Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
4 Scientific :Issues In Chapter 1, some of the nation's most pressing groundwater issues, along with their social importance, were introduced. This chapter pres- ents most of the same issues, with their corresponding tools or methods, as potential research topics for incorporation into the Ground-Water Re- sources Program (GWRP), and provides recommended actions for the USGS. The issues are the following: making. aquifer management, natural groundwater recharge, groundwater quality and movement in surficial materials, groundwater-surface water interactions, groundwater in karst and fractured aquifers, characterization of subsurface heterogeneity, modeling of flow, transport, and management, and facilitating the use of groundwater information in decision- One common thread that connects all the topics discussed below is the necessity of integrating geochemical investigations into many, if not most, groundwater studies. The committee recognizes that most ground- water problems have a significant geochemical component and that geo- chemistry can often provide important insights into hydrogeologic proc- esses. Historically, most regional groundwater investigations by the USGS have emphasized physical hydrogeology at the expense of geo- 66
Scientific Issues 67 chemical hydrogeology. Yet physical and geochemical problems are usually intertwined, and both affect sustainability. AQUIFER MANAGEMENT Scientific and Management Issues Water managers have the very real problem of trying to project wa- ter use and water supply for a future that includes population growth, climate variability, and unknown technological breakthroughs. They must make decisions about curbing growth, investing in technology, and balancing the various needs of stakeholders and ecosystems. Funda- mental to these decisions are water-budget issues. How much water can be used without drawing down the water table or potentiometric surface, thereby causing Toss of storage, salt-water intrusion, or property damage due to aquifer subsidence? How can managers avoid drying up streams or draining wetlands many of which retain suspended sediment, excess nutrients, and pesticides and maintain wildlife habitat? The sustain- ability of human communities, including the ecosystems that support them, needs to be considered as an integral part of aquifer management. Climate change over decades can also have a major effect on water re- sources, independent of local human influence. Changes in global weather patterns can cause marked changes in precipitation and evapo- transpiration rates and distribution, resulting in changes in recharge, streamflow, flooding, and drought patterns. Although fully understand- ing climate change is a global issue, the USGS has a useful role in as- sisting those predicting climate change at the regional level in the United States. Excessive pumping of groundwater for irrigation and other uses has caused water-level declines of greater than 100 feet in some regions. In addition to causing resource depletion, this reduces pore pressures and raises the effective stress on the aquifer, often leading to irreversible consolidation. Differential settling cracks foundations, which may not only be costly for structures such as roads or buildings, but also hazard- ous to dams, power plants, or pipelines. In some cases, subsidence caused by irrigation pumping has lowered the land surface to the point where rivers have changed course and have flooded agricultural lands. Subsidence caused by overpumping of both water and hydrocarbons has
68 Investigating Groundwater Systems submerged coastal areas below sea level, causing ecological damage to coastal wetlands and exacerbating hurricane damage (White et al., 1993; Kreitler, 1977~. Excessive pumping has caused salt-water intrusion in the majority of U.S. coastal states, including Massachusetts (Person et al., 1998), New Jersey (Pope and Gordon, 1999), South Carolina (Smith, 1994), Florida (Merritt, 1996), Louisiana (Tomaszewski, 1996), and California (Izbicki, 1996~. Even inland areas underlain by formations containing saline wa- ter are susceptible (Sophocleous and Ma, 1998~. Saline groundwater is present in most of the major basins of the United States, and as coastal cities grow, this problem may be expected to get worse. It may take decades before salinity is noticeable in well water and, by then, years may be needed to purge the saline plume even if pumping halts. Injection of freshwater hastens the purging process only slightly, assuming a fresh supply can be found (Kazmann, 1972~. For this reason, it is in the national interest to thoroughly understand the process of salt- water intrusion to assess and manage the risk before damage occurs. The position of the freshwater-saltwater contact can often be esti- mated using the Ghyben-Herzberg principle (Baybon-Ghyben, 1888), which predicts a density-controlled floating lens of freshwater with a "root" approximately 40 times the elevation of the water table above sea level, thinning toward the coastline. However, most salt-water intrusion problems are too complex for simplistic approaches. Wedges of relict freshwater occur far offshore, sandwiched between saline water. Tidal forcing, rainfall events, and storm surges are transient short-term proc- esses that influence salinity. Pockets of relict seawater and intrusion of salt water through failed well casings, joints, or sinkholes complicate the interpretation of salinity in well water. Heterogeneity in aquifer material properties also affects the location of the saltwater-freshwater interface. Such systems are generally studied today using either a sharp interface model (e.g., SHARP; Essaid, 1990) or a variable-density solute transport model (e.g., SUTRA; Voss, 1984~. In addition to modeling, methods used to investigate freshwater- saltwater interactions include tracers (Box 4.~) and geophysical tools, especially electrical methods that are sensitive to conductive water (Rozycki, 1996~. Uncertainty in and scale dependence of material properties and processes plague virtually all measurements of the salt- water-groundwater flux. Aquifer storage and recovery (ASR) technology has recently gained
Scientific Issues 69
70 Investigating Ground water Systems increased interest as a means for aquifer management. Aquifer storage ant] recovery projects involve the artificial storage of water in under- ground aquifers during times of water availability and the recovery of that water when the water is needed (Pyne, 1995~. Most projects involve the subsurface injection of water into aquifers and later extraction of the same water. Example ASR applications to meet aquifer management needs include, among others, seasonal storage of water, emergency stor- age of water, the prevention of salt-water intrusion, enhanced wellfield production, and hydraulic control of contaminant plumes. ASR technol- ogy has been used in various parts of the nation since the late 1960s. The use of ASR poses many technical challenges. These include as- sessing the hydraulic performance of the systems and determining the effects on nearby wells, the Tong-term geochemical changes caused by mixing waters of different chemical compositions in the subsurface, and contaminant migration away from ASR sites. USGS Roles in Aquifer Management The role of the USGS in aquifer management includes collecting, inventorying, and analyzing data on groundwater levels, developing im- proved techniques for acquiring such data, and developing and improv- ing analytical and numerical tools for aquifer management. Potentiometric and water-level maps are a key tool in assessing the effects of regional water use. Such maps were made for the major re- gional aquifer systems as part of the Regional Aquifer-System Analysis (RASA) Program. An unmet need is a national effort to track water lev- els over time in order to monitor water-level declines (Sun and Johnston, 1994~. This is being done on an ad hoc basis by individual states, but the creation of regional potentiometric maps is the responsibility of the federal government. Data to support potentiometric surface mapping are likely to be available from non-USGS entities, especially state geological surveys; the USGS must collaborate with these entities in sharing and interpreting water-level data. Traditional groundwater resources projects are still required in many areas. A growing U.S. population, especially in the arid and semiarid regions, points to the need for the USGS to explore and characterize al- ternative groundwater supplies in areas such as the Great Basin in Ne- vada.
Scientific Issues 71 The exact location and rates of subsidence depend on geology and duration of pumping. Although it is not possible to make exact predic- tions with respect to settlement, it should be possible to improve our per- formance in that area. More effective management will first require better definition of geologic heterogeneity. The USGS should continue to study the relationship between water levels and subsidence, modeling interactions among Ethology, clay content, recharge, pumping, storage, and subsidence. The goal should be to keep subsidence within safety limits for the strain of structures and to identify the critical pumping rate at which there is no permanent strain. For tracking regional subsidence, techniques such as synthetic aper- ture radar interferometry (Massonnet and Feig1, 1998; Amelung et al., 1999) and global positioning systems (GPS) should be fully exploited. These do not require a fixed datum, as does high-precision leveling, and are more cost-effective for large geographic areas. Likewise, arrays of piezometers with transducers should be used to track long-te~ regional changes in the potentiometric surface as an early warning system. Bore- hole tilt-meters or seismographs can be deployed in high-risk areas. The NRC identified the need to analyze links between water re- sources and climate change as one of eight key areas for USGS WRD research (NRC, 1991a). The difficulty of scaling hydrologic models to be compatible with coarse-meshed global circulation models, or vice versa, is a limitation that must be overcome. Predictions under a variety of scenarios must be wedded to decision-making models they must be presenter! in a form useful to water managers and decision-makers. Salt-water intrusion modeling has far to go before it reaches the stage where it can be used effectively by water resources managers. Surprisingly, few test cases exist for independently "verifying" the groundwater codes used for such modeling (Simmons et al., 1999~. Also, although three-dimensional models exist, computation time still limits most real-worId simulations to two-dimensional analysis. A fur- ther challenge is that the nonlinear coupling of the flow and transport equations creates difficulties in their numerical solution. Finally, inte- grated optimization tools are generally lacking, as are linkages to geo- graphic information systems (GIS). Geochemical methods can also use refinement. The radium tracer technique of Box 4.1 has potential for field ground-truthing. Various ratiose.g., CI:Br (Davis et al., 1998) and CI:F (Vengosh and Pankra- tov, 1998) have also shown promise in distinguishing CT from modern
72 Investigating Groundwater Systems seawater, "connate" water, wastewater, road salt, and domestic water- conditioning recharge effluent. It is likely that the USGS will become involved with ASR projects as they influence regional aquifer management. Appropriate roles for the USGS include regional modeling of ASR impacts, investigations of geochemical and hydraulic processes associated with ASR projects, and determination of aquifer properties (transmissivity, storage, heterogene- ity) relevant to ASR performance and design. NATURAL GROUNDWATER RECHARGE Ciroundwater recharge is a critical part of the water budget, and it is arguably the hardest component to quantify. The difficulty in measuring this "income" term in the water budget makes it no less important, espe- cially in arid and semiarid areas. It is also important in coastal areas, where lowering of the water table induces salt-water intrusion into water supplies, and in surficial aquifers, where recharge can carry surface and soil contamination into shallow water supplies. Scientific and Management Issues The critical attributes of recharge are its rate and spatial distribution. Combining the two yields volumetric recharge to an aquifer. In the past, the estimation of recharge rate, particularly in arid areas, was given the most attention. Recharge rates can be estimated using hydroclimatologi- cal approaches requiring measurement or estimation of rainfall, evapotranspiration, soil moisture, and runoff and treating recharge as the residual. Groundwater recharge rates and their spatial distribution can also be estimated using environmental tracers such as dyes, chloride, bromide, nitrogen-15, and chiorofluorocarbons (CFCs), the stable isotopes deute- rium and oxygen-1 S. and the radioisotopes carbon-14, tritium and chio- rine-36 (Clark and Fritz, 1997, pp. 80-99~. These techniques can be ap- plied on both the small scale (Cook et al., 1994) and large scale (Cam- pana and Boyer, 1996~. Agricultural chemicals for which application records exist can also serve as tracers. Recharge can also be estimated by intensive study of infiltration and
Scientific Issues 73 moisture redistribution in the unsaturated zone. Obviously, the scales of these two approaches are drastically different, from tens of kilometers to centimeters. The infiltration approach is physically based and rigorous; however, extrapolation to large scale presents an obstacle. Moreover, although the centimeter-scale process can be modeled using physically based models of the unsaturated zone, linking these models to aquifer models with a resolution of tens of meters or kilometers continues to be difficult. The lack of data to support centimeter-scare modeling of vast areas provides a strong disincentive to reconciling the two scales. Methods of measuring recharge directly have the advantage of inte- grating the sub-centimeter-scare changes to the meter scale. Although the controlling processes occur at the pore scale, they result in a percep- tible movement of moisture that can be measured at the field scale with appropriate field instruments. Noninvasive surficial methods include geophysical methods and remote sensing. Time-domain refractometry and micro-gravity surveys show promise in determining recharge rates (e.g., Young et al., 1997~. The USGS is monitoring micro-gravity at the University of Arizona's network near Tucson in the first basinwide ap- plication of micro-gravity methods to the measurement of changes in groundwater storage. Long-term monitoring, including two El Nino events already, will permit correlation of storage changes with climatic events, facilitating water-use planning and management. Recharge has been interpreted from remotely sensed data with some success. For ex- ample, high-resolution radar images, filtered by principal components analysis, show promise for quantifying the dependence of recharge on climate and topography (Verhoest et al., 1998~. Although the USGS and others have been researching various meth- ods of estimating recharge, the goal of straightforward regional applica- tion has yet to be achieved in most cases. For example, the use of ground-penetrating radar to determine travel times for establishing depth to water is confounded by variations in soil moisture. Unfortunately, surficial methods of recharge estimation will always be difficult because of spatial variability of hydrogeologic mate- rials and soils. For example, using a water-table rise as evidence of re- charge may be misleading if elevated areas are actually areas of lower hydraulic conductivity, because there is not a unique relationship among hydraulic conductivity, head, and recharge. Statistical methods of opti- mizing parameter estimates can be brought to bear on the problem of nonuniqueness, but describing the aquifer heterogeneity is still critical.
