4
NAWQA Cycle II Goals— Trends and Statistical Support for Understanding

INTRODUCTION

The second goal of the National Water Quality Assessment (NAWQA) Program that continues into Cycle II (albeit with different emphases) is the determination of observed trends (and lack of trends) in water quality and the assessment of empirical associations between land use and water quality (largely in support of the understanding theme discussed in Chapter 5). The U.S. Geological Survey (USGS) has a long and distinguished record of trends assessment in the hydrologic sciences. Indeed, many of the techniques currently employed for the estimation of trends in hydrologic time series are derived from work by USGS scientists. From a practical standpoint, this is a major focus in Cycle II because trends are often the result of water quality changes associated with human-related activities. Thus, data and information on an observed trend in water quality become particularly useful when the trend is linked to its underlying cause. The reliable and early detection of trends is of fundamental value because it can provide information on changes in water quality that might be useful for future decision making and scientific understanding relating to the management of water quality. If trends are successfully detected in a timely fashion, along with a scientific understanding of the reason for those trends, it may be possible to implement management strategies to help reduce future degradation in water quality (and/or to promote future improvements).

As in Chapter 3, the organization and content of much of this chapter is based on a careful review and subsequent committee deliberations of several iterations of the NAWQA Cycle II Implementation Team (NIT) report Study-Unit Design Guidelines for Cycle II of the National Water Quality Assessment



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 93
Opportunities to Improve the U.S. Geological Survey National Water Quality Assessment Program 4 NAWQA Cycle II Goals— Trends and Statistical Support for Understanding INTRODUCTION The second goal of the National Water Quality Assessment (NAWQA) Program that continues into Cycle II (albeit with different emphases) is the determination of observed trends (and lack of trends) in water quality and the assessment of empirical associations between land use and water quality (largely in support of the understanding theme discussed in Chapter 5). The U.S. Geological Survey (USGS) has a long and distinguished record of trends assessment in the hydrologic sciences. Indeed, many of the techniques currently employed for the estimation of trends in hydrologic time series are derived from work by USGS scientists. From a practical standpoint, this is a major focus in Cycle II because trends are often the result of water quality changes associated with human-related activities. Thus, data and information on an observed trend in water quality become particularly useful when the trend is linked to its underlying cause. The reliable and early detection of trends is of fundamental value because it can provide information on changes in water quality that might be useful for future decision making and scientific understanding relating to the management of water quality. If trends are successfully detected in a timely fashion, along with a scientific understanding of the reason for those trends, it may be possible to implement management strategies to help reduce future degradation in water quality (and/or to promote future improvements). As in Chapter 3, the organization and content of much of this chapter is based on a careful review and subsequent committee deliberations of several iterations of the NAWQA Cycle II Implementation Team (NIT) report Study-Unit Design Guidelines for Cycle II of the National Water Quality Assessment

OCR for page 93
Opportunities to Improve the U.S. Geological Survey National Water Quality Assessment Program (NAWQA) (Gilliom et al., 2000; see Appendix A). That report describes the design and implementation strategy for Cycle II investigations in NAWQA study units. In this chapter, the committee considers the extent to which the NAWQA Cycle II design supports the three themes and six related objectives concerning the determination of trends in water quality and the association of spatial or temporal variations in water quality with urbanization and agricultural practices. In a larger sense, this chapter serves as a bridge between the statistical inference focus on water quality status assessments discussed in Chapter 3 and the scientific inference focus on understanding factors and processes that affect water quality in Chapter 5. To provide continuity between Chapters 3 and 5, this chapter includes an examination of statistical issues in trend analysis and causal inference in nonexperimental studies. One cannot begin a discussion of trend assessment without first describing the design of water quality networks. It is only through carefully controlled water quality monitoring that trend detection is possible. Carefully controlled water quality monitoring includes attention to numerous issues including, but not limited to, site selection, determination of monitoring locations, sampling frequency and protocols, field operations, data reporting, laboratory protocols, and so on. Next, general water quality monitoring network design issues are discussed, including the development of concise monitoring objectives, the use of statistical methods for optimal design of water quality networks, and the accurate estimation of pollutant loads. When trends are the focus of a monitoring program, other important water quality monitoring design issues include the frequency and location of sampling, the length of the data series, the possible collection of collateral information that might be used to fill in missing data or augment the data series, and trend detection methods. A sensible approach to the design of water quality monitoring networks is essential to the detection and evaluation of water quality trends—the topic of the third section of this chapter. Ideally, a water quality data series collected for trend analysis would include samples evenly spaced in time (e.g., monthly) that exhibit temporal independence and minimal measurement error. In reality, water quality data typically exhibit nonnormal distributions, seasonality, missing values, values below detection limit, changes in analytical detection methods, changes in sampling frequency and location, and serial correlation. Thus, from a practical standpoint, effective trend detection and assessment requires methods to deal with these features of real environmental data. The field of environmental statistics has emerged over the past few decades in an effort to address these complexities. Trend detection requires a rigorous background in environmental statistics. In this regard, many fundamental contributions to the field of environmental statistics have been developed and described by USGS researchers (e.g., Helsel and Hirsch, 1992). The issue of causal inference (or scientific understanding) based on observational data is discussed in the next section of this chapter. The conventional view

OCR for page 93
Opportunities to Improve the U.S. Geological Survey National Water Quality Assessment Program is that scientific understanding results from controlled experiments, and non-experimental (observational) studies should therefore not be the basis for causal inferences. However, many established scientific fields, such as epidemiology, routinely make use of nonexperimental data for scientific inference. Beyond that, emerging work in statistical inference holds promise in this area. The NAWQA Cycle II plan proposes two research themes with a total of four related objectives to address scientific understanding of the effects of urbanization and agricultural management practices on water quality, respectively. The proposed assessments will be conducted as space-for-time studies. In this context, “space for time” implies the use of studies on many watersheds and aquifers in space, all characterized by different levels of urbanization and agricultural management practices. Analysis across many watersheds and aquifers in space, at one time, allows for an evaluation of the effects of urbanization and agricultural management practices that normally evolve over time in an individual watershed. Equivalent long-term sampling of the impacts of urbanization or agricultural management practices would require much greater budgetary commitments and far more time to get results; hence the proposed space-for-time studies are creative and efficient alternatives to time sampling. The final section includes a summary of conclusions and related recommendations discussed throughout this chapter. DESIGN OF WATER QUALITY MONITORING NETWORKS AND PROGRAMS Water quality monitoring networks exist throughout the world, and methods for the design of such networks have been organized into textbooks such as those by Sanders et al. (1994) and Harmancioglu et al. (1999). The design of water quality monitoring networks includes all activities involved in the planning and management of sampling activities to collect and process water quality data for the purpose of obtaining information about the physical, biological and chemical properties of water. In addition to collecting water quality data, monitoring activities include subsequent procedures such as laboratory analyses, data processing, storage, and ultimately the statistical analyses to produce desired information. Interestingly, the steps involved in the design of water quality monitoring systems are nearly identical to the steps involved in any data management system. An efficient data management system is required if one wishes to maximize the overall utility of the resulting data. Data collection and dissemination are costly procedures, requiring large investments. The ultimate goal of any water quality data management system is to support decision making for water quality management. To enable sensible environmental management and policy decisions, an effective and integrated water quality management system is required. The components of such a water quality management system include the following:

