INAUGURAL ARTICLE

Peak water limits to freshwater withdrawal and use

Peter H. Gleick1 and Meena Palaniappan
Pacific Institute, 654 13th Street, Oakland, CA 94612

This contribution is part of the special series of Inaugural Articles by members of the National Academy of Sciences elected in 2006.

Contributed by Peter H. Gleick, April 8, 2010 (sent for review February 22, 2010)

Freshwater resources are fundamental for maintaining human impact of human appropriations at various scales through the
health, agricultural production, economic activity as well as critical use of rainfall, surface and groundwater stocks, and soil moisture.
ecosystem functions. As populations and economies grow, new An early effort to evaluate these uses estimated that substantially
constraints on water resources are appearing, raising questions more water in the form of rain and soil moisture—perhaps
about limits to water availability. Such resource questions are 11;300 km3 ∕yr—is appropriated for human-dominated land uses
not new. The specter of “peak oil”—a peaking and then decline such as cultivated land, landscaping, and to provide forage for
in oil production—has long been predicted and debated. We pre- grazing animals. Overall, that assessment concluded that humans
sent here a detailed assessment and definition of three concepts of already appropriate over 50% of all renewable and “accessible”
“peak water”: peak renewable water, peak nonrenewable water, freshwater flows, including a fairly large fraction of water that is
and peak ecological water. These concepts can help hydrologists, used in-stream for dilution of human and industrial wastes (3). It
water managers, policy makers, and the public understand and is important to note, however, that these uses are of the “renew-
manage different water systems more effectively and sustainably. able” flows of water, which we explain below. In theory, the use of
Peak renewable water applies where flow constraints limit total renewable flows can continue indefinitely without any effect on
water availability over time. Peak nonrenewable water is observa- future availability. Still, although many flows of water are renew-

SUSTAINABILITY
ble in groundwater systems where production rates substantially able, some uses of water will degrade the quality to a point that

SCIENCE
exceed natural recharge rates and where overpumping or conta- constrains the kinds of use possible.
mination leads to a peak of production followed by a decline, In the past few years, various resource crises around water,
similar to more traditional peak-oil curves. Peak “ecological” water energy, and food have led to new debates over definitions and
is defined as the point beyond which the total costs of ecological concepts about sustainable resource management and use. Some
disruptions and damages exceed the total value provided by energy experts have proposed that the world is approaching, or
human use of that water. Despite uncertainties in quantifying has even passed, the point of maximum production of petroleum,
many of these costs and benefits in consistent ways, more and or peak oil (4–7). More recently, there has been a growing dis-
more watersheds appear to have already passed the point of peak cussion of whether we are also approaching a comparable point
water. Applying these concepts can help shift the way freshwater for water resources, where natural limits will constrain growing
resources are managed toward more productive, equitable, effi- populations and economic expansion. In this article, we define
cient, and sustainable use. the concept of peak water and we evaluate the similarities and
differences between water and oil, how relevant this idea is to
surface water ∣ water use ∣ sustainable water management actual hydrologic and water-management challenges, and the
implications of limits on freshwater availability for human and

T he Earth has substantial water resources, in numerous forms

and qualities, in various stocks and flows in the hydrologic
cycle. Overall, the planet has a stock of approximately 1.4 billion
ecosystem wellbeing.
Regional water scarcity is a significant and growing problem.
Many possible indicators have been developed to measure water
cubic kilometers of water, the vast majority of which (nearly 97%) scarcity, including both single-factor and weighted water mea-
is salt water in the oceans. The world’s more limited freshwater sures (8). The United Nations has offered a definition of water
stocks are estimated at around 35 million cubic kilometers. Most stress as regions where water consumption exceeds 10% of
fresh water, however, is locked up in glaciers in Antarctica and renewable freshwater resources. Other definitions set per-capita
Greenland, in permanent snow cover in mountains or high availability standards for defining scarcity (9–12). These kinds of
latitudes, or in deep groundwater inaccessible to humans for indicators inform decision making and offer insights into progress
practical reasons. Only small fractions are readily available to on addressing water problems, but no single measure can com-
humans in river flows, accessible surface lakes and groundwater, pletely describe the characteristics of water scarcity. Despite
soil moisture, or rainfall (1). Table 1 shows the distribution of the the lack of clear and specific measures of scarcity, it is increas-
main components of the world’s water. ingly apparent that some regions are experiencing limits to
Serious water challenges face humanity, including the failure growth in water use due to natural, ecological, political, or
to meet basic human needs for safe water and sanitation for economic constraints.
billions, growing contamination of water with human and indus-
trial wastes, the consequences of extreme events such as floods Concept of Peak Resource Production
and droughts, ecological disruption in aquatic ecosystems, in- The theory of peak resource production originated in the 1950s
creasing concerns about water shortages and scarcity, and the with the work of geologist M. King Hubbert and colleagues who
growing risks from climatic changes that will affect regional suggested that the rate of oil production would likely be
hydrology and water management. Considering the total volume characterized by several phases that follow a bell-shaped curve
of water on Earth, however, the concept of “running out” of (13). The first phase is the discovery and rapid increase in growth
water at the global scale is of little practical utility. There are huge in the rate of exploitation of oil as demand rises, production
volumes of water—many thousands of times the volumes that hu-
mans appropriate for all purposes. In the early 2000s, total global Author contributions: P.H.G. designed research; P.H.G. and M.P. performed research;
withdrawals of water were approximately 3;700 km3 per year, a P.H.G. and M.P. contributed new reagents/analytic tools; P.H.G. and M.P. analyzed data;
tiny fraction of the estimated stocks of fresh water (2). and P.H.G. and M.P. wrote the paper.
A more accurate way to evaluate human uses of water, how- The authors declare no conflict of interest.
ever, would look at regional stocks and flows of water and the 1
To whom correspondence should be addressed. E-mail: [email protected].