74 Investigating Groundlwater Systems In most watersheds, recharge is not spatially uniform because of variations in rainfall, evapotranspiration, infiltration, and runoff. Dis- charge, or negative recharge, may be a natural process occurring in wet- lands or stream valleys, or it may be a result of pumping. In any case, predicting the flow of water within aquifers requires specifying the fluxes of water into and out of the system. Until recently, standard practice was to assume a uniform recharge rate over an entire watershed, and for some purposes this assumption yielded practical results. The assumption becomes increasingly restrictive with decreasing scale and with increasing need for resolution. In 1991, the NBC recommended the development of methods to identify critical recharge areas on small spa- tial scales ~C, 199lb). The need for mapping recharge remains, de- spite locally notable efforts such as Sophocleous (1992~. Although methods have been proposed, they have not been widely used and are complicated by problems of scale. USGS Roles in Groundwater Recharge For regional studies of groundwater, it is essential that the USGS continue to develop and test methods that define recharge at scales ranging from local to regional (Box 4.2~. The required knowledge base includes (1) an improved understanding of basic controlling processes such as evapotranspiration and infiltration, (2) new modeling methods integrating centimeter-scale processes and linking them to large-scare models, including numerical methods for handling nonlinearity in satu- rated-unsaturated models, and (3) methods to measure or average or sta- tistically represent centimeter-scale heterogeneity. Improved knowledge of groundwater recharge will help water man- agers protect aquifer health under stresses imposed by increasing with- drawals or by drought, and it will help them avoid recharging aquifers with poor-quality (contaminated or salines water. From the point of view of health of aquifers regionally, it is critical that studies of recharge make the leap from local, intensive "case" studies to general principles, determining what controls recharge regionally and mapping those factors with a GIS to provide a basis for aquifer management. As appealing as this concept is, efforts to map groundwater vulnerability regionally for management have not always produced practical results. It is important that maps not be too generalized if they are to be useful in local man-
Scientific Issues 75 agement. If decisions about water or land use affect citizens preferen- tially, the map must be detailed enough to resolve local variations in soil, topography, and drainage perceived by an observant citizen.
76 Investigating Groundwater Systems GROUNDWATER QUALITY AND MOVEMENT IN SURFICIAL MATERIALS Over broad areas of the United States, groundwater occurs in shal- low surficial materials. These materials include glacial, alluvial, and lacustrine deposits as well as weathered bedrock residuum. In general, such materials are a few tens to a few hundreds of feet thick and often lie above deeper bedrock aquifers. Surficial materials can be quite discon- tinuous, as exemplified by eskers in the Northeast, or they can be very extensive, such as the tit! sheets in the northern Midwest. Where such materials are composed of permeable sand and gravel, they often form important aquifers. However, materials of Tower permeability, such as clayey till or silty lacustrine deposits, also contain and transport groundwater ant! have important functions in the overall water cycle. Scientific and Management Issues Occurring near the land surface, groundwater in shallow surficial materials is particularly vulnerable to contamination (see Chapter I, Box 1.~) by the hundreds of thousands of reported releases of gasoline from leaking underground fuel tanks nationwide, and the nation is currently spending hundreds of millions of dollars remediating contaminated sites in these materials. The USGS National Water-Quality Assessment (NAWQA) Program discovered many instances of nitrate and pesticide contamination of shallow groundwater in agricultural areas (USGS, 1999b). Likewise, onsite septic systems and lawn fertilization can also contaminate groundwater. Shallow groundwater contamination can move to adjacent lakes, rivers, and wetlands as well as to underlying deep aquifers used for water supply. Concern for the integrity of groundwater supplies has led to legisla- tion at all levels of government to protect aquifers from contamination by land use, much of it under welIhead protection clauses. Despite great effort expended on predicting how water and contaminants move under- ground, it is still difficult to state with confidence that a given land use will have a specific impact on a particular water supply. Clearly, though, regional deterioration of shallow water supplies has occurred and can be linked to land-use practices, with an example being agricul- tural fertilizers causing high nitrate levels in rural water supplies. Much
Scientific Issues 77 of the uncertainty in pinpointing sources arises from aquifer heterogene- ity and complexity, confounding welThead protection analyses. USGS Roles in Surficial Material Hydrogeology Knowledge of groundwater movement, quality, and quantity and of potential contaminant sources and pathways is needed to effectively manage surficial groundwater. Public-sector managers need water qual- ity and water-level monitoring information in areas geologically suscep- tible to degradation or characterized by high-risk land-use practices, and they need the educational tools to enlist the support of the public in their own self-interest. Where prevention fails or contamination has already occurred, the managers need tools to restore the water supply. The USGS is providing, through NAWQA, a regional assessment of water quality. The USGS is strongly encouraged to continue developing accu- rate and accessible information about groundwater using available tech- nologies. Aquifer restoration has developed to a high degree of sophistication in the private sector, given economic incentives driven by public regula- tions. However, because the public interest is at stake, impartial assess- ments of restoration technology are neecled. In particular, the cost and difficulty of aquifer restoration (e.g., difficulty in delivering nutrients and oxidants to contaminated zones for in situ bioremediation) has led to a closer look at intrinsic bioremediation in contaminated surficial aqui- fers (Barber, 1994; Smith et al., 1994~. The process of intrinsic biore- mediation, or natural attenuation, needs to be quantified before regula- tory decisions can be made (Chapelle, 1999~. Because the U.S. Envi- ronmental Protection Agency (EPA), other federal agencies, and univer- sities are also active in this area, study sites should be chosen carefully for their regional applicability. Setback requirements and other guiclelines for accepted natural- attenuation systems such as septic tanks must be reviewed and revised in the face of developments in the understanding of pathogen transport and lithologic controls. The transport of microbes can be quite variable. Sorption of pathogens is microbe-specific and depends on surface chemistry and flow velocity (Harvey, 1993; Hendry et al., 1999~. The USGS should be more involved in studies of natural attenuation, just as it has been involved with studies of contaminant transport at several well-documented sites.
