James E. (Jim) Harrison, Environmental Scientist

U.S. EPA Region 4, Atlanta, GA

E-mail: harrison.jim@epamail.epa.gov

Monitoring water resources means many things to different people and entities. Federal and state agencies have varying missions, legislative mandates, organizational cultures and stakeholder concerns. Political and organizational boundaries, and other real or imagined obstacles to cooperation, often tend to keep everyone’s monitoring systems separate and disjointed. Still, regulatory agencies and others who monitor water quality often share common key questions (see ITFM 1994 and SAMAB 1996b) including:

• What is the desired/reference condition (standards and criteria)?
• Where (and what) are our problems (screening)?
• How do we fix (or prevent) them? and,
• Are we making progress over time (evaluation) at multiple spatial scales: large, medium and small areas? [Natural resources agencies should emphasize evaluation of environmental results to meet the spirit and intent of the Government Performance and Results Act (GPRA).]

Other important questions and variations of those listed above can also tend to dominate agencies’ agendas, monitoring approaches, and, sometimes, monitoring resources. Some of these include:

• Are environmental laws being complied with? Compliance with environmental regulations and statutes through compliance inspections, sampling, self monitoring and audits will always be important to insure, for example, that end-of-pipe limits are met and that required management practices are in place.
• Are emergencies and spills being responded to quickly and effectively? Crises and accidents will always happen but should be minimized through effective prevention approaches and education.
• How do stresses and combinations of stresses affect aquatic ecosystem condition? Basic and applied scientific inquiry into the mechanisms and processes of environmental, social and economic factors affecting water resource integrity such as the national water and watersheds research program (see NSF and EPA 1998) and the U.S. Geological Survey’s National Water Quality Assessment Program (see Gilliom and others 1995) will always be crucial to eventual success in restoring and maintaining water quality.

Answering each of these questions may require different monitoring network designs. Managing the nation’s waters to restore and maintain aquatic ecosystem integrity requires reliable answers to all of these questions. Thus, a systems view (Senge 1990) is needed to simultaneously address multiple objectives, and to successfully answer all of our key questions for the least cost. Now, let’s consider each of the four key questions in turn.

Continuing development of protective standards and criteria for water resource integrity that go beyond existing chemical and toxicity based approaches will be essential to protect and restore waters affected by a wide variety of stresses caused by humans. Knowing what we want for the desired condition of waters will require increasing attention to biological condition measures and biocriteria for fish, benthic and other communities (Davis and Simon 1995), measuring and understanding habitat attributes influenced by both riparian and watershed factors such as clean sediment, and channel type/stability (Rosgen 1994), and development of protective criteria for nutrients and their effects on biological systems (USEPA 1998).

All of these new types of criteria will require integration of workable spatial frameworks to define where the criteria apply and to use them effectively. Vast differences in aquatic communities, soils, geology, vegetation and land use (for example) across the nation necessitate criteria tailored to different regions with similarity of environmental characteristics. Consistent multi-agency and multi-national approaches for defining ecological regions (see Commission for Environmental Cooperation 1997) are becoming available. Likewise, watershed boundaries can now be quickly and economically delineated for any point using digital elevation models (DEM’s). Thus, both ecological regions and watersheds (Omernik and Bailey 1997) will be essential spatial stratifications for applying our new criteria to real-world situations. Integration of landscape data (Naiman 1996), reference site data from relatively unimpacted reference areas (Hughes 1995) and impacted site in-stream data will be essential to better define attainable conditions for aquatic life. Science-based indicators of aquatic ecosystem conditions and stresses will evolve and improve over time as our knowledge and understanding of aquatic systems increases.

Figure 1 shows the locations of over 200 stream reference sites established by the eight Southeastern states of EPA Region 4 and the ecoregions that they represent. Note that the ecoregional boundaries cross state lines and administrative divides such as the EPA regions. Long term data gathering from the reference areas will be crucial to document natural ranges of the best expected biological, habitat, channel and substrate conditions and the variability of important stresses such as nutrients and sedimentation.

Figure 2 illustrates the ongoing rebalancing of monitoring program efforts. Two decades ago most monitoring centered on chemical variables. Toxicity testing was added to the mix during the 1980’s. Biological, habitat (Karr 1993), channel morphology and hydrologic measures (Poff and others 1997) are now being actively developed and added to the water quality toolbox.

Traditionally, finding water quality problems has usually relied on citizen complaints and on targeted monitoring networks that have often been biased toward known problem areas such as point sources. Screening to identify problem areas and significant stresses in complete and systematic ways remains a continuing need, especially for dispersed nonpoint source issues. Since in-stream/on-the-ground monitoring must be done within resource constraints, systematic screening strategies are needed to identify potential problem areas and to prioritize more intensive targeted monitoring to confirm problems and then solve them. The clear need for more comprehensive screening is underscored by numerous citizen suits targeting Total Maximum Daily Loads (TMDL’s). Many of these suits emphasize shortcomings in identification of impaired waters by existing monitoring efforts.