www.pnas.org/cgi/doi/10.1073/pnas.1004812107 PNAS ∣ June 22, 2010 ∣ vol. 107 ∣ no. 25 ∣ 11155–11162

 Table 1. Major stocks of water on Earth (34)
Distribution area, Volume, Percent of total Percent of
103 km2 103 km3 water, % fresh water, %
Total water 510,000 1.386 million 100 —
Total freshwater 149,000 35,000 2.53 100
World oceans 361,300 1.340 million 96.5 —
Saline groundwater — 13,000 1 —
Fresh groundwater — 10,500 0.76 30
Antarctic glaciers 13,980 21,600 1.56 61.7
Greenland glaciers 1,800 2,340 0.17 6.7
Arctic islands 226 84 0.006 0.24
Mountain glaciers 224 40.6 0.003 0.12
Ground ice/permafrost 21,000 300 0.022 0.86
Saline lakes 822 85.4 0.006 —
Freshwater lakes 1,240 91 0.007 0.26
Wetlands 2,680 11.5 0.0008 0.03
Rivers (as flows on average) — 2.12 0.0002 0.006
In biological matter — 1.12 0.0001 0.0003
In the atmosphere (on average) — 12.9 0.0001 0.04

becomes more efficient, and costs fall. Second, as stocks of oil are peak of production will only be identified in hindsight, and its
consumed and the resource becomes increasingly depleted, costs timing depends on the demand and cost of oil, the economics
rise and production levels off and peaks at a point now known as of technologies for extracting oil, the rate of discovery of new re-
peak oil. Finally, increasing scarcity and costs lead to a decline in serves compared to the rate of extraction, the cost of alternative
the rate of production more quickly than new supplies can be energy sources, and political factors.
found or produced. This last phase would also be typically accom-
panied by the substitution of alternatives. The phrase peak oil Comparison of Peak Production in Oil and Water
Does production or use of water follow a similar bell-shaped
refers to the point at which approximately half of the existing
curve? In the growing concern about global and local water
stock of petroleum has been depleted and the rate of production
shortages and scarcity, is the concept of peak water valid and use-
peaks (see Fig. 1). In a now-classic paper, Hubbert (1956)
ful to hydrologists, water planners, managers, and users? In the
predicted that oil production in the United States would peak
following sections, we consider the differences and similarities
between 1965 and 1970 (13). In 1970, oil production in the between oil and water to evaluate whether a peak in the produc-
United States reached a maximum and began to decline (Fig. 2). tion of water is possible, and in what contexts it may be relevant.
The concept of a roughly bell-shaped oil production curve has
been proven for a well, an oil field, a region, and is thought to Key Characteristics of Renewable and Nonrenewable Resources. In
hold true worldwide. any comparison among resources, it is vital to distinguish between
In recent years, the concept of peak oil has received renewed renewable and nonrenewable resources. The key difference be-
attention because of growing concern that the world as a whole is tween these is that renewable resources are flow or rate limited;
approaching the point of declining petroleum production. No one nonrenewable resources are stock limited (16). Stock-limited
knows when global oil production will actually peak, and forecasts resources, especially fossil fuels, can be depleted without being
of the date range from early in the 21st century to after 2025. One replenished on a timescale of practical interest. Stocks of oil, for
of many recent estimates suggests that oil production may peak as example, accumulated over millions of years; the volume of oil
early as 2012 at 100 million barrels of oil per day (15). The actual stocks is thus effectively independent of any natural rates of re-
plenishment because such rates are so slow. Conversely, renew-
able resources, such as solar energy, are virtually inexhaustible
over time, because their use does not diminish the production
Half of oil produced – Peak in of the next unit. Such resources are, however, limited by the flow
oil production rate, i.e., the amount available per unit time. Our use of solar
Annual Production of Oil

12000
Thousand barrels per day (average)

10000

8000

6000

4000

2000

0
1900
1905
1910
1915
1920
1925
1930
1935
1940
1945
1950
1955
1960
1965
1970
1975
1980
1985
1990
1995
2000
2005

Time
Fig. 1. There is no reason actual peak resource curves have to follow Fig. 2. Total annual US production of crude oil, 1900–2007. US production
symmetrical bell curves. peaked in 1970 (14, 33).

