Estimating Water Use in the United States: A New Paradigm for the National Water-Use Information Program (2002)

Among the components of the hydrologic cycle, water use has unique qualities. Foremost among these is that water use has an important social science component. Water use science, therefore, must encompass and integrate both the social and natural sciences. We use the term “integrative water use science” to describe this multidisciplinary, hypothesis-driven investigation of the behaviors and phenomena that determine spatial and temporal patterns of water use, and of the impacts of water use on aquatic ecosystems, the hydrologic cycle, and the sustainability and vulnerability of the nation’s water resources.

Water use is also an element of the geospatial data layer. In the past, the nation has not given water use data the importance it deserves; now, a reliable water use dataset can be developed with current technology. This was clearly illustrated in Chapters 3 and 7, which describe the emergence of site-specific water use databases and give many examples of how such a database, linked to a geographic information system (GIS), could illuminate the relations of water use to land use and hydrologic characteristics.

In this chapter, we expand upon the concept of integrative water use science, initially focusing on two very specific aspects of water use science that have not previously been addressed in the NWUIP: (1) the interrelationships among water use, water flow, land use, and water quality and (2) the estimation of instream flow for ecological needs. We argue that the NWUIP should consider expanding into these very important areas in collaboration with other U.S. Geological Survey (USGS) programs (e.g., National Water-Quality Assessment Program [NAWQA], National Streamflow Information Program [NSIP], and the Biological Resources Division), where the USGS can bring its many talents to bear upon a revitalized approach to water use science.

We conclude by presenting a holistic methodology for the NWUIP, primarily developed by Dr. Gregory E. Schwarz of the USGS, Branch of Systems Analysis. This approach would utilize the surface water quality model SPARROW (SPAtially Referenced Regressions On Watershed attributes) (Smith et al., 1997) to provide a framework for organizing and analyzing source data, a means for assessing data accuracy, and information on areas where data needs are most pressing.

All uses of water affect its quality (i.e., its physical, chemical, and biological characteristics) (Getches et al., 1991). The quality of water determines what it can be used for; likewise, how water is used will change certain aspects of its quality. An example of the latter includes the increase in total dissolved solids (TDS) of irrigation water as it is diverted from a stream or reservoir or pumped from an aquifer, is applied to the land surface, undergoes evapotranspiration, leaches salt from the soil, and emerges as irrigation return flow. Water diverted for municipal consumption also generally suffers reduction in quality, as the water discharged from a wastewater treatment plant is generally “degraded” relative to the original diversion. Aquifer storage and recovery, an increasingly common water management practice, may lead to increased concentrations of arsenic, methyl mercury, and other metals (NRC, 2001). Indeed, Getches et al. (1991) stated that control of water use is one approach to protecting water quality and that most uncontrolled water-quality degradation relates to water uses authorized by state water allocation systems. Along these lines, it is interesting to note that return flow from irrigated agriculture, a major user of water, especially in the western United States, is specifically exempted as a nonpoint source of pollution under the Clean Water Act.

Water-quality degradation from use falls into four categories (Getches et al., 1991):

1. Depletion degradation. The consumption of water results in a higher concentration of pollutants because the remaining water is less able to dilute them (e.g., consumptive use of a stream diversion reduces the stream’s ability to dilute pollution).

2. Physical alteration. Some uses of water directly alter the physical characteristics of the water (e.g., storage in a reservoir changes temperature, dissolved oxygen content, etc.).

3. Pollution migration. Water use causes preexisting pollution to contaminate other waters (e.g., pumping from an aquifer may induce the migration of

contaminated water into the aquifer; diversions from a stream may cause salt-water encroachment).

4. Incidental pollution. Water use causes pollutants to enter waterways other than from discrete point sources (e.g., irrigation water leaches salts from soils, causing the salts to enter surface waters and groundwaters as part of irrigation return flows).

Water use is also linked to land use. Different land uses will dictate different degrees of water use. Even within a given land use—agriculture, for example—water use can vary dramatically (see Table 3.1). Site-specific water use databases (see Chapter 7), such as the Arkansas database, will enable the Naitonal Water-Use Information Program (NWUIP) to assess the interrelationship between land use and water use.