One promising option for this type of systematic screening utilizes extrapolation of existing data by building and calibrating empirical or other models relating landscape and in-stream factors (Zucker and White 1996). Such an approach relies greatly on new but readily available satellite (or aerial photo) based landscape data, in addition to traditional in-stream monitoring information. This approach can provide estimates of in-stream condition with known confidence, assuming the calibration data comes from a statistical sample, where little or no in-stream information is available. One distinct advantage of a landscape modeling approach is that we will soon have consistent landscape classification information for the entire country through the Multi-Resolution Landscape Characterization (MRLC) consortium (Vogelmann and others 1998a). Integration of multi-resolution remote sensed data (satellite and air photo coverages) will be critical to systematic screening at multiple scales. Coarser resolution satellite data might be most effective for characterizing large areas such as ecoregions or river basins; while finer resolution data from air photo interpretation might be most accurate and cost effective for smaller areas such as watersheds or subwatersheds. Predictions of likely problem areas using extrapolation based on calibrated relationships between landscape and in-stream factors can be confirmed with more intensive in-situ sampling appropriate to the scale of the resource being studied.

The Southern Appalachian Assessment (SAMAB 1996a) provides some examples of potential landscape factors that might be used to construct empirical relationships linking landscape and in-stream factors. Figure 3 maps intensive human land use within ecoregions and hydrologic areas for the Southern Appalachians. Similarly, Figure 4 shows the percentage of riparian forest for the same area. The lower Nolichucky River drainage in Eastern Tennessee shows high potential stress for both of these factors. Looking more closely at the actual land use pattern in this area (Figure 5) and further analyses of smaller drainages can show where stresses are concentrated. Examination of specific stress factors such as riparian forest cover might also guide targeting of potential restoration actions: note the stream side zones dominated by agricultural crops and by urban land uses. A wide variety of potential landscape indicators of catchment health (Jones and others 1996) are worthy of consideration. More discussion on approaches and considerations for building models linking landscape and in-stream factors follows in the "building and extrapolating landscape/in-stream relationships" section below.

Detailed on-the-ground data, process modeling and higher resolution information are often required to prescribe and implement solutions to site-specific water quality problems. Point source restoration opportunities have traditionally been guided by site specific waste load allocation (WLA) studies and modeling of dissolved oxygen, nutrients and toxics. Nonpoint sources are addressed primarily by voluntary implementation of best management practices (BMP’s). New tools are becoming available through Environmental Protection Agency (EPA) and state efforts to develop total maximum daily loads (TMDL’s) for impaired waters. Some of these include the "BASINS" modeling package (Lanlou and others 1996) which uses the ArcView geographic information system (GIS) software as input to the HSPF (Hydrologic Simulation Package Fortran) hydrologic and water quality model; and air photo interpretation techniques (Malone and Bower 1998), such as the Tennessee Valley Authority’s (TVA) nonpoint source atlas process. Detailed attributes identified from low or medium altitude photographs aid evaluation of causes and sources of nonpoint source pollution, and can thus help prioritize and guide voluntary restoration activities.

The eventual success of water quality restoration and maintenance efforts will likely rest more and more on voluntary citizen actions focusing on local, multi-stakeholder watershed protection approaches (Montgomery and others 1995) in addition to regulatory and permitting actions. Local community action should involve monitoring of water quality (see Firehock and West 1995 for example) as well as direct restoration and management actions. Best management practices (BMP’s) implementation and other approaches such as stream channel restoration will all benefit from spatial data integration, clear priority setting (USEPA 1996) and careful evaluation of the effectiveness (at small to large scales) of management actions. Increasing local responsibility for environmental results such as local evaluation and enforcement of construction and stormwater runoff BMP’s are one trend in this direction. Community knowledge of the interrelationships between landscape activities and water resources (Arnold 1996) and a clear positive vision of the value of healthy waters can help to insure that growing human populations and infrastructure do not destroy our collective water resource heritage. Making explicit choices about development practices and land use patterns to protect water resources will be crucial to our future quality of life. All communities should take advantage of the best available scientific input and recommendations (see Nichols and others 1997 for example) to make these decisions.

Evaluation of environmental results (Committee on Government Affairs 1993) may be the bottom line, but thus far, our monitoring and assessment systems can not yet defensibly answer this question for large areas. Statistical samples examining specific types of waters in a geographic area are proving their utility for evaluating results for large areas over time. Examples from the Eastern US include the Savannah River Basin Regional Monitoring and Assessment Program (REMAP) sites (Figure 6) (Raschke and Howard 1996), the South Florida Ecosystem Assessment (Stober 1996) and the Mid-Atlantic Highlands Assessment of EPA Region 3. These studies, and many other similar examples from around the country, provide unbiased estimates of the distribution of aquatic ecosystem condition with known confidence. EPA’s guidance to states on preparing biennial water quality reports to Congress (USEPA 1997) explicitly encourages both probability based sampling approaches to describe water quality status for large areas, and the testing and potential use of modeling approaches for better screening of potential problem areas and targeting of more intensive monitoring.