78 Investigating Groundwater Systems GROUNDWATER-SURFACE WATER INTERACTIONS Scientific and Management Issues The USGS has long been a leader in the quantification and charac- terization of surface water and groundwater interactions. Winter et al. (1998) documents not just the role of the USGS, but also the evolution of knowledge in the field in general. Earlier works by Winter (1976, 197S, 1981) led to major advances in the science of groundwater-wet- land and groundwater-lake interaction. Groundwater and surface water interaction occurs in all types of hy- drogeologic and cTimatologic settings at all spatial and temporal scales (Winter et al., 1998~. Groundwater can interact with streams, lakes, es- tuaries, bays, wetlands, and coastal areas. The interconnection between groundwater and surface water means that they often behave as one res- ervoir and should be treated and managed as a single resource (Winter et al., 1998~. Groundwater, Lakes, and Streams. Early views of groundwater- stream interactions often treated streams simply as areas of recharge to groundwater or recipients of discharge from groundwater; groundwater flow paths in the vicinity of surface water bodies were generally two- dimensional and simplistic (Woessner, 1998~. Interactions between groundwater and streams or lakes were often characterized in terms of base flow or bank storage, the latter being recognized as an important storage and flood-wave attenuation mechanism (Whiting and Pomera- nets, 19971. Interactions in coastal regions were often cast in the context of saTine-water intrusion (Todd, 1959; Davis and DeWiest, 1966; Freeze and Cherry, 19791. Flow paths in heterogeneous aquifers near surface water bodies are far more complicated and dynamic than was previously thought (Woess- ner, ~ 998; Wroblicky et al., ~ 998~. To the simple "gaining" and "losing" streams (and other surface water bodies) can be added "zero exchange" and, perhaps more significantly, "flow-through" streams (Woessner, 1998~. Temporal changes in water and chemical exchange between groundwater and surface water occur because of variations in stream stage and discharge (Squiliace, 1996; Won~zell and Swanson, 1996; Morrice et al., 1997), which sometimes occur very rapidly (Wroblicky et al., 1998~.
Scientific issues 79 The interface between groundwater and surface water systems is recognized as a distinct zone, the hyporheic zone (Gibers et al., 1990; Vervier et al., 1992~. This refers to a region where active and dynamic exchange of water and nutrients occurs between the surface water and the adjacent groundwater system (Triska et al., 1989; VaTett et al., 1997; Wroblicky et al., 1998~. The hyporheic zone is an "ecotone" that allows for bi-directional flow of organisms and materials (Gibers et al., 1990; Vervier et al., 1992~. Temporal variation in the extent of exchange between a stream and associated aquifer is driven not only by the changing characteristics of the stream, but also by changes in aquifer status as well. The fluvial system includes the floodplain and hiTIsiope where infiltration and re- charge alter the hydraulic gradients linking the stream and aquifer. In this way, the variation in exchange reflects the combined effects of hy- drologic change in the stream and aquifer. Hyporheic studies, with synergistic collaboration among geologists, hydrologists, limnologists, ecologists, and geochemists, are becoming increasingly common (Triska et al., 1989; Harvey and Bencala, 1993; Stanford and Ward, 1993; Valett et al., 1996; Morrice et al., 1997; Dahm etal., 1998~. The scientific contributions of the USGS groundwater-lake interac- tion research program are numerous. They notably include detailed in- strumentation studies designed to determine how to calculate wetland and lake water budgets and their associated measurement errors (e.g., Rosenberry et al., 1993; Winter et al., 1995; LaBaugh et al., 1997), theo- retical and field investigations of wetIand-groundwater hydraulics (e.g., Winter and Woo, 1990; Harte and Winter, 1995; Winter, 1995a,b; Ro- senberry and Winter, 1997; Winter et al., 1998), investigation of wetland biogeochemistry related to climate change (e.g., McConnaughey et al., ~ 994; Schwalb et al., ~ 995; LaBaugh et al., ~ 996; Poiani et al., ~ 996) and, most recently, a multidisciplinary program incorporating hydrology, geochemistry, and ecology in a complex of lakes in Minnesota (Winter, 1997~. The NAWQA Program, a nationwide study of surface water and groundwater quality trends and their cause-and-effect relationships, seeks to treat groundwater and surface water as a single unit, where ap- propriate. The USGS also has contributed greatly to stream-groundwater inter- action. Bencala and Walters (1983) formulated the concept of "tran- sient storage" and its application to streams. Triska et al. (1989) studied
80 Investigating Grounclwater Systems the ecological consequences of groundwater and surface water exchange and its impacts on nitrogen transformations along stream corridors. Harvey and Bencala (1993) and Harvey et al. (1996) applied different hydrogeologic techniques to address how inferences on surface and groundwater interaction depend on the scale of measurement. Groundwater and Wetlands. The National Research Program of the WRD has historically supported wetland research, its first major contributions being in the field of wetlands classification (e.g., Cowardin et al., 1979; Novitski, 1979~. This research thrust has been expanded to include remote sensing and ecological research in wetlands focusing on how the distributions and kinds of submerged plants (macrophytes) may be hydrologic surrogates useful to identify water quality parameters such as salinity, turbidity, pH, nutrients, presence of various pollutants, or frequency and duration of inundation (e.g., Carter et al., 1983; Batiuk et al., 1992; Tennyson et al., 1993; Carter et al., 1996~. Wetlands, as places where groundwater and surface water interact, are similar to hyporheic zones with respect to their being ecological and hydrologic interfaces between freshwater- and marine water-rich and water-poor landscapes. Compared to nonwetIand areas, the unique chemical characteristics of wetland soils, their