11156 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1004812107 Gleick and Palaniappan

 INAUGURAL ARTICLE
energy has no effect on the next amount produced by the sun, but poses. And like energy, the efficiency of water use can be greatly
our ability to capture solar energy is limited to the rate at which it improved by changes in technologies and processes. Unlike oil,
is delivered. however, fresh water is the only substance capable of meeting
Water demonstrates characteristics of both renewable and certain needs. Thus, although other energy sources can substitute
nonrenewable resources. This dual characteristic of water has im- for oil, water has no substitutes for most uses.
plications for the applicability of the term peak water. Water is When limits to water availability in a given region are reached,
largely a renewable resource with rapid flows from one stock there are a few possible options to meet additional needs: redu-
and form to another, and the human use of water typically has cing demand, substituting one use of water for another that has
no effect on natural recharge rates. But there are also fixed or
higher economic or social value, physically moving the demand
isolated stocks of local water resources that are being consumed
at rates far faster than natural rates of renewal. Most of these for water to a region where additional water is available; or in-
nonrenewable resources are groundwater aquifers—often called vesting in a higher priced source of supply, including bulk imports
“fossil” aquifers because of their slow-recharge rates. Tiwari et al. or transfers of water. In this case, the cost of new supply, inclu-
(17) recently calculated that a substantial fraction of water used ding the cost of transporting water, is a limiting factor.
in India comes from nonrenewable groundwater withdrawals, and A relevant concept to both peak water and peak oil, therefore,
that water ends up in the oceans, incrementally raising sea levels, is the introduction of a “backstop” technology when the price of
but substantially depleting groundwater stocks. Syed et al. (18) the resource rises, a concept given prominence by Nordhaus (19).
found similar transfers of nonrenewable groundwater for a wide As oil production peaks and then declines, the price of oil will rise
variety of groundwater basins using new data from the GRACE in the classic “supply/demand” economic response. Prices will
satellite. Some surface water storage in the form of lakes or gla- continue to rise until the point when a substitute, or backstop,
ciers can also be used in a nonrenewable way where consumption for oil becomes economically competitive, at which point prices
rates exceed natural renewal, a problem that may be worsened by will stabilize at the new backstop price. Nordhaus (19) noted that
climate change, as noted below. a backstop alternative is one capable of meeting the demand and

SUSTAINABILITY
that has a virtually infinite resource base.
Consumptive vs. Nonconsumptive Uses. Another key factor in Similarly, for water, as cheaper sources of water are depleted

SCIENCE
evaluating the utility of the concept of a resource peak is whether
or allocated, more and more expensive sources must be found
resource use is “consumptive” or “nonconsumptive.” Practically
and brought to the user, either from new supplies or reallocation
every use of petroleum is consumptive; once the energy is ex-
tracted and used, it is degraded in quality.* Almost every year, of water among existing users. Fig. 3 graphs a potential water-
the amount of oil consumed matches the amount of oil produced, production scenario in a watershed, where incremental supply
and sometimes we consume more than is produced that year. increases through supply side projects, e.g., groundwater harvest-
Thus a production curve for oil is solely dependent on access ing, in-stream flow allocation, and reservoir construction are
to new oil. layered upon each other until the maximum cost-effective extrac-
Not all uses of water are consumptive and even water that has tion of surface and groundwater is reached.
been “consumed” is not lost to the hydrologic cycle or to future Ultimately, the backstop price for water will also be reached.
use—it is recycled by natural systems. Consumptive use of water Unlike oil, however, which must be backstopped by a different,
typically refers to uses that make that water unavailable for renewable energy source, the ultimate water backstop is still
immediate or short-term reuse within the same watershed. Such water, from an essentially unlimited source—for example,
consumptive uses include water that has been evaporated, tran- desalination of ocean water. The amount of water in the oceans
spired, incorporated into products or crops, heavily conta- that humans can use is limited only by how much we are willing to
minated, or consumed by humans or animals. As discussed in pay to remove salts and transport it to the point of use, and by the
the section on the renewability of water resources, some stocks environmental constraints of using it. In some regions, desali-
of water can be consumed locally, making them, effectively, non-
nation is already an economically competitive alternative, parti-
renewable resources. When withdrawals are not replaced on a
timescale of interest to society, eventually that stock becomes cularly where water is scarce compared to demand, such as
depleted. The water itself remains in the hydrologic cycle, in certain islands in the Caribbean and parts of the Persian
another stock or flow, but it is no longer available for use in Gulf (20, 21).
the region originally found. There are also many nonconsumptive
uses of water, including water used for cooling in industrial and
Production of Water from Various Sources