Spatial and temporal aggregation of comparable monitoring results of known quality from multiple agencies and groups can also, in principle, multiply the utility and effectiveness of everyone’s monitoring programs and projects. Aggregation can often be problematic, however, since most monitoring site networks have not been designed to allow easy aggregation. Usually, sampling sites are biased to known or suspected problem areas. Additionally, deciding what area a sampling point or points represents can be difficult. Representativeness of sample points must be known to defensibly extrapolate to larger areas.

Constructive synergies between evaluation and screening approaches are possible by using data from statistical surveys as input to models relating landscape and in-stream indicators. This allows the data from statistical networks to do double duty as the model building and calibration data that feed landscape/in-stream models as a promising screening tool.

Linking landscape ecology with in-stream indicators of biological, habitat and chemical quality is an active area of research and development. Numerous landscape pattern and structure metrics have been proposed (Riitters and others 1995) and many are being actively evaluated for their value in predicting water resource conditions (O’Neil and others 1997). Suites of potential indicators of watershed condition are already being produced for large areas of the United States. One extensive example is a recent landscape atlas of the Mid-Atlantic region—EPA Region 3 (Jones and others 1997). Models covering limited areas and one or more in-stream factors have also been developed. A decade ago, before widespread availability of GIS capabilities and techniques, Steedman pioneered this type of modeling by relating watershed and riparian forest and urban land cover to in-stream fish community condition—the Index of Biological Integrity or IBI (Steedman 1988). Significant relationships have also been shown between watershed and riparian agricultural uses, and in-stream chemistry (conductivity and nutrients)(Hunsaker and Levine 1995).

Considerable recent work elucidates a number of interesting aspects of landscape/in-stream relationships. Roth and others considered riparian and watershed factors at multiple spatial scales. They found that whole watershed measures, including those for the riparian zone, had higher predictive capability. Their results also suggest that presence of wetlands may be important for some regions where they naturally occur (Roth and others 1996). Work by Richards and others emphasizes that natural or anthropogenic landscape features can potentially dominate in-stream end points. Watershed soils, geology, land form and other natural characteristics should be considered along with human induced land uses (Richards and others 1996). Some studies are beginning to focus on large areas. Wang and others study of the effect of land use on Wisconsin-streams suggests levels of agricultural influence which may begin to dominate in-stream biological and habitat integrity. Their work also suggests factors that may mitigate agricultural influences; some natural, such as stream gradient, and some human influenced, such as intact riparian vegetation (Wang and others 1997).

Numerous investigators are considering the influence of urban and suburban development on water resource integrity. Schueler has reviewed available studies relating impervious cover to in-stream biology and habitat. He suggests that watershed impervious area of greater than 10% usually results in significant impact to aquatic systems and that imperviousness greater than 25% usually results in severe impacts. Hydrologic changes to in-stream flows due to impervious pavements, buildings, etc. are considered to be the instigating factor for in-stream habitat deterioration such as eroding stream banks that result in significant decreases in biological integrity (Schueler 1994). May and others’ study of urbanization effects on Puget Sound Lowland Ecoregion small streams found that riparian buffer integrity, availability of large woody debris (an important habitat indicator), and biological integrity (benthic and fish) was strongly related to watershed urbanization as measured by total impervious area. They were also able to establish significant interrelationships between watershed urbanization, riparian corridor integrity, and in-stream biological end points (May and others 1997). Maxted and Shaver found significant nonlinear relationships between impervious cover and biological community integrity measures for streams in Delaware. Their results also raise questions about the effectiveness of storm water retention or detention systems versus infiltration designs. Study sites with retention/detention storm water BMP’s did not appear to improve biological conditions (Maxted and Shaver 1996). Since effective impervious area can be influenced by the pattern, density and drainage connection of impervious surfaces to surface water, and by soil compaction, for example, effective imperviousness may be a more accurate indicator of potential hydrologic changes (Alley and Veenhuis 1983) and ensuing aquatic stresses.

In addition to the work just mentioned, numerous scientific teams continue to investigate landscape—water relationships. Some of this ongoing work is summarized in, Proceedings: 1998 Water and watersheds program review, a progress report on research jointly funded by the National Science Foundation and EPA (NSF and EPA 1998).

The fundamental assumption for building empirical models relating landscape and in-stream factors is that in-stream condition (biology, chemistry, habitat) can be predicted as a function of watershed landscape patterns and practices (forest, agriculture, urban, population, etc.). Mathematically:

in-stream condition = f(landscape factors).

Relationships of this sort will likely differ by ecological region since the sensitivity, resilience, recovery potential and stresses affecting ecological areas with similar environmental characteristics will be different depending on the distribution of soils, geology, vegetation, land form, land use and other factors that are important in defining ecological regions.

Four key challenges are important for further development and use of landscape/in-stream models: