4 tern moral ‘where /yp=optimal bandwidth; 0 denotes the standard devise tion ofthe eistsbution in dimension d and p=number of dimen. sions, eg. p= fora univariate kere estimator and p=2 fora bivariate Keoel estimator. ha e4 a variate Analysis the univariate analysis for droughts using the nonpara- ue, time series of drought durations were con- time series have PDSI thresholds of 0.0, -0.5, 0, and ~40 forthe Conchos River Basin. For exam) a) shows the érought duration time series for & teshold@PPD31=0.0. The sample statistics for the sets of rought duredGGgfRe shown in Table 4, where 8 is the number of drought event? by the data period (years) (n/N), which he sample size for estimating the PDF increases, both the’ the PDE. Lal and Moon (1993) je estimator is more useful for ns, since the relative ef ‘mean square eror Ihave shown thatthe ssnall samples than para ciency calevsted from Table 4. Sample Statistics of Drought Conchos River Basin, 1935-1998, ‘mete and Nonparametric Method for Conchos River Basin, 1935- Number of evens 22000 23,000 26000 od 1998 ‘Mean 18360 16910 13.380, 5430 Steodad devinion 18500 1530 13470 10980 Rinsion NP ___Lopnoraal Gomiel_—-Peanoni = uns nee feaes aon oom, : PDF 00s 00880 0z0—=AzaT Kursis 0680 ~0400 1220 420 5.256 cor 03902__assos oer? 0.9062 ® 03390384 040 _046i_0277 STheabal ‘Now NF =nonpermerie: JOURNAL OF HYDROLOGIC ENGINEERING © ASCE / SEPTEMER/OCTOBER 2003/5 PROOF COPY (HE/2001/022449] 0023050HEPROOF COPY [HE!2001/022440] 002305aNE A ° i a yen ge / a aod : J fF oor a | * Ze 40-5 0 5 10 15 20 25 90 36 40 45 80 OS OO 1 10 00 $000 Dates Senet @ Fig. 6. Retum periods of drought durations as function of threshold ioe oes Rv aos ae 7 eres OF fi well Mon and Lt 198, Th ou pr Str of dg dd fe he trees ee onl me Exosedence Prob Fig. . Comporison of marginal PDF end exosodance probability for ought durations estimated by nonparametti method foe threshold levels (TH), Conehos iver Basin, Mexico, 1935-1998: ) marginal PDFs; (b)exceodance probabilities [MSE(7))/MSE(Q,)], where Op is nonparamettic quantile estimator and Ty is the parsmetic estimator, is greater than 1 (Harrell and Davis 1982). ig. S shows the marginal PDF and exceedance probability for Doan] @ 10 Tw yo a i °] Mees a Joint Conditional [stvearrent aOR © e, Fig. 10, Retun period of bso sererst doug 7 River Basin, 1838-1998 () nia enely, () bia Sis Pacsnan-azor | Tacsoienains Fuego Taqmsnigt 631)» from Left] '8/ JOURNAL OF HYDROLOGIC ENGINEERING @ ASCE / SEPTEMERIOCTOBER 2003 PROOF COPY [HE!2001/022449) 0022050HE‘sH0s0¢200 [srrzz0/t00z/IH] Ad09 400He {6/6002 ¥360100R3"131435 / 39SV@ ONIONS O}DOTOUAAH 40 TwNeINOF 28 Ho Tea osu Kp PEN, (96) ZS 20K aN Vo soreon Pao Ps oops Ansty oN (EI) 9S SSI=LE1 (Oy Hog Hon cganensegoad yas pois 0 eves, (9002) "YM ‘ists051 (on)or “Beg mio 205 wy cE onwoni OSes ym BOONE, “(O0e) f TPUBLINS 161-161 BK, on ‘yeaapeye 2x Bayon fo jooqpuogy ste oan ajopom pus ehway., (Zee) HS ‘aoe Osi REA SA ‘sp oy andog you woBoj10o (5961) 9 A UL in yours euorouan mp svete, (Si61) “Lf WRMNO pai mea aye) Prony ur, (002) "es iggy por oms99 a09eN IDIe-geie (ae a flagD any «see Snot sono as poten, (Os po EN OTe-Se0e IDOE “4 ing emyePbatee fone pon 495 sqoupse vorpay ojneeb fay, “eT po" “OOH ‘52-60% (9 PE 10 32420 tom. 4p sine Seco 29) sana Pa un sewed 5p 9m aL (2651) NNSY POE ROR! nosy 234 «suonenp YH aS 30 HE 403 pears oy, (OR61) “VF “ous pe“ ‘rac-08 nosy 2m ,cwoneyand Sp aye [9p yoarundn y, (8661) °D "a "wg pe g uesdo =S001 (60 a non soe ORD YD Pe OHA -poeg mveumso Konanbog pony eusy, (Cg) 20094 Pos" mace “pe “eBag a'A pO nang STL HN TY THM A SHE 1A Berson jonas pa Seopa poe 0 Ase peo oo ep Sap Sap oe on J oR uy MAN, O61) "sOaNEE Poe TET Zo1T=E601 "syn 104 sto mAUD oso yoy soy woonny anon”) NPL ‘ey ote (Oz “COMBE we Jay Sogo ap aT FOO 39 Foe ines pe devant, (coe) ras poe TE "LCE 9-569 (049 Pyamworg rowan sty sag vom sa, 61) 4-9 seg mTOR oy yang At 2am ms hot ayy a Poe oun L451) LOH, ye-S SH “OP 7 mo sc | J whey xsd (666) 1 PS807 BE HEN “T'S m “Ste-s0e Ty "ier aay 7, seognddy 1 sno 2Ao}oH 30 8 Pos pound wag, (Ast) CT HHS pu epENEE o-is ‘ “Bug aot coopmanes EMDR | 9 OHO 490 pn pol me, Bog) es Poe “a “apaENL Bsi-6u8 (ne "a sony my «Komnbay pou 0 murat, (est) Swett ae Lez 91“ sy aD JP son 980, (OI) 0 HEEL PS ET Y Tees «Gems a osray ‘rower “eu 0202-9661 MORI 9° POI IP HOE 019 eAMGAPL Hid. (LEI) NO) FRY HP EODAN, AID sse-6se (0s “iq aojapy 1 sosoesd soapy wapuaiay Jo I Ho sporngeand suns mnt, (0002) "aT RS Pie “HD BID poh AoA AMOEN “mp pouty (W561) 11a OH HP “LORD yok oN ang gan Rx "onan as Yom ony aL mow He of santos Barons pay ugg)" wneary poe yee, hop (Oz “Bg OPA 509 -tonboy aop-no|o vonoussauruINdION, (8661) "A TMOREY ‘681-5881 “IIe “sy nog soy sooeN 2 poy Jo woes fue stmoumretaoN,, (S961) "EMORY ‘saouss9}oy -suounoo signe Bmpyoid pus woresedaid ‘ep sum Sims sy ‘oonayy ‘aSGOID “ydesour299 10 -tofgJo watmimdag 2930 sezeax> wasn. aq pw AONE, sayy Jo amare, meray a Jo oRUNdY Jae aC oF Ty axe 24 0oR92R6raVa seqUMN soUENABY “tonepmnog 20:08 (Poop 2p Jo umeong Ou om opm euoALY 30 Kus|ONI — ow ve seary waxy pow ABoqOIpAH Pung Jo Age “51g 265 Jaan 2m panoddns Som wo poseq sXe SL sjuowiBpejmouyoy -ssfqvur fovonbou 10 pian nour oroduou orueaig 99 uy AIsnoouranss pw Apse9 alous pau: ep 2q Aew Sonos pue vor wRoxp wog sous uaKOAE aur soo pu uBisop qjoq oy yes a Aprys SI MH preUIED spousd umjer owtag ay ‘ijnonied uy ysed ax) ut pounos0 1p s48noI am Suenyerd pus symeAp v Jo pouad wir ayy ‘unexznp 205 ryosn st poxpow posodend ty ye) aeaiput Apes Snip mF symsar eso oy “spomou oxnoumsed Jo suoRRIN 249 Jo aus sieumae prnco snaxp Jo uornaINSEP Aysqegaid 249 seuss 0 pou tnoweredcou & Buxsidde “expan “IyaNO 'S1E~10¢ “wonsog ‘omuepeay CG. ‘zou pooisiapan 2q pinoo syyBnoup Jo souATyEq TULA an pu ‘sunarsno 29 pivos sséqene Kouonbayy aperAran otra 249 ‘suonngunsip reuompucD pue wf Susp, ‘us 8 uopeinp wnoxp Jo woRNgiNsp |suORpLeD axy 30) om Jo poved wumanv pay o4Amyo eye aM _ Su 8 Stok Qo] usyyaxoa o porssd wns emia & oust 6 ‘auied paooai oxy Sump woxBox oq 1480p ssaroAat 450661 Om ut pono20 wom amp “use {any so ‘suo qu80up ayezeanrur punsiapun ob 2epio ur 25 0} peat ‘suonngusp feureun sq) $e (am 8 ‘Slorargeg ansqigegghd pouorrpnoo pue wif soxp “yeod pun ‘fysucqu “wones 8 SAUER wopueL parsaNIOD Ap x 22294 03P.OyY UH SEE BAR SI St yom 58 Knaxos oxy Susn ‘Kpros SU ou ye UE TO soy osntaag 'isSjune Sunvumse 203 Pes B5C-6¥2 ‘6 “AP 7 pom aN onagEoN ot -pumsdeou YE wg sous HoRE fds se ponoad 9 eo ‘aHDs0ez00l6r¥zz01L002/2H AdO9 sOOMd

Water Resources Yield

Uploaded by

Copyright:

Available Formats

Water Resources Yield

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

Water Resources Yield

Uploaded by

Copyright:

Available Formats

You might also like