energy production, and water used for washing, flushing, or other

residential uses if that water can be collected, treated, and reused. Shift to higher cost
This water recycles into the overall hydrological cycle and has no sources of water
effect on subsequent water availability in a region. (e.g., desalination,
water transfers) to
Incremental meet increased
Substitutability. Another important characteristic of peak re- increases in supply water needs
source discussions is the potential to substitute alternatives for from various sources
Maximum cost-
nonrenewable sources. As oil production declines and prices in- effective extraction
crease, substitutes that offer the same benefits become increas- of surface and
ingly attractive. Oil serves particular functions in industrial ground water
reached
society that can be satisfied by other means or resources (solar,
natural gas, biofuels, etc.). In this sense, any depletable resource
such as fossil fuels must be considered a transition option, useful
only as long as its availability falls within economic and environ- Time
mental limits. Like energy, water is used for a wide variety of pur-
Fig. 3. A potential water-production (supply) scenario in a watershed or
region. As demand increases, incremental supply projects (new dams, reser-
*Due to the law of conservation of energy, energy is never consumed—simply converted voirs, pumping) increase water availability. Once the maximum cost-effective
to another form. But in this case, the use of oil converts concentrated, high-quality extraction of surface and groundwater is reached, there is a final shift to a
energy into low-quality, unusable waste heat, effectively “consuming” the oil. higher cost backstop supply of water such as desalination or water transfers.

Gleick and Palaniappan PNAS ∣ June 22, 2010 ∣ vol. 107 ∣ no. 25 ∣ 11157
 Table 2. Summary comparison of oil and water
Characteristic Oil Water
Quantity of resource Finite Literally finite, but practically unlimited at a cost
Renewable or nonrenewable Nonrenewable resource Renewable overall, but with locally
nonrenewable stocks
Flow Only as withdrawals from fixed stocks Water cycle renews natural flows
Transportability Long-distance transport is economically viable Long-distance transport is not economically viable
Consumptive versus Almost all use of petroleum is consumptive, converting Some uses of water are consumptive, but many
nonconsumptive use high-quality fuel into lower-quality heat are not; overall, water is not consumed from
the hydrologic cycle
Substitutability The energy provided by the combustion of oil can be Water has no substitute for a wide range of
provided by a wide range of alternatives functions and purposes
Future prospects Limited availability; substitution inevitable by a backstop Locally limited, but globally unlimited after
renewable source backstop source (e.g., desalination of oceans) is
economically and environmentally developed

Transportability. The concept of running out of water at the global of the economic limits to transport and total volumes moved any
scale is of little practical use. Huge volumes of water over and significant distance are a tiny fraction of total urban or agricul-
above the volumes used by humans for all purposes are distri- tural demands.
buted around the world in various stocks. Because the Earth will
never “run out” of fresh water, concerns about water scarcity Three Peak Water Concepts
must, therefore, be the result of something other than a fear that Table 2 summarizes the characteristics of oil and water discussed
we are literally consuming a limited resource. And, of course, above. Given the physical and economic characteristics of
they are: Water challenges are the result of the tremendously un- resources reviewed above, how relevant or useful is the concept
even geographic distribution of water (due to both natural and of a peak in the production of water? We offer here three defini-
human factors), the economic and physical constraints on tapping tions where the concept of a peak is useful in the context of water
some of the largest volumes of freshwater (such as deep ground- resources and we introduce a term that is useful when thinking
water and ice in Antarctica and Greenland), human contamina- about maximizing the multiple services that water provides: “peak
tion of some readily available stocks, and the high costs of moving ecological water.” These peak water concepts should help drive
water from one place to another. important paradigm shifts in how water is used and managed.
This last point—the “transportability” of water—is particularly
relevant to the concept of peak water. Oil is transported around the Peak Renewable Water. A significant, albeit poorly quantified
world because it has a high economic value compared to the cost of fraction of total human use of water comes from water taken from
transportation. For example, one of today’s supertankers carries as renewable flows of rainfall, rivers, streams, and groundwater
much as 3.6 million barrels of oil. At a price of $70 per barrel, that basins that are recharged over relatively short time frames. Such
oil is worth over $250 million dollars and the cost of transportation systems experience stochastic hydrology, but use of water does
is minor. As a result, regional limits on oil availability can be over- not affect the ultimate renewability of the resource, much like
come by moving oil from any point of production to any point of solar energy use. Because a particular water source may be re-
use. In contrast, water is very expensive to move any large distance, newable, however, does not mean that it is unlimited. Indeed,
compared to its value. That same supertanker filled with fresh- the first peak water constraint is the limit on total water that
water would have an economic value of only around $500,000†— can be withdrawn from a system. The ultimate limit is the com-
far too little to support long-distance shipping and regional plete renewable flow.
constraints become a legitimate and serious concern. As shown in Fig. 4, when the production of renewable water
As a result, media attention to the concept of peak water has from a watershed reaches 100% of renewable supply, it forms a
focused on local water scarcity and challenges, for good reason. classic logistics curve, similar to a biological carrying capacity
But there has been little or no academic research or analysis on this
concept. In regions where water is scarce, the apparent nature of
water constraints—and hence, some of the real implications of a
“peak” in availability—are already apparent. Because the costs
Production of Renewable Water Per Year