ecological community compositions, and the relative degree to which their soils are saturated vary widely across the nation's climatic regimes. The minimal essential characteristics of a wetland are (1) recurrent, sustained inundation or saturation, and (2) the presence of physical, chemical, and biological features reflective of this condition. Common diagnostic features are hydric soils and hydrophytic vegetation (NRC, 199Sb). Hydric soils ac- cumulate more organic matter than do upland soils because organic matter cannot readily decompose during the anoxic conditions that de- velop when wetland soils are saturated. Determining how long and how frequently soils need to be saturated to form an ecologic wetland community is very difficult. The NRC (1 995b) recommended a series of regional studies to address "how wet is wets to make a wetland while still maintaining the overriding ecosystem- based view on wetlands characterization that typified earlier cIassif~ca- tions. Winter (1988) and Brinson (1993) challenged the long-standing view that wetlands are fundamentally ecosystems. They argued instead that wetlands are physiographic places with hydrologic and meteoro- Togic conditions that lead to more water, which in turn leads to a unique
Scientific Issues 81 ecosystem evolution and unique soils over time. Brinson's hydromor- phic wetland classification is largely based on water budgets and has generated wide interest in the wetland scientific community. Brinson's and Winter's classifications both fully incorporate riparian areas as wet- lands. The renewed scientific emphasis on wetland hydrology has led to fruitful research designed to understand the complex interactions among wetland ecosystems, geochemistry, and hydrology. Driving some of this research are National Science Foundation initiatives on climate change and biodiversity maintenance and the regulatory need to manage wetland resources. Wetlands are a major terrestrial sink for carbon (as peat) and are significant sources of the atmospheric greenhouse gases of carbon dioxide and methane. Wetlands are also critical landscapes that main- tain marine fisheries and freshwater bio`diversity (Mitsch and Gosselink, 993; NRC, 1 995b). Research on the interactions among hydrology, geochemistry, and biology in wetlands has led to a better appreciation of solute cycling and transport. For example, we now know that (~) nutrients are processed and methane is generated by bacteria in distinct wetland soil microenvi- ronments, best identified by sampling at the centimeter scale (e.g., Ro- manowicz et al., ~ 995; Hunt et al., 1997; Steinmann and Shotyk, ~ 997), (2) the extent to which this methane is generated may be related to cli- mate change (Romanowicz et al., 1995), (3) wetland soils have signifi- cant macropores through which solutes can move by groundwater ad- vection (e.g., Chanton et al., 1995), and (4) trace metal contents, pollen distributions, and isotopic contents of peat profiles provide important information on Holocene climate change and anthropogenic contamina- tion (e.g., Shotyk et al., ~ 998~. Finally, the coupling of GIS technology with data on stream-water quality and discharge at the landscape scale has led to the conclusion that headwater depression and other small wetlands proportionally re- move more suspended material and store more water than do rip arian wetlands further downgradient in higher-order stream reaches (Johnston et al., 1990~. This kind of research has broad implications with respect to how we manage our natural environment. Together, the detailed studies of wetland hydrology, geochemistry, and biota and the more regional studies of wetland function in the land- scape have fostered an almost entirely new scientific discipline of multi- disciplinary study, complete with new journals such as Wetlands, pub- lished by the Society of Wetland Scientists.
S2 Investigating Groundwater Systems USGS Roles in Groundwater-Surface Water Interactions The central role of groundwater-surface water interactions in re- gional assessment suggests a number of key research areas where the USGS expertise in process-based groundwater science is essential to support the analysis and sustainable management of regional groundwa- ter systems: Water Availability Models. Water Availability Models (WAMs)- complex models that have their roots in conjunctive-use modelsseek to simulate the flow and legal availability of surface and subsurface water in a basin, region, or political subdivision. in their most sophisticated form, WAMs also include economic, sociological, water quality, and legal models embedded within the hydrologic models. Depending on the kinds of input required, it may be appropriate for the USGS (~) to generate or apply these models directly, (2) to work in cooperation with other organizations, or (3) to be the supplier of infor- mation to third parties. WAMs are especially useful in simulating water distribution sce- narios during droughts and in managing excess water. Targeted to water rights, permitting, and planning, groundwater-surface water interactions in most water-availability models are at best represented with simplified loss coefficients, representing average estimated losses and returns (TUNIC, 1997~. In many basins, regional permitting and planning deci- sions need to accurately incorporate the complex interactions between surface and subsurface flows in water-availability modeling. This growing need represents a timely opportunity for the USGS to integrate its expertise in process-based regional groundwater science with regional assessments of sustainable surface water and groundwater resources. Groundwater and the ecology of surface systems. The USGS has been one of the leaders in the important area of groundwater and the ecology of surface systems. The influence of the physical and geo- chemical characteristics of the coupled surface-subsurface hydrologic system on the associated ecosystems, and vice versa, is not well under- stood. Organisms in the hyporheic zone may be important indicators of ecosystem health; this avenue merits further exploration. The USGS is encouraged to engage in integrated physical, geochemical, and isotopic studies of these interactions.