OCR for page 93
Opportunities to Improve the U.S. Geological Survey National Water Quality Assessment Program overall water quality objectives and constraints; design of the water quality monitoring network; water quality sample collection; laboratory analyses; water quality data storage, distribution, and analyses; water quality modeling; and decision making for water quality management. Since water quality monitoring network design depends on program objectives, it is vital to identify and prioritize all monitoring objectives. Typical objectives of water quality monitoring include, but are not limited to, (1) trend monitoring, (2) biological monitoring, (3) ecological monitoring, and (4) compliance monitoring. Other common goals of water quality monitoring programs include an evaluation of the impact of pollution control efforts, a determination of the current status of water quality and the detection of causes of changes in water quality processes in both space and time. As noted previously, trend monitoring forms the basis of this chapter and is required to evaluate changing water quality conditions in a watershed or aquifer as well as the results of past and future water quality management control measures. As noted, the stated goals of a program logically dictate the resulting monitoring network design, and this can be seen clearly in the early formulation of the NAWQA program. During the late 1970s and early 1980s, the national characterization of water quality provided by the National Stream Quality Accounting Network (NASQAN) was insufficient for national policy purposes because trends identified by the NASQAN program could not be related (even grossly) to natural or anthropogenic causes (Hooper et al., 1996; see also Chapter 7). The resulting river basin assessments were well received by local constituents; however, they lacked the national perspective that would enable them to be useful to Congress. More recent versions of both the NASQAN and the NAWQA programs reflect the need (goal) to relate long-term trends (or lack of trends) in water quality to the major factors that affect observed water quality trends and conditions. In spite of very large expenditures on water quality monitoring networks, there still remains substantial uncertainty about many aspects of water quality regimes in lakes, rivers, and coastal waters in the United States and elsewhere. As a result, there are still important advances to be made with respect to understanding the spatial and temporal patterns of water quality. Such gains will come in part from improvements in our ability to design and implement water quality monitoring networks. According to Harmancioglu et al. (1999), there is a significant gap between information needs and the information provided by current water quality monitoring networks. These researchers argue that the adoption of integrated approaches to data management appears to be the only means by which the existing gaps in information can at least be minimized. Integrated approaches to data management involve careful planning a priori to ensure that the databases

OCR for page 93
Opportunities to Improve the U.S. Geological Survey National Water Quality Assessment Program resulting from a monitoring program are useful for a variety of programmatic objectives. One of the most common deficiencies associated with monitoring programs is the lack of coordination between different local, state, and federal monitoring agencies. (As described in Chapter 1, this was the general condition of national monitoring programs at the initiation of NAWQA in the 1980s.) This lack of coordination can lead to redundancies and inefficiencies in data collection efforts and to databases that are useful only for limited purposes. Integrated design of data collection efforts would require all data collection efforts to coordinate their activities. Water quality monitoring networks can benefit from integrated design because environmental processes are interdependent, so it is necessary to develop monitoring and management systems that permit information transfer across environmental processes. They should also be redesigned periodically to incorporate the knowledge obtained from current monitoring networks, as was accomplished by both the NAWQA and the NASQAN programs during their most recent redesign efforts. The Cycle II NAWQA trends network design is an excellent example of the application of an integrated approach to water quality data management. NAWQA seeks to retain the local perspective enabled by its (local and regional) river quality assessments with a consistent and integrated water quality data collection approach across basins and aquifers that enables a national synthesis of results (Gilliom et al., 2000). On the one hand, two primary but separate NAWQA trends networks were developed for Cycle II—one for streams and one for groundwater—because of the different hydrological mechanisms, temporal and spatial characteristics, and assessment requirements corresponding to these two sources of water. On the other hand, ground- and surface water networks share the objectives of (1) achieving a balanced representation of the primary hydrologic landscape, ecoregion, and land-use settings of the nation and (2) emphasizing the most important resources for drinking water and aquatic organisms. A third trend network planned for Cycle II is for contaminants in sediment. This network will focus on particle-associated trace elements and organic contaminants and is the primary approach for addressing Objectives T4 and T6 (see more below and in Appendix A). The National Trend Network of NAWQA Cycle II is integrated from a national perspective because sampling networks are designed to assess long-term trends for streams, groundwater, and sediment constituents using sites distributed among a wide range of environmental settings across the nation that are sampled systematically over time to evaluate trends and change. Further evidence of the integrated nature of NAWQA is provided by the dual use of both a temporal trends network and the space-for-time approach to assessing change, which is discussed earlier. The committee recommends that NAWQA continue emphasis on an integrated approach to water quality monitoring network design that attempts to coordinate efforts among various local, state, and federal agencies in an effort to make study unit designs as efficient and cost-effective as possible.

OCR for page 93
Opportunities to Improve the U.S. Geological Survey National Water Quality Assessment Program Central to NAWQA is a water quality monitoring network; thus, there should be a clearly established protocol for its design. Hirsch et al. (1988) defined the specific objectives for the design of both surface and groundwater quality investigations, and subsequent water quality monitoring networks have been designed at the study unit level based on those objectives. Trade-offs exist between the cost of monitoring, the number of water quality gages in space and the temporal frequency of measurements. The committee was, however, unable to discern any established method for evaluating the trade-offs that exist between the cost of monitoring and the spatial and temporal frequency of different monitoring networks designed to meet various objectives. Such methodologies are now commonly used in the design of water quantity networks. For example, Moss and Tasker (1991) compared regional hydrologic methods for the design of stream-gaging networks. Such methods that attempt to maximize regional information within a limited budget and time horizon have been in use for several decades. These methods exploit regional hydrologic regression methods that relate the spatial and temporal frequency of stream discharge gage sampling to the precision with which various regional streamflow characteristics are estimated. A natural extension of this methodology would be to apply the Spatially Referenced Regressions on Watershed Attributes (SPARROW) multivariate statistical water quality model for use in the design of water quality monitoring networks similar to the way in which regional multivariate statistical models of various streamflow statistics are used in the design of stream-gaging networks. The committee recommends that NAWQA attempt to extend the Network Analysis Using Generalized Least Squares (NAUGLS) approach introduced by Moss and Tasker (1991) for use in the design of water quality monitoring networks. NAWQA should develop an adequate and generalized approach to optimize the design of water quality monitoring networks that is quantitative and can be tailored to satisfy all relevant NAWQA objectives. There is now a significant and growing body of literature on the design of water quality monitoring networks. For example, Dixon and Chiswell (1996) reviewed 150 studies related to the design of water quality monitoring networks between 1970 and 1995. They argue that this literature is quite fractured, which led them to conclude that this subject area is not evolving in a cohesive way. They further concluded that there are too few examples of case studies that represent the types of studies most useful to the water industry. Perhaps the most difficult challenge associated with the design of water quality monitoring networks relates to the multiple and competing programmatic objectives that exist. For example, competing objectives might include meeting ambient water quality targets, assessment of trends, and regulatory monitoring. Multiple objectives are commonplace in the NAWQA program since it attempts to satisfy many different constituencies. Although Harmancioglu and Alpaslan (1992) describe a procedure for the design of water quality monitoring networks that are subject to multiple and competing objectives, such research is still in its infancy. The