Total annual renewable flow in a watershed

of transporting bulk water from one place to another are so high,
once a region’s water use exceeds its renewable supply, it will begin
tapping into nonrenewable resources, such as slow-recharge
aquifers. Once extraction of water exceeds natural rates of reple-
nishment, the only long-term options are to reduce demand to sus-
tainable levels, move the demand to an area where water is avail-
able, or to shift to increasingly expensive sources, such as
desalination or imports of goods produced in regions with adequate
water supplies, the transfer of so-called virtual water (22).
There are very few exceptions to the economic limits on trans-
porting water. Bottled water is sometimes consumed vast
distances from where it was produced only because it commands
a premium far above normal costs. Growth in bottled water con-
sumption may expand in some markets, but overall, bulk water is Time
not currently a significant export in commercial markets because
Fig. 4. Theoretical logistics curve showing increasing annual production of
renewable water from a watershed. Annual renewable water production
†
Assuming a price equivalent to what industry and urban users pay for high-quality increases exponentially and then levels off as it reaches the total annual
reliable municipal supplies. renewable water supply in the watershed.

11158 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1004812107 Gleick and Palaniappan

 INAUGURAL ARTICLE
model. Each watershed only has a certain amount of renewable drop, leading to a peak of production, followed by diminishing
water supply that is replenished every year. If the annual produc- withdrawals and use. This kind of unsustainable groundwater
tion of renewable water from a watershed began to increase ex- use can be found in the Ogallala Aquifer in the Great Plains
ponentially, it approaches the natural limit of the total annual of the United States, the North China plains, California’s Central
renewable supply of water (shown as a dashed line). This limit Valley, and numerous basins in India (23). Tiwari et al. (24)
varies, of course, with natural variation in hydrology, but it is estimate that nonrenewable use of water in India averaged 54 
the ultimate limit in terms of appropriation of renewable water 9 km3 per year between 2002 and 2008 or around 8% of India’s
supply. The appropriate practical limit may be substantially less total water withdrawals.
than this, as discussed below under Peak Ecological Water. As shown in Fig. 6, even when the rate of withdrawals from a
Increasing annual renewable water use to the theoretical groundwater aquifer passes the natural recharge rate for the
renewable limit has been shown to result in tremendous ecologi- aquifer (shown as a dashed line), the production of water from
cal, environmental, and human damage. the aquifer can continue to increase until a significant portion of
For a number of major river basins, peak renewable water the groundwater has been harvested. After this point, deeper
limits have already been reached as human demand consumes boreholes and increased pumping will be required to harvest
close to the entire annual supply. The Colorado River in the Uni- the remaining amount of water, potentially reducing the rate
ted States, for example, is shared by seven US states and Mexico, of production of water.
and in an average year little or no water reaches the delta (see It is also possible that the production of water from the aquifer
Fig. 5). For this watershed, the limit of peak renewable water is an will continue to increase until all the economically affordable
average of around 18 billion cubic meters annually—the total groundwater is harvested, after which the production of water
average annual flow. Other rivers are increasingly reaching their drops quickly. In both these cases, the important point is that ex-
peak limits as well, including the Huang He (Yellow River) in traction will not fall to zero, but to the renewable recharge rate
China, the Nile in Northern Africa, and the Jordan in the Middle where economically and physically sustainable pumping is possible.
East, where formerly perennial river flows now often fall to zero. In some places, climate change will affect the nature and
Under other circumstances, as noted below, peak renewable magnitude of peak water. Where local communities are currently

SUSTAINABILITY
water systems can sometimes be turned into nonrenewable dependent on river runoff from glacier melt, the loss of glaciers in

SCIENCE
systems through physical or chemical processes. coming years will lead to a “peak nonrenewable water” effect: the
diminishment of water supply over time. Communities dependent
Peak Nonrenewable Water. In some watersheds, a substantial on groundwater recharge that suffer a decrease in recharge rate
amount of current water use comes from stocks of water that will also experience an effect akin to peak water. In this case, the
are effectively nonrenewable, such as groundwater aquifers with concept of peak water is slightly different: It is not affected by the
very slow-recharge rates or groundwater systems that lose their magnitude of human use, but by physical or climatic factors that
ability to be recharged when overpumped due to compaction diminish the rate of, or potential for, replenishment. Similar to
or other physical changes in the basin. When the use of water peak oil, however, when the stock is gone, alternative sources will
from a groundwater aquifer far exceeds the natural recharge rate, have to be found.
this stock of groundwater will be quickly depleted. Or when
Peak Ecological Water. For many watersheds, a more immediate
groundwater aquifers become contaminated with pollutants that
make the water unusable, a renewable aquifer can become non- and serious concern than running out of water is exceeding a
renewable. point of water use that causes serious or irreversible ecological
In these particular situations, the groundwater aquifer is clo- damage. Water provides many services: Not only does it sustain
sely analogous to an oil field or oil-producing region. Continued human life and commercial and industrial activity, but it is also
fundamental for the sustenance for animals, plants, habitats, and
production of water beyond natural recharge rates will become
environmentally dependent livelihoods (25, 26, 27).
increasingly difficult and expensive as groundwater levels
Each new incremental supply project that captures water for
human use and consumption decreases the availability of that
source to support ecosystems and diminishes the capacity to pro-
vide services. The water that has been temporarily appropriated
Production of Water from Groundwater Aquifer