The above discussion suggests that water use, land use, water flow, and water quality, although often considered separate from one another, are inextricably linked. They should not be treated independently of the other, although this is often the case insofar as water allocations are concerned. Box 8.1 provides an illustrative example from one of the nation’s most heavily allocated basins—the Colorado River basin of the western United States.

The current NWUIP deals with use and flows only. The usefulness of the NWUIP would be increased were it to consider water quality. The USGS could lead the way in the integration of land use, water use, water flow, and water quality data by expanding the current NWUIP. But given the budget constraints of the present, how can water quality be integrated into the NWUIP short of calling for an expensive program to monitor our nation’s waters? Fortunately, there are a number of current programs with which the NWUIP might collaborate. A few of the more visible programs are described in the next section.

Water-quality data may come from many sources, both within and outside of the USGS. These sources include the USGS National Water Quality Assessment (NAWQA) and National Stream Quality Accounting Network (NASQAN) Programs, the U.S. Environmental Protection Agency (EPA) the U.S. Bureau of Reclamation (USBR) and the U.S. Army Corps of Engineers (ACE).

The NAWQA Program (http://water.usgs.gov/nawqa; see also Gilliom et al., 1995; Hirsch et al., 1988; Leahy et al., 1990) is an ambitious program of the USGS designed to describe the status and trends in the quality of the nation’s groundwaters and surface waters and to understand the natural and anthropogenic

factors affecting their status. Water-quality investigations are being conducted in 59 “study units” (SUs), which cover about half the land area of the United States and about 60 percent to 70 percent of the nation’s water use and population served by public water supplies.

The NAWQA Program began as a pilot program of 7 study units in 1988. In 1991, the number of SUs was increased to 20, and the program reached its full complement of 60 SUs in 1997 (subsequently reduced to 59 by combining two adjacent SUs in New England). The SUs represent portions of major river basins and aquifer systems. Budget limitations will necessitate a cut in the number of SUs to 42 (this is the estimate as of August 2001) as the NAWQA Program enters its Cycle II (second 10 years) phase in 2001. Even with the planned reductions, the NAWQA Program will still cover about 40 percent of the nation’s land area, comprising about 60 percent of its drinking water use. The NAWQA Program is far more than just a data collection/monitoring program, as it seeks to establish cause-and-effect relationships and quantify the effects of land use on water quality. It also has a “national synthesis” component; a report on nutrients and pesticides (USGS, 1999) has already been produced; the next reports will be on VOCs (volatile organic compounds) and trace elements. Because the NAWQA Program does not cover the entire nation, any of its water-quality information that would be used in an expanded NWUIP would have to be augmented by data from other programs.

Although much of the existing water use data are organized by political unit, it may be feasible in some cases for the NAWQA study units themselves to serve as modeling units to develop the science behind water use estimates. Statistical estimation models for water use could be developed and tested. Sampling protocols could be examined where land use information, water withdrawal, and water discharge points are usually known. The NAWQA study units could effectively become water use investigation laboratories for developing the science behind the water use estimates.

Although NAWQA activities alone could not provide the data required for a redesigned NWUIP, NAWQA’s overall themes of water-quality status, trends, and understanding fit well with the NWUIP’s objective of providing information on status of and trends of the nation’s water use. The NAWQA Program was originally conceived to determine if the nation’s water quality was improving and whether all the money that had been spent cleaning up the nation’s waters had had any effect. It would be well worthwhile to link the NAWQA and NWUIP programs, as they both deal with important aspects of water—its use (and flows) and quality. An expanded NWUIP could provide a snapshot of the nation’s water use and quality every five years, augmenting the numerous NAWQA reports and complementing NAWQA by assessing the relationships between land use and water use.

The USGS NASQAN Program has been in effect since 1973 (http://water.usgs.gov/nasqan/), collecting water-quality information at over 500 locations. Such a suite of data would fit nicely with water use data. However, since

1995, the program has been cut drastically and now monitors only 39 sites in four major river basins: the Columbia, Mississippi-Missouri, Colorado, and Rio Grande. Barring a restoration of this program, the current NASQAN Program could only provide a small set of streamflow quality measurements.