OCR for page 93
Opportunities to Improve the U.S. Geological Survey National Water Quality Assessment Program NAUGLS methodology (Moss and Tasker, 1991) for the design of stream-gaging networks could be extended for use in the multiobjective design of water quality monitoring networks. Even if one assumes a single objective for the design of a water quality monitoring network, there are still many unsolved problems. For example, one programmatic objective of NAWQA is to estimate loads of selected water quality constituents at key locations. Water quality monitoring networks have been designed heuristically, with this objective in mind, so many aspects of this problem have been solved. For example, Cohn (1995) summarizes decades of research relating to the estimation of sediment and nutrient loads in rivers using statistical methods. Nearly all such estimates of long-term loads are based on statistical relationships between stream discharge (Q) and water quality constituent concentrations (C). Relationships between C and Q are also exploited to extend and fill in missing observations in trend evaluations, as discussed in the next section. Although a sizable literature exists relating to estimation of relationships between C and Q, it does not provide much guidance on the conditions under which such relations are reliable or the physical basis for such relations. Furthermore, none of the existing methods reported by Cohn (1995) seem to account for the fact that the C-Q relationship must be combined with a much longer record of Q measurements, and the resulting long-term load estimate depends heavily on the fact that there are relatively few measurements with which to estimate the C-Q relationship, compared to the nearly continuous record of Q used to estimate the long-term load. Clarke (1990) pointed out this fact when he stated that the problem of long-term load estimation is really “just an old problem (streamflow record extension and augmentation) in a new disguise.” The current literature ignores the statistical issues relating to the “record extension” aspect of the load estimation problem. Furthermore, it is not clear from existing literature what method can be used to determine the relationship between the frequency of streamflow discharge sampling and water quality constituent sampling that is necessary for reliable estimation of long-term loads. Finally, comparatively little research has been performed relating to the frequency distribution of concentrations and/or loads. It is hoped that future NAWQA and other USGS research will address all relevant issues relating to the objective of designing water quality monitoring networks to ensure accurate estimates of long-term loads. Guidance is needed on the trade-off that exists between cost of sampling and the precision associated with resulting estimates of long-term loads associated with a wide range of water quality constituents. Cost of sampling relates to the frequency of those samples in both space and time and the availability of streamflow measurements. Accuracy of resulting long-term loads is dictated by the accuracy of the concentration-discharge relationship combined with the spatial and temporal frequency of water quality and quantity samples. Quantitative methods are needed to account for

OCR for page 93
Opportunities to Improve the U.S. Geological Survey National Water Quality Assessment Program these issues. Extensive exploratory analyses are required to understand the probabilistic and mechanistic structure of water quality data. EVALUATION OF TRENDS IN WATER QUALITY “The primary emphasis of Cycle II is to assess long-term trends in water quality and improve our understanding of the factors and processes that govern water quality” (Gilliom et al., 2000). It is the emphasis on an explanation and description of the major factors that affect water quality conditions and trends that distinguishes NAWQA from previous USGS national water quality programs. During the 1980s, many studies identified trends in water quality (Smith and Alexander, 1985; Smith et al., 1982, 1987). The most common approach was to use the seasonal Kendall’s tau test, a nonparametric test for monotonic trend detection introduced by Hirsch and others (1982). Although water quality trends were frequently observed, it was difficult or impossible to determine whether those trends arose from climate variations, natural changes, or anthropogenic influences. Furthermore, the types of trend tests applied were sensitive to contamination and analytical laboratory methods because the null hypothesis of “no trend” could be rejected for a variety of reasons, many of which had no bearing on the experiment. Problems existed with water quality data assurance management and control programs during this period, particularly for constituents present in very low concentrations, such as dissolved trace elements. The NAWQA program was designed to help reduce these types of problems and produce more meaningful regional water quality assessments and trend detection and analysis studies (Hirsch et al., 1988). A central theme planned for Cycle II is an examination of the effects of land use, including urbanization and agriculture, on the biological and chemical aspects of ground- and surface water quality (Gilliom et al., 2000). The premise is to evaluate trends in both ground- and surface waters, corresponding to a wide range of land uses and watershed use categories. For example, NAWQA hopes to highlight the detection of water quality trends in watersheds, rivers, and aquifers that are important sources of drinking water supply. Ideally, NAWQA hopes to assess water quality trends nationally, while coincidentally relating detected trends to upstream land use changes. The first two objectives of this theme of Cycle II are the following: Objective T1: Determine long-term trends and changes in the concentrations of NAWQA target constituents in (a) the most important principal aquifers used for drinking water supply and (b) recently recharged groundwater upgradient of these principal aquifers in a nationally representative range of hydrologic and land use settings. Objective T2: Determine long-term trends and changes in concentrations and loads of NAWQA target constituents, and in the condition of stream ecosystems, for (a) streams representative of the primary hydrologic landscape and ecoregion

OCR for page 93
Opportunities to Improve the U.S. Geological Survey National Water Quality Assessment Program settings present in the study units, (b) a diversity of important stream ecosystems present in study units, (c) streams representative of agricultural, urban, reference, transitional, or mixed land-use settings in the nation, and (d) the most important streams used for drinking water supply. There are now hundreds of studies that have examined trends in water quality across the nation. Water quality observations usually consist of intermittent series of sparse records, so that application of classical methods for trend detection has generally produced unsatisfactory results. Rigorous statistical methods are now available for testing trends in water quality data. Many of those methods are nonparametric because such methods do not require assumptions regarding the probabilistic or stochastic structure of time series of water quality data. There are also many statistical complications associated with testing trends in water quality data such as nonnormal distributions, seasonality, missing values, values below detection limit, changes in analytical detection methods, changes in sampling frequency and location, and serial correlation, among others. Fortunately, most of these problems have now been dealt with effectively through methodological improvements. For example, many time series data sets for streamflow and associated constituent concentrations exhibit autocorrelation or serial correlation. Autocorrelation or serial correlation is a property of a data series (e.g., daily measurements of contaminant concentration) that is indicative of persistence in behavior from one data point to the next. In hydrology, autocorrelation tends to be positive, which reflects the fact that higher-than-average measurements tend to follow higher-than-average measurements (and vice versa). For example, during dry weather, streamflow is fed primarily by groundwater; thus, each day’s streamflow is related to the previous day’s flow. In that situation, positive autocorrelation would mean that low daily flow would tend to follow low daily flow. This temporal-structural phenomenon can confound one’s ability to detect trends that may be due to longer-term causative mechanisms. To address this problem, a variety of techniques have been recommended; failure to use one of these techniques in the presence of positive autocorrelation means that a hypothesis test for trend is more likely to result falsely in the conclusion of a significant trend (see Hamed and Rao, 1998; Hirsch and Slack, 1984). As noted above, positive serial correlation is common in water quality time series because it often reflects unmodeled “short-term persistence” in the data. Short-term persistence in water quality measurements can arise from transient (or short-term) human, hydrologic, or climatic factors that are not modeled, or adjusted for, in the trend assessment. For example, prolonged periods of higher-than-average precipitation might result in the short-term persistence of higher concentrations, which is manifested in positive autocorrelation when the resultant time series data are analyzed. This is not to be confused with the long-term anthropogenic influences that tend to cause meaningful trends. Ideally, short-