Scientific Issues 83 Watershed planning and management. Principles of groundwater- surface water interactions must be integrated into the increasingly com- mon watershed-based approach to planning and management. Ground- water flow systems will often not coincide with watershed boundaries, adding to the complexity in evaluating flow paths, source areas, and dominant processes controlling streamflow, contaminant transport, and nonpoint pollution. Storm flows can be dominated by subsurface flow (Barnes, 1939; Anderson and Burt, 1980) with the chemical signature of "old water" (Sklash and Farvolden, 1979; Rice and Hornberger, 1998~. The origins and transport of pesticides in streamflow are not easily re- solved without detailed understanding of the response time and of inter- actions between direct surface runoff, bank storage, and subsurface flow (SquiTIace et al., 1993~. Groundwater-surface water interactions em- body critical processes, essential for understanding basin hydrology and watershed-scare fate and transport, developing watershed-scare manage- ment plans, and calculating total maximum daily Toads. GROUNDWATER IN KARST AND FRACTURED AQUIFERS Scientific and Management Issues Unlike many of the nation's major aquifers that owe their productiv- ~ty to primary (i.e., intergranular) porosity, "karst" aquifers such as the Floridan and Edwards aquifers are characterized by secondary porosity due to dissolution, usually of carbonate rock, along pores and fractures. Fractures in less-soluble, low-permeability rocks such as granite, gneiss, shale, and clayey till may also yield aquifers that are relatively poor but regionally important because they are a region's only water supply. These karst and fractured aquifers present a variety of challenges to quantitative hydrogeology. Karst Aquifers. Of all geologic settings, karst may be the most susceptible to groundwater contamination because water infiltrates rap- idly, the aquifer has poor filtering ability, and residence times are short (Figure 4.1~. Many regions of the United States, especially the midwest em and southeastern states, have groundwater problems associated with karst. Flow velocities can exceed 1,000 m/day, and dissolved sub- stances, microbes (E. cold and Giardia cysts), sewage or manure parti-
84 Investigating Ground water Systems FIGURE 4.1 Greer Spring in the Missouri Ozarks is the largest spring on U.S. Forest Service land, and averages 187,000,000 gallons per day (courtesy of Randy Orndorff, USGS). cles, and even crustaceans can move through conduits in karst; apertures are sufficiently large that sediment can be transported with adsorbed toxic metals or organic molecules. The use of karst sinkholes for rural waste disposal puts water supplies in rural areas at special risk. In regulating septic setbacks, it is not clear that ambient mechanisms in fractured or karst aquifers can meet regulatory requirements (Chapelle, 1999~. Effective management means protecting the actual source area for potential water supplies, as is done in Florida by creation of conservancy recharge parks. Mapping recharge areas, however, is not straightfor- ward. Recharge areas are controlled by fractures that create anisot- ropy a strong preferred flow direction elongates the source area for a well or spring. Solutions widen certain fractures to create conduits, the locations of which are hard to determine. In fact, our ability to represent the architecture of actual karst systems is far from satisfactory. Although conceptual models of flow in karst areas have been con- structed (Bogli, ~ 980), numerical modeling of karst has either treated the conduits as discrete fractures, assuming no matrix flow, or has assumed
Scientific issues 85 the whole system to behave as a "black box." In the latter approach, re- charge is the input and spring flow is the output. Modeling of karst aq- uifers as equivalent porous media may be feasible if the cell size is suffi- ciently large, and such a model provides general information on flow direction. However, head, permeability, and water quality measured in wells cannot be expected to be representative. Traditional tools of the hydrogeologist often yield ambiguous data. Placement of monitoring wells is critical, and interpretation of pumping tests, head values, and geochemistry is difficult because the concept of the water table breaks down in this highly transient, dual-porosity system. Methods need to be developed to either represent the true aquifer complexity or, if this proves infeasible, to sample the aquifer parameters in some integrated fashion. Fractured-Rock Aquifers. Fractures in otherwise impermeable rock present some of the same problems as karst conduits. The bulk hy- draulic conductivity of a fractured aquifer may be Tow, but flow veloci- ties and contaminant transport rates in fractures or solution cavities can be high, resulting in early contaminant appearance at target wells or streams and in poor filtration of pathogens. Locations of fracture zones are extremely important to the development of water supplies in areas with generally unproductive aquifers. Fractured zones may be concen- trated along lithologic contacts or near the surface in the case of stress- relief fractures. They may cluster around regional fault traces. Predic- tion of the occurrence of fracture zones based on tectonic or structural models would assist in water-supply development. Knowledge of frac- ture-flow dynamics and fracture orientation and location is critical be- cause fractures may cause markedly anisotropic behavior, especially in contaminant transport. There has been an evolution in the understanding of basic processes in fractured aquifers paralleled by developments in the mathematical simulation of fractured systems, and much of this work has been con- ducted by USGS scientists (e.g., Shapiro and Hsieh, 1996~. Early mod- els represented fractures as parallel plates of varying aperture, but recent models accommodate two- or three-dimensional networks of fractures with statistically distributed apertures and spatial correlation of aperture width. Representation of fractured-aquifer porosity has evolved from fracture porosity to a dual-porosity representation including matrix po- rosity, and even a fracture-skin porosity (Robinson et al., 1998~. Diffu-
86 r Investigating Ground water Systems sign, adsorption, and degradation of contaminants in fractured aquifers can be modeled. Realistic-Iooking fracture networks based on a growing number of field studies of fractured aquifers can be modeled, although there is still a need for a geostatistical description of fractured systems. Modeling continues to pose many challenges, the greatest of which is the data deficiency relative to the needs of a good predictive fracture- flow model. Practical conceptual models of fractured systems are still needed. For example, in a sedimentary sequence in New Jersey, shallow flow was found to occur via bedding-plane partings, but deeper flow oc- curred via high-angle fractures (Morin et al., 19971. We still know little about the control of the compressive-tensile stress regime on the frac- tured-flow system, and there is an accompanying data need for sound management models of fractured systems. Determining reactions along groundwater flow paths requires a sound conceptualization of the hydrogeochemical system, including ve- locities, residence times, surface area, oxidation state, availability of particles such as clay and organic matter, and microbes. Advection dominates transport in fractures; diffusion dominates in blocks (Moench, 1995~. Contaminant pathways may be complex, irrespective of averaged heads and gradients. Natural attenuation depends on the complex reac- tions expected in these highly buffered systems, including dilution, re- dox reactions, dissolution, precipitation, adsorption/desorption, com- plexation, and ion exchange. USGS Roles in Karst and Fractured-Rock Studies Representing the architecture of a small number of actual karst or fractured-flow systems via intensive studies is a prerequisite to under- standing these systems. Such studies should aim at developing methods to either represent the true aquifer complexity (perhaps as a geostatisti- cal description of fractured systems) or to sample the aquifer parameters in some integrated fashion. The USGS Geologic Division should be re- cruited to integrate tectonic and structural models with hydrologic mod- els, in order to predict the occurrence of karst or fracture zones. In par- ticular, how does the compressive-tensile stress regime relate to the fractured flow system? Neotectonically active zones are primary candi- dates for preferential flow paths in karst, for example. A sound concep- tualization of the hydrogeochemical system in fractured or karst regions
Scientific Issues 87 is needed as well. Specifically, research on the fate and transport of nu- trients and microbes in karst aquifers is urgently needed. Drilling, sampling, and flow-determination protocols need to be tested and developed. For example, in drilling, borehole location rela- tive to fractures is critical: well productivity is dependent on proximity of the well to fracture-correlated lineaments or faults (Allen and Michel, 1998; Mabee, 1999~. Boreholes should be fracture-oriented, with frac- ture locations confirmed by trenching before drilling and by down-hole photography and caliper Togs after drilling. Drilling has been found to dilute contaminants in fractures, while pore water is relatively unaf- fected (McKay et al., 1998~. Because of this, some researchers suggest that the extent of contamination is better measured by pore water rather than by fracture water concentrations (McKay et al., 1998~. Drilling has also caused cross-contamination, and some researchers use dry-augering to minimize this problem (Nativ et al., 1999~. Improved aquifer-test analytical methods are needed for fractured- rock systems. In quantifying the heterogeneity of a fractured system with extremely variable well yields, a test of over 4,000 wells showed that Ethology accounted for 12 percent of the variation. Well construc- tion and aquifer-test duration accounted for 24 percent, with 64 percent of the variation presumably due to (fracture) permeability variation (Knopman and Hollyday, 1993~. This test underscores the need for testing and developing field protocols for fractured aquifers. Methods are needed to locate transmissive fractures. Fractured zones that intersect boreholes can be detected by acoustic televiewers, direct-current resistivity, nuclear magnetic resonance, seismic tomogra- phy, electromagnetics, heat-pulse flowmeters, tracers (borehole-to- borehole or single borehole), straddle-packer pump tests, and borehole radar. Heat-puise flowmeters and thermometers can show which frac- tures are actively transmitting water even at very Tow velocities. The acoustic televiewer, which works in muddy water, discriminates bed- ding-plane and high-angle fractures and reveals conjugate joint sets (Morin et al., 1997~. Surface geophysical methods including airborne electromagnetism or infrared airborne spectroscopy may be useful for mapping fracture trends or large individual fractures. Large fractures or fracture zones can be detected by analysis of linear features in air pho- tos, satellite images, and digital elevation model data, using filtering to enhance lineaments. These methods need to be refined and tested, in- cluding development both of technology and protocol.