The EPA is another organization concerned with water quality and has programs that could provide water-quality data to an expanded NWUIP. A number of EPA-sponsored programs collect water-quality data; only two will be mentioned here. The Environmental Monitoring and Assessment Program (EMAP) is monitoring and assessing water quality as part of its mission (http://www.epa.gov/emap/). The National Water Quality Inventory, submitted to each Congress under Section 305(b) of the Clean Water Act, contains a wealth of information on surface waters and groundwaters (U.S. EPA, 2000). The inventory is based upon data collected by the states and Indian tribes; these Section 305(b) quality data could be obtained by NWUIP and integrated into its reports. Because NWUIP already relies on state cooperation and data for its work, its use of additional state data would be appropriate.

Both the USBR and the ACE provide water for drinking, recreation, irrigation, and aquatic life, primarily through reservoirs they construct and operate. These agencies perform water-quality monitoring for a variety of uses: aquatic life, fish consumption, primary contact, secondary contact, drinking water supply, and agriculture.

The USBR is currently concerned with the quality of irrigation water supplied by its projects through its National Irrigation Water Quality Program (http://www.usbr.gov/niwqp/). Because irrigation return flow often empties into reservoirs—e.g., Kesterson Reservoir in the San Joaquin Valley, whose wildlife has suffered from selenium poisoning (NRC, 1989)—the USBR is monitoring quality in a number of its reservoirs. These USBR-collected data could be integrated into NWUIP’s work.

Some of the ACE districts/regions have very active reservoir water-quality monitoring programs. The Northwestern Division of the Corps’ Missouri River Region (Omaha and Kansas City Districts) has its Water Quality Management Program–Missouri River Region Lake Projects. The Omaha District alone is conducting water-quality studies in over 30 lakes and reservoirs. The NWUIP could use these data in its effort to characterize the nation’s water quality.

Numerous other agencies have water-quality data that may be made available for the NWUIP. Prominent among these are the U.S. Department of Agriculture (USDA), U.S. Forest Service, the National Park Service (NPS), the Bureau of Land Management (BLM), and the National Oceanic and Atmospheric Administration (NOAA).

An expanded (i.e., water use and water quality) NWUIP and the NAWQA Program would have much common ground for collaboration. The NAWQA Program realizes the interconnection between use (land and water) and quality in its effort to understand cause-and-effect vis-à-vis water quality. Further, it requires water use data: the NAWQA Program needs specific public supply and domestic water use information and will be affected by the elimination of some categories in the 2000 NWUIP report (M. Maupin, USGS, personal communication, 2000).

An excellent example of a NAWQA project where water use and water quality are interconnected and integrated is the Santa Ana Basin SU in Southern California (http://water.wr.usgs.gov/sana_nawqa/). Although this study unit is rather small (2,700 square miles) with a highly “engineered” hydrologic cycle and does have a number of agencies collecting both use and quality data, the interdependence of the two types of data is clearly indicated and appreciated. Perhaps this study unit could be used as a model for similar water use–water quality studies.

In addition, the SPARROW model (Smith et al., 1997), which has been used extensively in the NAWQA Program as a regional water-quality assessment tool, may also be useful in an expanded NWUIP. Although SPARROW was not developed under the aegis of the NAWQA Program, it has been implemented in a number of NAWQA studies. The relationship between SPARROW and NWUIP will be further explored at the end of this chapter.

Coordinating the two programs, or at least elements of them, makes sense. Both are concerned with status and trends. The NAWQA Program already has a national team in place, and it does have a demonstrated need for water use data. It has a national synthesis component and seeks to link land use and water quality. Its study units do not cover the entire country, but many USGS district offices have NAWQA teams.

Estimation of the instream flows needed for environmental (or ecological) uses is becoming increasingly important as greater emphasis is placed on the

importance of keeping water in streams for ecosystems (Gillilan and Brown, 1997). Development has put heavy demands upon streams, especially in the arid West, where already-scarce surface water is subject to many demands. Each additional diversion leaves less water in stream channels to satisfy a variety of ecological uses: fish and other aquatic organisms, wildlife, and riparian vegetation.

Knowledge of how much water is used for instream ecological use will be critical, as there is growing awareness that the benefits of instream flow extend well beyond the immediate benefits to anglers and recreationists; this awareness is a powerful force changing traditional water management institutions (Thompson, 1999, p. 272). Leaving water in streams to protect aquatic and riparian ecosystems provides value to people who rarely, if ever, visit a stream, as well as to society as a whole (Gillilan and Brown, 1997). Environmental and recreational interests are increasingly pursuing an agenda that includes watershed restoration and protection of instream flows to restore and sustain a stream’s historic ecological and hydrogeologic functions (Tarlock, 1999). A growing body of evidence indicates that the maintenance or restoration of whole riverine ecosystems requires a range of flow conditions, because different species have different flow optima and may be dependent on natural disturbances such as low and high flows (Poff et al., 1997; Sparks, 1992). Estimation of the range and timing of flows needed for instream ecological use is a major scientific challenge, but one that has been given increased significance by the Endangered Species Act, which in some cases could require mimicking of the natural flow regime to protect endangered organisms (e.g., Muth et al., 2000).

According to Richter et al. (1997), the growing need to predict the biological impacts (or recovery) associated with water management activities, and to set water management targets that maintain riverine biota and socially valuable goods and services associated with riverine ecosystems, has spawned a new approach to modeling instream flow needs. These newer models have as their primary aim the design of environmentally acceptable flow regimes (i.e., pattern of flow variation) to guide river management (Richter et al., 1997). Unfortunately, recent advances in understanding the relationship between hydrologic variability and ecosystem integrity have had minimal influence on the setting of instream flow requirements or on river ecosystem management (Richter et al., 1997).

Despite the interest in and importance of instream ecological flows, the most recent NWUIP report does not document instream uses except for hydroelectric power generation, which is relatively easy to quantify (Solley et al., 1998); the upcoming 2000 report will also follow this approach. Quantitative estimates for most instream uses, not just ecological ones, are difficult to compile on a national scale (Solley et al., 1998). However, because instream uses compete with offstream uses and affect water quality and water quantity, effective water resources

planning and management dictate that such uses be individually quantified. California, for example, is documenting and calculating various instream uses, including certain kinds of environmental uses (Department of Water Resources, 1998).

Because the NWUIP relies upon state data for its national compilation, it is apparent that the program might have to calculate or estimate ecological instream uses for many states. With the relatively low level of funding currently provided to the NWUIP, this is not a realistic expectation. However, given the increasing importance of instream flows, the NWUIP should start exploring ways to estimate instream flows for ecological use for each state. It would be entirely appropriate for the NWUIP to develop techniques for estimating instream flow uses on a large-scale basis, perhaps by using resources within the USGS, such as the Midcontinent Ecological Science Center (MESC; http://www.mesc.usgs.gov).

One of the difficulties in estimating instream flows for ecological use is that different species have different instream flow needs. Riparian ecosystems require instream flows different in timing and magnitude from aquatic organisms, which often differ among themselves (e.g., fish vs. macroinvertebrates). As an example, throughout the West, inundation of alluvial floodplains is required to produce appropriate conditions for seed germination and seedling establishment of cottonwood trees, a major component of riparian forests. If floodplain inundation does not occur or if floodwaters recede too rapidly from the floodplain, cottonwood gallery forests decline (Auble et al., 1994; Rood and Mahoney, 1990).

Most methods to estimate environmental flows focus primarily on one or a few species that live in the wetted river channel. Thus, they cannot estimate instream flows for multipurpose ecological uses or for an entire ecosystem (NRC, 1992; Poff et al., 1997). Most of these methods are narrowly intended to establish minimum allowable flows. The simplest make use of easily analyzed flow data, of assumptions about the regional similarity of rivers, and of professional opinions of the minimal flow needs for selected fish species. Some of the more popular methods—e.g., Tennant; Modified Tennant; Habitat Quality Index (HQI); Wetted Perimeter; Aquatic Base Flow (ABF)—are briefly described and referenced in Lamb and Doerksen (1990) and Stalnaker et al. (1995).