OCR for page 93
Opportunities to Improve the U.S. Geological Survey National Water Quality Assessment Program term persistence or serial correlation of the time series is removed (or modeled) prior to trend detection, similar to the way in which time series are standardized prior to distributional hypothesis testing. Recent NAWQA reports that have attempted to evaluate trends in surface (Vecchia et al., 1997) and groundwater resources (Wagner, 1997) demonstrate that most of the complexities described above have been accounted for. Such reports have clearly demonstrated NAWQA’s ability to discern trends in water quality due to land use and other modifications from the basic temporal stochastic structure of the water quality time series, such as short-term persistence. Unfortunately, one of the key statistical issues—spatial correlation among water quality time series—is still usually ignored. Douglas et al. (2000) found that when they accounted for the fact that streamflow time series are correlated in space, there was very little evidence of trends in either flood and drought observations across thousands of flow sequences in the United States. However, when spatial correlation was ignored, many of the regions of the United States seemed to exhibit significant trends. These researchers show that most previous studies that ignored the correlation among flow records led to erroneous claims regarding detection of trends in streamflow. The committee suspects that the same phenomenon is relevant when testing water quality data for trends, since both concentrations and loads for many constituents may be correlated with streamflow, which is itself correlated in space, in addition to the fact that concentrations are probably correlated in space themselves. The committee recommends that future NAWQA research relating to detection of trends in water quality data use methods similar to those described by Douglas et al. (2000) to account for the spatial correlation of water quality time series. Without a proper accounting for the spatial correlation of the water quality time series, resultant conclusions regarding regional trend assessments are likely be flawed. If it is found that the water quality records being evaluated for evidence of trends do not exhibit significant spatial correlations, then it is unnecessary to modify trend detection approaches to account for such correlations. Another common problem arises when the relationship between streamflow and concentration is weak and that relationship is then used to determine whether trends in water quality exist. Most water quality monitoring networks do not yield continuous measurements of concentration, yet continuous measurements of discharge are commonplace. Thus, the concentration-discharge relationship (also known as the “rating curve” approach) is often used to “fill in” or “reconstruct” a time series of concentration that is subsequently tested for trend. Misspecification of the form of the concentration-discharge relationship can lead to misspecification of resulting conclusions regarding trends. Specification of the concentration-discharge relationship has received considerable attention in the literature. Although a log-linear concentration-discharge relationship may be the most commonly used assumption in practice, many other more complex relationships have been suggested. Particularly given the wide range of constituents

OCR for page 93
Opportunities to Improve the U.S. Geological Survey National Water Quality Assessment Program and their properties with which NAWQA is dealing, these relationships must be reviewed carefully. Most recent NAWQA attempts to quantify the concentration-discharge relationship have used multivariate statistical relationships such as in the SPARROW model (Smith et al., 1997) and the models summarized by Cohn (1995) and Helsel and Hirsch (1992). The performance of multivariate statistical approaches varies greatly from site to site and between constituents. In many cases, there is not enough information in the discharge record to accurately predict the concentration. Significant improvements to the log-linear concentration-discharge relationship usually result from the inclusion of model terms that account for (1) time trends, (2) seasonality, (3) serial correlation of model residuals, (4) retransformation bias, (5) hysteresis in the concentration-discharge relationships, and (6) other explanatory variables (Cohn, 1995; Helsel and Hirsch, 1992). For example, Miller (1951) showed that adjustments to the basic rating-curve model for different seasons increased the overall model performance for sediment concentrations. Additionally, Cohn et al. (1992) showed that the explanatory power of the rating-curve model increased for nutrient loads entering Chesapeake Bay when seasonality and trend parameters were added to the model. Beyond the log-linear concentration-discharge relationship, more complex transfer function-type time series models (e.g., see Lemke, 1991) are designed to handle the type of serial structure associated with concentration and discharge measurements. The committee recommends that NAWQA place greater attention on the use of transfer function methods in its development of statistically based water quality models, because such methods properly account for the stochastic structure of time series of both water quality and its correlates. Simple bivariate and even more sophisticated multivariate statistical models are attractive because they are straightforward to estimate and employ in practice, and confidence intervals for resulting concentration and load estimates are available. However, these statistical models can exhibit lack-of-fit problems and consequently have poor predictive capability in trend studies and other assessments. Hybrid physical and statistical models are available that employ physical dynamic and spatial relationships between concentration and discharge but are parsimonious and estimable using classical statistical methods such as regression. For example, studies by O’Connor (1976) and others have developed both a physical and a statistical basis for the spatial and temporal distribution of conservative dissolved solids in rivers. Such studies could be integrated into the statistical rating curve framework currently employed by NAWQA. Similarly, simple dynamic models of the type introduced by Duffy and Cusumano (1998) can exhibit the type of time-dependent hysteresis one observes in the relationship between streamflow and concentration. The committee recommends that NAWQA continue to emphasize practical and efficient models of the concentration-discharge relationship, because such models have applicability to trend assessments as well as many other important