8s Investigating Groundwater Systems ~ determination of flow paths, tracers are a promising method for determining fracture-flow. Isotopes, dyes, dissolved chemicals, bacte- ria, and even lanthanide-labeled clay have been used successfully. Ma- trix porosity and fracture aperture can be determined with accuracy and are relatively insensitive to type of tracer experiment (Himrnelsbach et al., 1998~. There is a need for tracer-test protocol. For example, in- duced-gradient tracer tests may underestimate the importance of disper- sion relative to advection because under low-velocity and Tong-resi- dence-time natural conditions, dispersion dominates transport (Raven et al., 1988~. There also needs to be a testing and cataloging of suitable tracers, including natural or isotopic tracers. Parameter-estimation mod- els of fractured systems will help direct data collection; these models have shown that permeability is a poor estimator of fracture aperture, but that flow velocities and tracer breakthrough times are good estimators of aperture (Tsang et al., 1988~. CHARACTERIZATION OF SUBSURFACE HETEROGENEITY Aquifer heterogeneity arises from the complex history of geologic deposition, erosion, lithification, and tectonic deformation of rocks. The importance of heterogeneity to groundwater occurrence and movement is apparent in the wide range of hydraulic conductivities commonly ob- served from ~ 0~~ ~ to 1 o2 cm/see (Freeze and Cherry, ~ 979~. Given this range, the determining characteristic of an aquifer in controlling fluid movement is its hydraulic conductivity distribution, or heterogeneity. Despite its importance, characterizing heterogeneity remains elusive. Scientific and Management Issues The need for better characterization of heterogeneous aquifers is driven by scientific and public needs for groundwater protection and remediation. Classic hydrogeology has often described aquifers only in teas of bulk hydraulic characteristics (transmissivity, storage coeffi- cient, and porosity) that are relevant to groundwater resources issues. The RASA models, which combined many complex stratigraphic units into a few conceptual layers, are examples of this approach. However,
Scientific Issues ~9 bulk properties are rarely, if ever, adequate to determine flow paths and travel times necessary for contaminant transport studies or welIhead protection. instead, a more detailed knowledge of the distribution of hydraulic properties is critical. Efforts to cope with heterogeneity fall into three categories. First, there have been attempts to map heterogeneity by intensive drilling and geophysical surveying. Second, some researchers have attempted to logically relate rock or soil properties to the depositional process, using geologic facies architecture. Facies modelsconceptual models of the expected distribution of facies based on the geologic depositional history of an areacan be used to define hydrostratigraphic units (Maxey, 1964; Seaber, ~ 988; Anderson, ~ 989~. The petroleum industry interprets relatively scarce borehole data and abundant "soft" data such as three- dimensional seismograms using facies models. Third, heterogeneity has been treated as a stochastic process, initially as a purely random distri- bution of properties, more recently adding realism with correlation, non- stationarity, and nonrandomness. Predicted hydraulic conductivities Took increasingly plausible with these advanced methods, but they still need to be conditioned with in- formation including "soft" data (electrical resistance tomography, seis- mic tomography, radar tomography, etc.~. An abundance of small-scale data are required to detect the underlying stochastic processes for a vari- ety of geologic settings. Detailed studies are needed at heterogeneous sites such as those at the MADE (MAcroDispersion Experiment) site in Mississippi. Stochastic process models will have to be incorporated into facies models to cope with the nonstationarity that appears at the large scale. In the past, detailed characterization usually was not attempted because numerical models, the fundamental too] of modern hydro- geologic prediction, were largely unable to handle this complexity. This situation has changed with the advent of fast, inexpensive computers and improved modeling codes. Many hydrogeologists have encountered the so-called "scale effect" of hydraulic conductivity (K), which suggests that the effective K of a given material varies with the scale of either the testing method used or the field problem being addressed (Hsieh, 1998~. For example, a small- scale contamination study might collect field data and interpret hetero- geneity based on wells located only a few meters or tens of meters apart. For a subregional groundwater model (for example, for a small town), heterogeneity might be studied on the scale of hundreds of meters. What
9o Investigating Groun~lwater Systems is the effective K in these two cases? The question pertains to both the method of measuring K (aquifer test vs. slug test, for example) and the appropriate K to use when simulating aquifer behavior with a numerical model. The number of papers published on the topic of scale since the early 1990s shows there is growing interest in this topic. Different in- vestigators have examined possible causes of the scale effect in several ways, including field-testing and modeling studies. However, there is no consensus on the causes of the effect or on factors that might control its magnitude. Indeed, some hydrogeologists claim there is no physical ba- sis for the scale effect (Butler et al., 1996~. Because heterogeneity results from small-scare (and larger) proc- esses, understanding these processes requires a microscale investigation. Paradoxically, the results will eventually be applied at a larger scale, especially in numerical modeling. So in addition to needing methods to define small-scale features, methods are needed to realistically represent these processes at larger scales. USGS Roles in Characterization of Subsurface Heterogeneity The USGS should continue studies of groundwater in a variety of complex settings to reveal important principles and processes controlling water supply and quality. The Survey should also continue its inventory of aquifer properties in order to develop regional databases. The science is by no means complete, as is evident from new developments in the understanding of natural attenuation of contaminants. Translating lithostratigraphy to hydrostratigraphy rests on a foundation of detailed hydrogeologic studies at representative sites such as the Cape Cod toxic waste research site. Detailed studies at sites representative of important (common or especially susceptible to damage) hydrogeologic settings should continue and should be encouraged. It is important, however, that the significance of these studies for generalizing the results to broader areas be understood and emphasized by the USGS and stressed in its reports to the public. The USGS must justify the investment of resources at these intensive-study sites. In terms of regional groundwater investigations, there is a need for better integration among geologic disciplines: hydrogeology, stratigra- phy, sedimentology, and structural geology. The USGS should continue to develop methods of deducing hydrologic information from geologic