A more sophisticated assessment of how changes in river flow affect aquatic habitat is contained in the Instream Flow Incremental Methodology (IFIM; Bovee and Milhous, 1978). The IFIM couples two models, one that describes the physical habitat preferences of fishes (occasionally macroinvertebrates) in terms of channel width, depth, velocity, and substrate, and a hydraulic model that estimates how available habitat space for fish life stages varies with discharge. The IFIM has been widely used as an organizational framework for formulating and evaluating alternative water management options related to production of

one or a few fish species (Stalnaker et al., 1995). The IFIM has also been widely used for instream flow modeling (Lamb and Doerksen, 1990); however, in recent years it has been increasingly criticized for its lack of biological realism (Castleberry et al., 1996) and the accuracy of its physical simulations (Williams, 1996). The IFIM is labor- and data-intensive and requires field measurements and hydraulic modeling, and it is too costly and time-consuming for the NWUIP. Stalnaker et al. (1995) estimated that 80 percent of IFIM studies for a single stream reach might take 12 months and cost $45,000. It may be possible to extend IFIM techniques to the ecosystems planning and management realm, although it is too early to tell (Stalnaker et al., 1995).

Newer modeling techniques are “holistic” in that they incorporate the premise that environmental conditions that sustain the ecosystem will sustain the constituent individual species (Meyer et al., 1999). The details of species’ responses to shifting conditions cannot be accurately modeled; however, the environmental regime can be. Thus, the guiding framework for these newer approaches is to describe the natural environmental regime usually in terms of the natural flow regime (e.g., Poff et al., 1997; Richter et al., 1996), although other environmental drivers such as temperature and sediment flux are also used. In this view, the integrity of riverine ecosystems varies in response to the deviation of the prevailing flow regime from the pre-impaired state. The natural regime is characterized in terms of the magnitude, frequency, duration, seasonal timing, and rate of change of flows—factors that are known to have demonstrable effects on aquatic habitat and ecological processes.

The NPS has developed an instream flow needs quantification known as “departure analysis” (Gillilan and Brown, 1997). This approach initially assumes that 100 percent of a stream’s flow is required to maintain the natural riverine ecosystem. The method then incrementally subtracts—through modeling—small amounts of water until the reduced streamflow produces a quantifiable or observable impact on the environment. This is termed a “departure.” Any deviation from the natural condition is considered an “impairment.” The NPS believes the departure analysis method to be valid, and although it has not been tested in court, it has resulted in several successful water rights settlements (Gillilan and Brown, 1997).

A recent method proposed by Richter et al. (1997), the RVA or “Range of Variability Approach,” focuses on the critical role of hydrologic variability in sustaining aquatic ecosystems. This method uses measured or synthesized daily streamflow data from a period during which human alteration of the hydrologic system was negligible; this streamflow record is then characterized using 32 variables defined by Richter et al. (1996). A range in variation of each of these 32 parameters is then determined and translated into management strategies. The RVA is not intended to be prescriptive; rather, it provides river managers with an interim management strategy to estimate the hydrological needs of aquatic and riparian ecosystems.

Despite the difficulty in estimating instream flows for ecological uses, the NWUIP should initiate efforts to develop such techniques on a nationwide basis. The NWUIP should collaborate with other branches of the USGS, particularly the MESC group in Fort Collins, Colorado and other agencies in the federal government.

In October 2000, the committee heard Dr. Gregory E. Schwarz, USGS Branch of Systems Analysis, present a holistic approach for the NWUIP. Dr. Schwarz discussed the 1978 Second National Water Assessment of the Water Resources Council, which provided an overview of the nation’s 99 water resource assessment regions and of the nation as a whole. This assessment provided estimates of natural streamflow, groundwater overdraft, reservoir overdraft, and consumptive use and then used a mass balance to calculate the streamflow leaving each assessment region. Dr. Schwarz argued for such a holistic approach because it provides (1) important checks on the accuracy of the source data, (2) a framework for organizing and analyzing source data, and (3) information on where water resources are scarcest and accurate source data are most needed (G.E. Schwarz, USGS, personal communication, 2000).

The approach proposed by Dr. Schwarz would incorporate the surface water quality model SPARROW (Smith et al., 1997) with water use data, including consumptive and nonconsumptive uses. SPARROW is capable of integrating watershed data over multiple spatial scales. It would be used with the EPA’s River Reach File 1 (RF1) (DeWald et al., 1985), a 1:500,000-scale digital stream dataset that is attributed with stream-reach length, average stream discharge, and average flow velocity. The idea behind this approach is to determine the extent to which water use affects measured flow; land use could also be determined, thus providing a means of assessing land use–water use relationships. In theory, streamflow estimates could be made for every reach in the RF1 stream network. A streamflow model, similar in structure to the SPARROW model for nutrient loads (Preston and Brakebill, 1999), would be estimated using actual gaging station data as the dependent variable. The regression model would preserve mass balance and could be used to make predictions for every reach in the network.