OCR for page 93
Opportunities to Improve the U.S. Geological Survey National Water Quality Assessment Program having a management impact in New Jersey; presentations by the New Jersey Department of Environmental Protection (NJDEP) (Karen Schaffer, NJDEP, personal communication, 2000) and the New Jersey Office of State Planning (NJOSP) (Jim Reilly, NJOSP, personal communication, 2000) attested to the usefulness of the LINJ data and information. Detailed comments on the study design for these objectives are not possible at this time because the final guidelines for these studies have not been completed. However, based on the LINJ results, the space-for-time approach is a worthwhile one and should provide valid results as to the effects of urbanization on water quality in both surface water and groundwater systems. When coupled with modeling, this approach will lead to significant results that will greatly assist managers in protecting our nation’s waters. In addition, the following recommendations are made: The NAWQA Cycle II study design has focused on current land-use conditions and their relationship to stream and groundwater attributes. As data analysis proceeds, attention must also be given to urban watershed management practices and land-use history so that land-use legacies can be incorporated. Habitat assessment and measures of fish and invertebrate communities appear to be more powerful measures of urban impact than proposed measures of the algal community, which are sensitive to recent storm events. Thus, if resources are limiting, the algal community analyses could be eliminated with the least loss to the program. RESPONSE TO AGRICULTURAL MANAGEMENT PRACTICES Agriculture is one of the dominant land uses in the United States. More than 20 percent of land is in crop production for the United States as a whole, and in some regions this percentage is much higher (e.g., 60 percent in the “Corn Belt”). Crop production necessitates major changes to the land surface, an intensive use of chemical fertilizers and pesticides, and where irrigation is used, changes to hydrology and the water budget itself. These changes have the potential for impacting the quality of surface water and groundwater resources. For example, agriculture has been identified as the major source of sediment, nutrient, and pesticide pollutants in surface water and groundwater resources and as the single largest source of water impairments (EPA, 2000). Trend analysis using NASQAN data and other research has suggested relationships between sediment concentrations and conservation programs (Smith et al., 1993), between nitrogen concentrations and the introduction of inorganic fertilizer (Goolsby et al., 1999; Smith, et al., 1993) and between the use of pesticides and groundwater quality (Barbash et al., 1999). A systematic study of how changes in agricultural production practices affect the water quality of a watershed may better enable resource managers to identify

OCR for page 93
Opportunities to Improve the U.S. Geological Survey National Water Quality Assessment Program appropriate strategies for reducing agricultural pollution. Agriculture is a dynamic system, responding to economic forces and changes in technology. The dynamic nature of cropping and livestock systems requires complementary monitoring of watershed changes in livestock numbers, cropping patterns, and other land uses and sources of pollution. Runoff from agriculture is affected by weather, which is highly variable over time. In such a setting, long-term monitoring and assessment are required to account for all stochastic influences. Although much research has been devoted to quantifying the impact of individual management practices and cropping systems on water quality, the focus of this research has largely been at the field scale. There is a need to understand the linkage among field practices, off-site movement of pollutants, and the effect on the environment at a watershed or landscape scale as it relates to surface water quality (MSEA, 1995). There is a lack of information on which fields and landscapes may be more vulnerable to off-site movement of sediment and chemicals. Long-term monitoring of water quality and changes in production and management practices should quantify the relationships between agriculture and water quality. Over the past 20 years, hundreds of projects and programs have provided resources to farmers for reducing agriculture-related pollution such as the Rural Clean Water Program, Water Quality Incentives Program, President’s Water Quality Initiative, Environmental Quality Incentive Program, Conservation Technical Assistance, Agricultural Conservation Program, Wetland Reserve Program, and Conservation Reserve Program. Changes in management practices brought about by these programs have been documented for watersheds and for larger regions. However, the impacts on water quality of most of these efforts are largely unknown. Most programs report progress in terms of surrogate parameters, such as reduced soil erosion or chemical use, rather than in terms of water quality improvements (Swader, 1993). Without direct evidence of improving water quality, farmers and others may become indifferent to the voluntary use of practices that produce environmental gains (Ribaudo et al., 1999). This could hinder the success of conservation programs. Findings from NAWQA Cycle II research can provide information to farmers and soil conservation agents about the ability of conservation practices to protect water quality. Water quality monitoring is often incompatible with an agricultural conservation program’s time scale. A water quality baseline must be established before a project is started if impacts on water quality are to be determined. This can take a number of years. It is not often feasible to delay the start of an authorized conservation program for which funds have been appropriated to establish a baseline. The length of time needed to document water quality changes may far surpass the time frame of the conservation program itself. Improvements in water quality from farmers’ efforts to reduce pollutant loadings often take years to detect and document. The links between improved management and observed changes in water quality are complex. As many as 10 consecutive years of water quality data are needed before long-term changes can be distinguished from short-term

OCR for page 93
Opportunities to Improve the U.S. Geological Survey National Water Quality Assessment Program fluctuations (Smith et al., 1993). Phosphorus accumulated in bottom sediments will affect water quality long after conservation practices have dramatically reduced phosphorus loadings in runoff. Similarly, fish, insects, and other biological indicators of a healthy stream may not reach acceptable levels until many years after water quality improves and riparian habitat is restored. Aquifers may take decades to show improvements in quality after chemical management is improved. A particular Cycle II theme to be investigated is the response of water quality to long-term changes in agricultural management practices such as tillage methods, chemical use, and cropping patterns. This theme is consistent with the original framework for the NAWQA program and will provide valuable information to the U.S. Department of Agriculture (USDA), states, and agencies responsible for conservation and other agricultural programs. In particular, states have a need for this information as they develop total maximum daily load management plans to address persistent water quality problems. Many of these plans will have to include measures to reduce agricultural pollution. The U.S. General Accounting Office (GAO) concluded in 1990 that a lack of monitoring data on the scope of nonpoint source pollution and on the effectiveness of potential solutions was restricting states’ ability to deal with the problem (GAO, 1990, 2000). Cycle II Plans and Adequacy of Approach Spatial studies of the effects of land-use changes on water quality will be undertaken to provide explanation of findings of the trend network (Gilliom et al., 2000). The NAWQA national trend networks for streams and groundwater will monitor trends by systematic sampling over time at carefully selected sites and study areas that represent key water resources and land uses. As pointed out in the NIT Cycle II design report, the trend networks only partially address issues related to changes in urban and agricultural areas, and are inadequate for assessing the effects of urban and agricultural land uses on aquatic biota and stream ecosystems. To address Cycle II trend themes aimed at water quality changes related to agricultural practices, a range of spatial studies are planned. These will be closely integrated with the trend network and topical studies undertaken to examine the factors that govern water quality, and presumably with the activities of the Hydrologic Systems Team (HST). Some of the issues that could be addressed by these studies include the effectiveness of best management practices, the impact of animal production on water quality, and the impact of changes in chemical use practices on water quality. Two objectives are identified under this theme: Objective T5: Determine the effects of long-term changes in agricultural management practices on the concentrations and distributions of NAWQA target constituents in groundwater.

OCR for page 93
Opportunities to Improve the U.S. Geological Survey National Water Quality Assessment Program Objective T6: Determine the effects of long-term changes in agricultural management practices on the concentrations and distributions of NAWQA target constituents in streams and watersheds and on stream ecosystems. To address the first objective (T5), a subset of Cycle I LUSs and shallow groundwater study-unit survey areas that have undergone documented major changes in agricultural management practices since Cycle I will be resampled. (See Chapter 1 for further information about Cycle I water quality monitoring and study design.) Additional work will be conducted to link the history of land use change to the quality of groundwater. Special emphasis will be given to areas in which the shallow groundwater sampled through land-use studies is recharged for deeper groundwater used for drinking supplies. The studies will be located in areas where local changes in agricultural practices have occurred over several years, where hydrogeology has been well defined, and where groundwater flow can be simulated with numerical models. Flowpath studies will be conducted to investigate the links between shallow groundwater, which is most easily contaminated, and deeper groundwater used for drinking water. Flowpath studies will also be conducted to investigate links between shallow groundwater and streams. These studies will be coordinated with targeted spatial studies for streams and related understanding themes. A space-for-time approach will be used to meet the second objective (T6) under the theme, premised on the hypothesis that networks of watersheds can be identified based on gradients in agricultural factors. Spatial studies will use a design that investigates a subset of watersheds chosen from the population of watersheds present in a geographic area of interest. Focus will be on management practices that vary regionally or nationally. The NIT Cycle II design report (Gilliom et al., 2000) suggests that an initial focus of studies may be provided by linking agricultural spatial studies to topical studies relevant to Objective U13 (nutrient enrichment; see Chapter 5) and topical studies on nutrient sources and transport. The USGS and NAWQA are well positioned to carry out this work. NAQWA has established water quality baselines and monitoring networks in the study units and is operating at a time scale sufficient to establish relationships between production practices and water quality at a watershed scale. This research will not be able to identify the water quality benefits of individual practices, but the USDA research can provide this information. Specifically, the Agricultural Research Service (ARS) evaluates new and improved strategies for reducing water contamination from agricultural lands. ARS and its State Experiment Station partners (and state Water Resources Research Institutes) develop field practices to reduce impacts of nutrients, pesticides, and other synthetic chemicals; pathogens and other bacterial contaminants; sediments; and trace elements on surface water and groundwater. These management practices and systems are evaluated for their water quality and other environmental benefits at

OCR for page 93
Opportunities to Improve the U.S. Geological Survey National Water Quality Assessment Program field, farm, watershed, and basin scales on irrigated and rain-fed cropland and grazing land. An example of what NAWQA can do regarding this theme is provided by the Lake Erie-Lake St. Clair Drainages (LERI) Study Unit (see Figure 1-1). Retrospective and other monitoring data were coupled with land-use information to evaluate the relation between suspended sediment discharges, conservation tillage practices, and soil loss in the Maumee River Basin (Myers and Metzker, 2000). In brief, increases in conservation tillage (and decreases in soil loss) were related to decreases in suspended sediment discharge from streams. This research provides a model for other study units’ undertaking Cycle II analyses of the relationships between agriculture and water quality. The NIT Cycle II report (Gilliom et al., 2000) does not elaborate on how a landscape as diverse as agriculture’s would be characterized in a study that includes long-term changes in agricultural management practices. A watershed may contain hundreds of farms and thousands of individual fields. These fields may be growing a variety of crops and be farmed using a variety of tillage, conservation, and other production practices. It seems that some sort of classification scheme is needed to reduce the diversity of agriculture to a manageable set of variables that can be tied to water quality. Furthermore, changes in agricultural practices from year to year will have to be monitored, including chemical application rates. Will this information be obtained from aerial surveys, personal interviews with producers, or some other means? To ascertain agriculture’s impacts on water quality, changes in other land-use activities and point source discharges must also be accounted for. Methods for obtaining this information on a periodic basis have to be outlined. An important component of Cycle II that should support the understanding of agriculture’s impacts on water quality trends is the greater emphasis on quantitative hydrologic analysis and the mathematical modeling of processes. The proposed HST is a national study team whose role will be the application and support of hydrologic and water quality models in NAWQA studies, including the use of models to aid in interpretive analysis of water quality. It seems logical that the HST will support research on the effects of changes in agricultural management on ground- and surface waters (Objectives T5 and T6). In the Cycle II report (Gilliom et al., 2000), the USGS proposes that an Agricultural Chemical Source and Transport Team (ASTT) be formed to determine how physiographic characteristics and agricultural management practices affect runoff and recharge processes and the quality of shallow groundwater and streams. The Cycle II NIT report could be improved if it included a better presentation of how the HST and subsidiary ASTT will be coordinated with teams undertaking the various trend and understanding objectives. The committee provides the following related recommendations for Cycle II, based on recent changes in agriculture: Assess the impacts on water quality of state water quality protection pro-

OCR for page 93
Opportunities to Improve the U.S. Geological Survey National Water Quality Assessment Program grams (e.g., California’s Pesticide Contamination Prevention Act, Nebraska’s Ground Water Management and Protection Act, the Everglades Forever Act, North Carolina’s Nutrient Sensitive Waters). Consider evaluating changes in sediment in areas where cropland has been retired by the Conservation Reserve Program. Examine the length of time between the introduction of a new pesticide and its appearance in water resources. Study changes in nitrogen and phosphorus loadings in regions where confined animal feeding operations have concentrated. Assess impacts on water quality of conservation tillage and other low-input agricultural practices. Consider tasking the proposed HST team to develop model(s) relating agriculture to water quality that can be applied in major agricultural areas. CONCLUSIONS AND RECOMMENDATIONS The committee notes that many of the topics discussed in this chapter on the trend goal of NAWQA for Cycle II are pertinent to addressing the committee’s statement of task (see Preface to this report). This is particularly true for the interrelated issues on extrapolation and aggregation of information at regional and national scales. In this regard, the USGS has long been a leader in the regionalization of hydrologic (especially streamflow) data, which has involved both aggregation and extrapolation efforts. As noted elsewhere in this report, NAWQA’s consistent protocols and sampling regimen and the integrated approach to network design provide a good baseline that minimizes many problems that can make extrapolation and regionalization of its data and information problematic. The USGS has a long and distinguished record of experience in hydrologic trends assessment. Indeed, many of the techniques currently employed for the estimation of trends in hydrologic time series are based on work by USGS scientists. The reliable and early detection of trends is of fundamental value because it can provide information on changes in water quality (especially related to anthropogenic sources) that might be useful for decision making and scientific understanding relating to the management of water quality. If trends are successfully detected in a timely fashion, along with a scientific understanding of the cause of those trends, it may be possible to implement management strategies to help reduce future degradation in water quality. Thus, information on a trend in water quality becomes particularly useful when the trend is linked to its underlying cause. To support this type of causal assessment, the Cycle II NAWQA program has established three themes and six objectives for determination of the trends in the status of water resources and the effects of urbanization and agricultural practices on water quality. The committee finds that the USGS and NAWQA are well positioned to carry out this important work in Cycle II. In this regard, NAQWA

OCR for page 93
Opportunities to Improve the U.S. Geological Survey National Water Quality Assessment Program has established water quality baselines and monitoring networks in the Cycle I study units and is operating at time and spatial scales sufficient to establish these relationships. Several recommendations regarding the ability of the Cycle II NAWQA program to meet these trends and related objectives are provided below: NAWQA should continue its emphasis on an integrated-approach to water quality monitoring network design that attempts to coordinate efforts among various local, state, and federal agencies in an effort to make study unit designs as efficient and cost effective as is possible. The use of the NAUGLS approach introduced by Moss and Tasker (1991) should be extended to the design of water quality monitoring networks in Cycle II. This may be accomplished by extending the SPARROW multivariate statistical water quality model for use in the design of water quality monitoring networks similar to the way in which regional multivariate statistical models of various streamflow statistics have been used in the design of stream-gaging networks. NAWQA should develop a generalized quantitative approach to optimize the design of water quality monitoring networks that can be tailored to satisfy relevant NAWQA objectives. It is hoped that future NAWQA research will address all relevant issues relating to the objective of designing water quality monitoring networks to ensure accurate estimates of long-term loads. Guidance is needed on the trade-off that exists between cost of sampling and the precision associated with resulting estimates of long-term loads associated with a wide range of water quality constituents. Cost of sampling relates to the timing of these samples in both space and time and the availability of streamflow measurements. Accuracy of resulting long-term loads is dictated by the accuracy of the concentration-discharge relationship combined with the spatial and temporal frequency of water quality and quantity samples. Quantitative methods are needed to account for these issues. Extensive exploratory analyses are required to understand the probabilistic and mechanistic structure of water quality data collected by NAWQA. If trend evaluations are to be meaningful, they must account for all relevant statistical complexities to enable meaningful conclusions. Although most statistical complexities have already been dealt with, the issue of spatial correlation for regional inferences has not commonly been considered. Without a proper accounting for the spatial correlation of the water quality time series, resultant conclusions regarding trend assessments are likely be flawed. Future NAWQA research relating to detection of trends in water quality data should use methods similar to those described by Douglas et al. (2000) to account for the spatial correlation of water quality time series. NAWQA should place greater attention on the use of transfer function methods in its development of statistically based water quality models, because such methods properly account for the stochastic structure of time series of both water quality and its correlates.

OCR for page 93
Opportunities to Improve the U.S. Geological Survey National Water Quality Assessment Program NAWQA should continue to emphasize practical and efficient models of the concentration-discharge relationship, because such models have applicability to trend assessments as well as many other important societal problems relating to water quality management. The committee also recommends that NAWQA place much greater emphasis in the future on the integration of physical and statistical models of the concentration-discharge relationship in the hopes that such integrative research will lead to much more credible and useful models. The reliance on observational data analysis for causal inferences is always subject to some controversy. Accordingly, NAWQA should consider application of recently developed methods in probability-based causal inference (e.g., Bayesian techniques) for its cause-and-effect studies. The NAWQA Cycle II study design has focused on current land-use conditions and their relationship to stream and groundwater attributes. As data analysis proceeds, attention must also be given to urban watershed management practices and land-use history so that land-use legacies can be incorporated. Habitat assessment and measures of fish and invertebrate communities appear to be more powerful measures of urban impact than proposed measures of the algal community, which are sensitive to recent storm events. Thus, if resources are limiting, the algal community analyses could be eliminated with the least loss to the program. The proposed NAWQA HST modeling team should include examination of strategies that combine information from site-specific data and regional model forecasts. To assess the water quality impacts related to agriculture, NAWQA should consider studies related to exemplary state water quality protection initiatives (e.g., California’s Pesticide Contamination Prevention Act, Nebraska’s Ground Water Management and Protection Act, the Everglades Forever Act, North Carolina’s Nutrient Sensitive Waters). NAWQA should consider evaluating changes in sediment in areas where cropland has been retired by the Conservation Reserve Program. The length of time between the introduction of a new pesticide and its appearance in water resources should be examined. Changes in nitrogen and phosphorus loadings should be studied in regions where confined animal feeding operations have concentrated. The impacts on water quality of conservation tillage and other low-input agricultural practices should be assessed. REFERENCES Barbash, J. E., G. P. Thelin, D. W. Kolpin, and R. J. Gilliom. 1999. Distribution of Major Herbicides in Ground Water of the United States. U.S. Geological Survey Water-Resources Investigations Report 98-4245. Sacramento, Calif.: U.S. Geological Survey. Clarke, R. T. 1990. Statistical characteristics of some estimators of sediment and nutrient loadings. Water Resources Research 26(9):2229-2233.

OCR for page 93
Opportunities to Improve the U.S. Geological Survey National Water Quality Assessment Program Cohn, T. A. 1995. Recent advances in statistical methods for the estimation of sediment and nutrient transport in rivers. U.S. Natl. Rep. Int. Union Geod. Geophys. 1991-1994. Review of Geophysics 33:1117-1123. Cohn, T. A., D. L. Caulder, E. J. Gilroy, L. D. Zynjuk, and R. M. Summers. 1992. The validity of a simple log-linear model for estimating fluvial constituent loads: An empirical study involving nutrient loads entering Chesapeake Bay. Water Resources Research 28(9):2353-2363. Dixon, W., and B. Chiswell. 1996. Review of aquatic monitoring program design. Water Resources Research 30(9):1935-1948. Douglas, E. M., R. M. Vogel, and C. N. Kroll. 2000. Trends in floods and low flows in the United States: Impact of spatial correlation. Journal of Hydrology 240:90-105. Duffy, C. J., and J. Cusumano. 1998. A low-dimensional model for concentration-discharge dynamics in groundwater stream systems. Water Resources Research 34(9):2235-2247. EPA (U.S. Environmental Protection Agency). 2000. The Quality of Our Nation’s Water: A Summary of the National Water Quality Inventory. 1998 Report to Congress. EPA841-S-00-001. Washington, D.C.: U.S. Environmental Protection Agency, Office of Water. Frick, E. A., D. J. Hippe, G. R. Buell, C. A. Couch, E. H. Hopkins, D. J. Wangsness, and J. W. Garrett. 1998. Water Quality in the Apalachicola-Chattahoochee-Flint River Basin, Georgia, Alabama, and Florida, 1992-95. U.S. Geological Survey Circular 1164. Reston, Va.: U.S. Geological Survey. GAO (U.S. General Accounting Office). 1990. Greater EPA Leadership Needed to Reduce Nonpoint Source Pollution. GAO/RCED-91-10. Washington, D.C.: General Accounting Office. GAO. 2000. Key EPA and State Decisions Limited by Inconsistent and Incomplete Data. GAO/ RCED-00-54. Washington, D.C.: General Accounting Office. Gilliom, R. J., K. Bencala, W. Bryant, C. A. Couch, N. M. Dubrovsky, L. Franke, D. Helsel, I. James, W. W. Lapham, D. Mueller, J. Stoner, M. A. Sylvester, W. G. Wilber, D. M. Wolock, and J. Zogorski. 2000. Study-Unit Design Guidelines for Cycle II of the National Water Quality Assessment (NAWQA). U.S. Geological Survey NAWQA Cycle II Implementation Team. Draft for internal review (11/22/2000). Sacramento, Calif.: U.S. Geological Survey. Goolsby, D. A., W. A. Battaglin, G. B. Lawrence, R. S. Artz, B. T. Aulenbach, R. P. Hooper, D. R. Keeney, and G. J. Stensland. 1999. Flux and Sources of Nutrients in the Mississippi-Atchafalaya River Basin: Topic 3 Report . Silver Spring, Md.: National Oceanic and Atmospheric Administration. Hamed, K. H., and A. R. Rao. 1998. A modified Mann-Kendall trend test for autocorrelated data. Journal of Hydrology 204:182-196. Harding, J. S., E. F. Benfield, P. V. Bolstad, G. S. Helfman, and E. B. D. Jones III. 1998. Stream biodiversity: The ghost of land-use past. PNAS (US) 45:14843-14848. Harmancioglu, N. B., and N. Alpaslan. 1992. Water quality monitoring network design: A problem of multi-objective decision making. Water Resources Bulletin 28(1):179-192. Harmancioglu, N. B., O. Fistikoglu, S. D. Ozkul, V. P. Singh, and M. N. Alpaslan. 1999. Water Quality Monitoring Network Design, Water Science and Technology Library. Volume 33. Dordrecht, The Netherlands: Kluwer Academic Publishers. Helsel, D. R., and R. M. Hirsch. 1992. Statistical Methods in Water Resources. Amsterdam: Elsevier. Hirsch, R. M., and J. R. Slack. 1984. A nonparametric trend test for seasonal data with serial dependence. Water Resources Research 20(6):727-732. Hirsch, R. M., W. M. Alley, and W. G. Wilber. 1988. Concepts for a National Water-Quality Assessment Program. U.S. Geological Survey Circular 1021. Reston, Va.: U.S. Geological Survey. Hirsch, R. M., J. R. Slack, and R. A. Smith. 1982. Techniques of trend analysis for monthly water quality data. Water Resources Research 18:107-121. Hooper, R. P., D. Goolsby, S. McKenzie, and D. Rickert. 1996. National Program Framework, U.S. Geological Survey, NASQAN Redesign. Reston, Va.: U.S. Geological Survey.