Peak in production of fossil groundwater

from aquifer

Natural Groundwater Recharge Rate

Fig. 5. Annual flows (in million cubic meters) of the Colorado River into the
delta from 1905 to 2005 at the Southern International Border station. Note
Time
that, in most years after 1960, flows to the delta fell to zero as total
withdrawals equaled total (or peak) renewable supply. The exceptions are Fig. 6. This theoretical curve shows the progression of unsustainable
extremely high-flow years when runoff exceeded demands and the ability water extraction from a groundwater aquifer, hypothesizing a peak-type
to store additional water (International Boundary Waters Commission data production curve for water after the production rates surpass the natural
on Colorado River flows at the Southern International Border. Water data groundwater recharge rate and production costs rise. Long-term sustainable
from http://www.ibwc.state.gov/wad/DDQSIBCO.htm). withdrawals cannot exceed natural recharge rates.

Gleick and Palaniappan PNAS ∣ June 22, 2010 ∣ vol. 107 ∣ no. 25 ∣ 11159
or moved was once sustaining habitats and terrestrial, avian, and
aquatic plants and animals. By some estimates, humans already

Total Social and Ecological Value of Water

appropriate almost 50% of all renewable and accessible fresh-
water flows (28), leading to significant ecological disruptions.
Since 1900, half of the world’s wetlands have disappeared
(29). The number of freshwater species has decreased by 50%
since 1970, faster than the decline of species on land or in the
sea. River deltas are increasingly deprived of flows due to
upstream diversions, or receive water heavily contaminated with
human and industrial wastes.
Fig. 7 is a simplified graph of the value that humans obtain
from water produced through incremental increases in supply
(e.g., drinking water and irrigation), plotted against the declining Point of “Peak Ecological Water” –
Maximized combined benefits to
value of the ecological services (e.g., water for plants and ani- human society and ecosystems
mals) that were being satisfied with this water. The graph assumes
that ecological services decrease as water is appropriated from
watersheds (though in nature such declines may be nonlinear). Water Appropriations by Humans
The pace or severity of ecological disruptions increases as in-
creasing amounts of water are appropriated. Because ecological Fig. 8. This graph charts the overall value of water, a combination of social,
services are not easily valued in dollar terms, the y axis should economic, and ecological value, as water appropriation by humans increases.
be considered the overall (economic and noneconomic) “value The value increases to a peak, where benefits to society and ecosystems is
provided by water.” maximized, but then declines as increased appropriations lead to excessive
ecosystem and social costs. Nonmonetary costs and benefits are hard to
At a certain point, the value of ecological services provided by
quantify, but must be included to avoid exceeding the point of peak eco-
water is equivalent to the value of human services satisfied by that logical water.
same use of water. After this point, increasing appropriation of
water leads to ecological disruptions beyond the value that this
increased water provides to humans (the slope of the decline accurately measuring water withdrawals from countless diverse
in ecological services is greater than the slope of the increase sources to meet agricultural, industrial, commercial, and domes-
in value to humans). We define this point to be peak ecological tic needs. As a result, identifying peak water limits will be diffi-
water—where society will maximize the total ecological and cult. Nevertheless, there is some strong evidence that the United
human benefits provided by water. As shown in Fig. 8, the overall States may have already passed the point of peak water, including
value of water, combining ecological and social benefits, then peak renewable, nonrenewable, and ecological water. Fig. 9
declines as human appropriation increases. Economists and re- shows US gross domestic product (in 2005 dollars) plotted with
source analysts have long noted the difficulty of quantifying this total water withdrawals in the United States, for all purposes,
point because of problems in assigning appropriate valuations to from 1900 to 2005, based on data from state and federal water
each unit of water or each unit of ecosystem benefit in any agencies, compiled largely by the US Geological Survey’s water
watershed (30). But the mistaken assumption that such values use assessments (31). These two curves grew exponentially, in
are zero has led to them being highly discounted, underappre- lockstep, through the first three-quarters of the 20th century.
ciated, or ignored in 20th century water policy decisions. After the late 1970s, however, the two curves split apart, and total
water withdrawals in the United States are now below their
Peak Water in the United States maximum level. Per-capita water withdrawals have fallen even
Data on total water use is sparse. Few countries or regions collect more, as population has also continued to grow. Some of the
such data because of the physical or political difficulties of reasons for this dramatic change include improving efficiency
of water use, changes in the structure of the US economy, the
implementation of the Clean Water Act, which led to reductions
in industrial water use and discharges, and physical, economic,
Value of ecological
and environmental constraints on access to new supplies. Some
services provided of the reasons for this change are explored in more detail else-
where (32). But the graph suggests that the United States may
Value Provided by Water