One of the explanatory variables that could be used in such a regression model would be the amount of water withdrawn by use category. A coefficient would be estimated for each category. The negative value of this coefficient might be interpreted as the share of withdrawn water that is consumed by that use

category. The statistical significance of these coefficients would give an indication of whether the water use data are of reasonable accuracy to statistically reflect the effect of the water use on the availability of water in the stream. The magnitude of the coefficients would determine if there are any gross biases in the water use estimates of the proportion of water that is consumed (G.E. Schwarz, USGS, personal communication, 2000).

An error analysis of the results could also be performed, whereby squared values of the estimated errors would be correlated with various terms in the model. The significance of these correlations would indicate potential sources of error in the model. Presumably, if the water use variables were correlated with the squared error, then they would represent a significant source of error in the model. There are two interpretations that could be drawn from this correlation: the error is due to systematic error in the water use estimate, and/or the error is due to random variation in the use coefficient. In either case, the magnitude of the correlation and the size of the error could be used to improve the estimate of consumptive use in a watershed. Thus, such a model would represent a holistic analysis of water availability that would parallel the approach of the Second National Water Assessment (G.E. Schwarz, USGS, personal communication, 2000).

Such an approach would not represent an alternative method for estimating water use. Rather, it would be a tool for assessing the relevance of water use information in determining water availability in the stream, which might prove useful in estimating instream flows for ecological needs.

Various sampling and other methods could be used to make independent estimates of water use in the absence of detailed water use estimates in every state. Site-specific survey information, wherever it is available, could be linked to ancillary data that are available everywhere. Predictions of water use for regions with no water use information could then be based on extrapolations of the survey data using the ancillary information. Standard errors of the estimates could also be computed. The primary challenge for this approach would be the identification of reasonable variables for the ancillary information. As has been described in earlier chapters, the accuracy of such variables varies with water use category.

If successful, the advantages of such a model would include the following: (1) model water use estimates would include a measure of error, (2) model estimates would be consistent with streamflow (and perhaps groundwater) data, improving the integrity of the estimates, (3) land use and water use relationships could be discerned, and (4) the model would represent a holistic approach to water resource assessment. Implementation of this model approach would require an inventory of the information available in each state for calculating water use, identification of the population of water users for extrapolation of survey data, and determination of the location of this population relative to the reach network

for synthesis with other water data (G.E. Schwarz, USGS, personal communication, 2000).

It is becoming increasingly apparent that water quality, land use, and water use cannot be considered in isolation. An approach that integrates land use, water quality, and water use would provide a great service to the nation. The USGS, through the NWUIP, could provide desperately needed leadership in this arena.

The NWUIP should be expanded to encompass water quality and the effects of land use on water use. Collaboration between the NWUIP and NAWQA programs should be strengthened.

The estimation of instream flows for ecological needs, although quite daunting, will nonetheless become increasingly important as society seeks to maintain aquatic and riparian ecosystems while also using as much of a stream’s flow for nonecological needs. Most current methods for estimating instream flows are species-specific and cannot estimate instream flow requirements for multispecies assemblages or entire aquatic/riparian ecosystems. The specter of global climate change also looms large, as it may diminish flows in already overallocated streams.

The NWUIP should encourage the development of methods to estimate instream flows for ecological needs. Collaboration with organizations already engaged in this research (e.g., MESC, The Nature Conservancy) should be pursued.

Some water-quality models such as SPARROW are capable of integrating watershed data over multiple spatial scales. Such a model could be used with River Reach File 1 (RF1) to provide estimates of how water use affects streamflow, which could then be incorporated into a national water use model. This approach would not provide an alternative method for estimating streamflow, but rather would serve as a tool for assessing the relevance of water use and land use information in determining water availability in a stream. It would form part of larger holistic approach to national water resource assessment and water use.

The integration of water use data into water-quality models has the potential to greatly strengthen the present NWUIP. Steps should be taken to determine the feasibility of this approach.