OCR for page 93
Opportunities to Improve the U.S. Geological Survey National Water Quality Assessment Program Lemke, K. A. 1991. Transfer-function model of suspended sediment concentrations. Water Resources Research 27(3):293-305. McMahon, G., and T. F. Cuffney. 2000. Quantifying urban intensity in drainage basins for assessing stream ecological conditions. Journal of the American Water Resources Association 36:1247-1260. Miller, C. R. 1951. Analysis of Flow-Duration, Sediment-Rating Curve Method of Computing Sediment Yield. Denver, Colo.: U.S. Bureau of Reclamation. Moss, M. E., and G. D. Tasker. 1991. An intercomparison of hydrological network-design technologies. Hydrological Sciences Journal 36(3):209-221 MSEA (Management Systems Evaluation Area). 1995. Management Systems Evaluation Areas: Report of the Progress from 1990-1995. Washington, D.C.: U.S. Department of Agriculture. Myers, D. N., and K. D. Metzker. 2000. Status and Trends in Suspended-Sediment Discharges, Soil Erosion, and Conservation Tillage in the Maumee River Basin—Ohio, Michigan, and Indiana. U.S. Geological Survey Water-Resources Investigations Report 00-4091. Columbus, Ohio: U.S. Geological Survey. O’Connor, D. J. 1976. The concentration of dissolved solids and river flow. Water Resources Research 12(2):279-294. Pearl, J. 2001. Causality: Models, Reasoning, and Inference. Cambridge, U.K.: Cambridge University Press. Porter, S. D., T. F. Cuffney, M. E. Gurtz, and M. R. Meador. 1993. Methods for Collecting Algal Samples as Part of the National Water-Quality Assessment Program. U.S. Geological Survey Open-File Report 93-409. Raleigh, N.C.: U.S. Geological Survey. Ribaudo, M. O., R. D. Horan, and M. E. Smith. 1999. Economics of Water Quality Protection from Nonpoint Sources: Theory and Practice. Agricultural Economic Report 782. Washington, D.C.: U.S. Department of Agriculture, Economic Research Service. Sala, O. E., F. S. Chapin, J. J. Armesto, E. Berlow, J. Bloomfield, R. Dirzo, E. Huber-Sanwald, L. F. Huenneke, R. B. Jackson, A. Kinzig, R. Leemans, D. M. Lodge, H. A. Mooney, M. Oesterheld, N. L. Poff, M. T. Sykes, B. H. Walker, M. Walker, and D. H. Wall. 2000. Global biodiversity scenarios for the year 2100. Science 287:1770-1774. Sanders, T. G., R. C. Ward, J. C. Loftis, T. D. Steele, D. D. Adrian, and V. Yevjevich. 1994. Design of Networks for Monitoring Water Quality. Littleton, Colo.: Water Resources Publications. Smith, R. A., R. M. Hirsch, and J. R. Slack. 1982. A Study of Trends in Total Phosphorus Measurements at NASQAN Stations. U.S. Geological Survey Water Supply Paper 2190. Reston, Va.: U.S. Geological Survey. Smith, R. A., R. B. Alexander, and M. G. Wolman. 1987. Water quality trends in the nation’s rivers. Science 235:1607-1615. Smith, R. A., R. B. Alexander, and K.J. Lanfear. 1993. Stream Water Quality in the Conterminous United States—Status and Trends of Selected Indicators During the 1980s, National Water Summary 1990-91. U.S. Geological Survey Water Supply Paper 2400. Reston, Va.: U.S. Geological Survey. Available online at http://water.usgs.gov/nwsum/sal/index.html. Smith, R. A., G. E. Schwarz, and R. B. Alexander. 1997. Regional interpretation of water-quality monitoring data. Water Resources Research 33(12):2781-2798. Smith, R. A. and R. B. Alexander. 1985. Trends in Concentrations of Dissolved Solids, Suspended Sediment, Total Phosphorus and Inorganic Nitrogen at U.S. Geological Survey National Stream Quality Accounting Network Stations. U.S. Geological Survey Water Supply Paper 2275. Reston, Va.: U.S. Geological Survey. Spirtes, P., C. Glymour, and R. Scheines. 2001. Causation, Prediction, and Search. Cambridge, Mass.: MIT Press. Swader, F. N. 1993. Agricultural research: The challenge in water quality. Pp. 16-20 in Proceedings of Agricultural Research to Protect Water Quality Conference, Minneapolis, February 21-24. Ankley, Iowa: Soil and Water Conservation Society.

OCR for page 93
Opportunities to Improve the U.S. Geological Survey National Water Quality Assessment Program USCB (U.S. Census Bureau). 2001. About Metropolitan Areas. Available online at http://www.census.gov/population/censusdata/urpop0090.txt. Vecchia, S., G. Wiche, and W. Berkas. 1997. Evaluation of Sampling Procedures for Monitoring Trends in Surface Water Quality for the NAWQA Program. Internal draft. Bismark, N.D.: U.S. Geological Survey. Wagner, B. 1997. Evaluation of Alternative Sampling Strategies for Long-Term Groundwater Quality Trend Detection in NAWQA. July 28. Internal draft. Menlo Park, Calif.: U.S. Geological Survey.