by water
well be past the point of peak water.
Value of “human
Whether this change is permanent or temporary is unknown.
services” provided In theory, US production of oil could increase again and even
by water exceed, for a while, the previous peak, although resource, eco-
nomic, and environmental constraints make this unlikely. Simi-
“Peak ecological water” larly, total water withdrawals could certainly begin to increase
again, but many factors suggest this is unlikely in the long run.
Significant expansion of irrigated agriculture, which dominates
US water use, seems improbable, especially in the western US
where almost all major rivers and aquifers are already tapped
Water Appropriation by Humans out—at the limits of their renewable and nonrenewable supplies.
Another major driver of US freshwater use is power plant
Fig. 7. This graph charts the value of water provided by increasing supply cooling, and a significant expansion of cooling demand also seems
from various sources in a watershed against the loss in value of ecological
unlikely because of constraints on water withdrawals, even in
services provided by that water. As water withdrawals for human needs
increase (solid line), the ecological services provided by same water are in
relatively well-watered regions, and because efforts to move from
decline (dashed line). At a certain point, the value of water provided through central water-intensive thermal plants to less-water-intensive
new supply projects is equal to the value of the ecological services. Beyond renewable systems are gaining traction. Certainly, some regions
this point, ecological disruptions exceed the benefits of increased water have passed the point of peak ecological water and efforts are
extraction. We call this point peak ecological water (see Fig. 8). now focused on how to restore some water for the environment,

11160 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1004812107 Gleick and Palaniappan

 INAUGURAL ARTICLE
SUSTAINABILITY
SCIENCE
Fig. 9. US gross domestic product (GDP) in 2005 dollars from 1900 to 2005 (left axis) plotted with total water withdrawals for all purposes in cubic kilometers
per year (right axis). Data on GDP come from the US Bureau of Economic Analysis; data on water use comes from the US Geological Survey (31).

not how to take more out. In short, we think it possible that the the value provided by additional increments of water use by hu-
United States is past the point of peak water and the focus of mans for economic purposes. Defined this way, many regions of
future water use efforts will be on improving efficiency of current the world have already surpassed peak ecological water—humans
uses and reallocating water from one existing user to another. use more water than the ecosystem can sustain without significant
deterioration and degradation.
Conclusions: Implications of Peak Water The concepts around peak water are also important in driving
As the world anticipates a resource-constrained future, the
some paradigm shifts in the use and management of water. There
specter of peak oil—a peaking in the production of oil—has been
predicted. Real limits on water are far more worrisome, and far are growing efforts to quantify peak ecological limits and to
more difficult to evaluate, than limits on traditional nonrenew- develop policies to restore water for ecosystem services in basins
able resources such as petroleum. Water is fundamental for eco- where serious ecological disruptions have already been reco-
system health and for economic productivity, and for many uses it gnized. Improvements in the ability to identify groundwater
has no substitutes. This paper offers three separate definitions of basins suffering from nonrenewable withdrawals are increasing
peak water, for renewable and nonrenewable water systems and the pressure on water managers to reduce withdrawals to more
also introduces the concept of peak ecological water. We also sustainable levels, or to better integrate surface and groundwater
raise the possibility that the United States has already passed management. And the realization that there are limits to peak
the point of peak water. renewable water use are forcing new discussions about improving
The concept of peak water does not mean we will run out of water-use efficiency and developing innovative technologies for
water. Water is a renewable resource and is not consumed in the water treatment and reuse as alternatives to expanding tradi-
global sense: Hence, water uses within renewable peak limits can tional supply projects to further mine overtapped renewable
continue indefinitely. But not all water use is renewable; indeed water sources. The bad news is that we are increasingly reaching
some water uses are nonrenewable and unsustainable. Ground- peak water limits. The good news is that recognizing and under-
water use beyond normal recharge rates follows a peak-oil type standing these limits can stimulate innovations and behaviors that
curve with a peak and then decline in water production. Such
can reduce water use and increase the productivity of water,
peak nonrenewable water problems are increasingly evident in
shifting water policy toward a more sustainable water future.
major groundwater basins with critical levels of overdraft, such
as the Ogallala and California’s Central Valley in the United
ACKNOWLEDGMENTS. Thanks to our colleagues at the Pacific Institute, includ-
States, the North China Plains, and in numerous states in India, ing Heather Cooley and Nancy Ross, for comments and feedback as we were
such as Andhra Pradesh, Rajasthan, and Tamil Nadu. Peak eco- preparing this analysis. Thanks also to Professors Robert Glennon and Robert
logical water refers to the point after which the cost of disruptions Wilkinson for reviewing the manuscript and offering thoughtful and helpful
that occur in the ecological services that water provides exceeds suggestions for improving it.

1. Shiklomanov IA (2000) Appraisal and assessment of world water resources. Water Int 3. Postel SL, Daily GC, Ehrlich PR (1996) Human appropriation of renewable fresh water.
25(1):11–32. Science 271:785–788.
2. Gleick PH (2006) The World’s Water 2006–2007 (Island Press, Washington, DC), Data 4. Bentley RW (2002) Global oil and gas depletion: An overview. Energ Policy 30
Table 2, pp 228–236. (3):189–205.

Gleick and Palaniappan PNAS ∣ June 22, 2010 ∣ vol. 107 ∣ no. 25 ∣ 11161
 5. Duncan RC (2003) Three world oil forecasts predict peak oil production. Oil Gas J 101 20. Cooley H, Gleick PH, Wolff G (2006) Desalination: With a Grain of Salt A California
(21):18–20. Perspective (Pacific Inst, Oakland, CA).
6. Kerr RA (2007) The looming oil crisis could arrive uncomfortably soon. Science 316:351. 21. National Research Council (2008) Desalination: A National Perspective (National
7. Bardi U (2009) Peak oil: The four stages of a new idea. Energy 34(3):323–326. Academy Press, Washington, DC).
8. Gleick PH (2002) Soft water paths. Nature 418:373. 22. Allan JA (1999) Water in International Systems: A Risk Society Analysis of Regional
9. Falkenmark M, Lundqvist J, Widstrand C (1989) Macro-scale water scarcity requires Problemsheds and Global Hydrologies. SOAS Occasional Paper 22 (School of Oriental
micro-scale approaches: Aspects of vulnerability in semi-arid development. Nat Resour and African Studies, University of London, London).
Forum 13(4):258–267. 23. Chatterjee R, Purohit RR (2009) Estimation of replenishable groundwater resources of
10. Lundqvist J, Gleick P (1997) Comprehensive Assessment of the Freshwater Resources of India and their status of utilization. Curr Sci India 96(12):1581–1591.
the World. Sustaining Our Waters into the 21st Century (United Nations Educational, 24. Tiwari VM, Wahr J, Swenson S (2009) Dwindling groundwater resources in northern
Scientific, and Cultural Organization, New York), pp 1–51. India, from satellite gravity observations. Geophys Res Lett 36:L18401 doi: 10.1029/
11. United Nations World Water Assessment Program (2003) World Water Development 2009GL039401.
Report: Water for People, Water for Life (United Nations World Water Assessment
25. Daily GC, et al. (1997) Iss Ecol 1:1–18.
Program and Berghahn Books, Barcelona), pp 1–575.
26. Daily GC, et al. (2000) The value of nature and the nature of value. Science
12. United Nations Environment Programme (2007) Global Environmental Outlook 4:
289:395–396.
Environment for Development (United Nations Environment Programme, Malta),
27. Gleick PH (1998) Water in crisis: Paths to sustainable water use. Ecol Appl 8(3):571–579.
pp 1–540.
28. Postel SL, Daily GC, Ehrlich PR (1996) Human appropriation of renewable fresh water.
13. Hubbert MK (1956) Nuclear Energy and the Fossil Fuels (Shell Development Co.,
Science 271:785–788.
Houston), pp 1–44.
29. Katz D (2006) The World’s Water 2006–2007, ed PH Gleick (Island Press, Washington,
14. United States Energy Information Agency (2008) International Petroleum Monthly
http://www.eia.doe.gov/ipm/. DC), pp 29–39.
15. Gold R, Davis A ( 10, 2007) Oil Officials See Limit Looming on Production. Wall Street 30. Daily GC, et al. (2000) The value of nature and the nature of value. Science
Journal p A-1. 289:395–396.
16. Ehrlich P, Ehrlich A, Holdren JP (1977) Ecoscience: Population, Resources, Environment 31. Kenny JF, et al. (2009) Estimated Use of Water in the United States in 2005 (United
(Freeman, San Francisco). States Geological Survey, Reston, VA) Circular 1344 (see also the entire series for earlier
17. Tiwari VM, Wahr J, Swenson S (2009) Dwindling groundwater resources in northern years).
India, from satellite gravity observations. Geophys Res Lett 36:L18401 doi: 10.1029/ 32. Gleick PH (2003) Water use. Annu Rev Env Resour 28:275–314.
2009GL039401. 33. United States Energy Information Agency (2009) US Field Production of Crude Oil,
18. Syed TH, Famiglietti J, Chambers DP (2009) GRACE-based estimates of terrestrial fresh- Independent Statistics and Analysis. (United States Energy Information Agency,
water discharge from basin to continental scales. J Hydrometeorol 10(1):22–40. Washington, DC).
19. Nordhaus WD (1973) The allocation of energy resources. Brookings Pap Eco Ac 1973 34. Shiklomanov IA (1993) World Fresh Water Resources. Water in Crisis, ed PH Gleick
(3):529–570. (Oxford Univ Press, Oxford), pp 13–24.

11162 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1004812107 Gleick and Palaniappan