Scientific Issues 91 models and geophysical methods (Jorgensen, 19884. The current Middle Rio Grande basin projects illustrate how this integration can be done successfully. Currently, however, there are no generally accepted "rules" or measures of heterogeneity and its importance; developing such measures would be a fruitful area for research. There are many possible research directions for the improved simulation of the spatial heterogeneity of aquifers (geostatistical models, fractal methods, and process models). Other areas for investigation include better integration of subsurface stratigraphy with hydrogeology, innovative geophysical tools (down- hole logging, geotomography, flowmeters, radar, etc.), measurements of hydraulic parameters such as hydraulic conductivity at a variety of scales, and correlation of these measurements with stratigraphic facies. Tracer experiments, especially experiments that test/verify fieldwork and modeling experiments in heterogeneous aquifers, are needed. It should be noted that the Cape Cod and Borden tests, which have become litera- ture classics, were both conducted at relatively uniform sites. NUMERICAL MODELING Scientific and Management Issues During the last two decades, numerical modeling has become stan- dard practice in most groundwater studies. Better modeling codes, faster and cheaper computers, and user-friendly interfaces have put sophisti- cated modeling within the facilities and budgets of most groundwater projects (Figure 4.2~. However, these advances are a mixed blessing. A 1983 editorial titled "Groundwater Modeling: The Emperor Has No Clothes" (Anderson, 1983) examined the pitfalls of using sophisticated groundwater-flow models without a clear understanding of the modeling process and/or without proper data and model calibration. A follow-up abstract titled "Modeling Complexity: Does the Emperor Have Too Many Clothes?" (Anderson and Hunt, 1998) discussed what has hap- pened to groundwater modeling in the intervening 15 years. The prolif- eration of model add-one such as pre- and postprocessors and various optional packages (transport, streamflow routing, lake interactions, evapotranspiration, etc.) has made extremely complex models compara- tively easy to construct. Such complex models offer a false sense of ac-
92 . .. ....- ,~ .~ ~ i ., ., ~ a.... an. ~ l ~ ~ i. - Ct F: T ~ ~ ~ - -am- --; ~.~.~.. ~.~ ~. -I- ....... ' t Y~.~-:.~.~ ~ ~ I :L,... - - .... ~ ~ _: i. ii..~. . Ct · _ Cal a' .......... a, ~ a. o ~ Cal a.. i...... ~ U:) .~ Cal Fiji. ,~jjjii,. au Cal o sol o sly Ct C) O X ~0 o a' ~ 5-, a; Cal X ~ ._ t . U) Ct .O
Scientific Issues 93 curacy and precision if the model complexity cannot be supported with appropriate field information and the model uncertainty is not quanti- f~ed. The ability to evaluate uncertainty and sensitivity an important re- cent trend in mode] developmentaddresses concerns about misreading model results. Parameter estimation codes such as UCODE and MODFLOWP (Poeter and Hill, 1998) allow modelers to estimate opti- mum sets of model parameters, consistent with field data, and they also provide rigorous measures of the sensitivity of the model solution to changes in parameters. Such uncertainty analyses improve models as tools for decision-making. Aquifer management-optimization codese.g., AQMAN (Lefkoff and Gorelick, 1986), AQMAN3D (Puig et al., 1990), and various com- mercial products are a significant step forward in decision-making. Such codes provide optimal groundwater-management solutions, such as the most favorable well placement or pumping rates, under various physical and economic scenarios. They enable, for example, a munici- pality to maximize groundwater extraction subject to the limitation that heads near a sensitive stream or hazardous waste site do not fall below a threshold value. Traditional methods of measuring and modeling flow in porous me- dia are being used only cautiously in fractured-rock systems. Significant advances have occurred in the understanding of fractured-rock hydro- geology (NRC, 1996~. Most water movement occurs through open fractures, while most storage occurs in the porous matrix. A number of analytical models (e.g., Moench, 1995) now exist for such dual-porosity systems, while sophisticated numerical codes such as FracMan/MAFTC (Golder Associates, 1987) allow evaluation and simulation of discrete fracture networks using stochastic techniques. Field methods are also being developed to characterize these aquifers. One of the major im- pediments to progress in fractured-rock hydrogeology is a lack of well- characterized field sites for model evaluation. The USGS fractured-rock hydrology research site at Mirror Lake, New Hampshire (Shapiro and Hsieh, 1996), is one of only a few such sites in the United States. USGS Roles in Numerical Modeling The USGS has a strong history of innovation and achievement in the
94 Investigating Groun~lwater Systems development of fundamental groundwater models such as MODFLOW. Efforts should continue in conceptual and theoretical aspects of numeri- cal modeling (flow, reactive chemical transport, management), espe- cially in the near-surface environment, so that increasingly sophisticated models will be available to help diagnose cause-and-effect relationships and perform predictive simulations. However, the committee strongly feels that, in the context of flow modeling, the USGS should devote its efforts to conceptual and theoretical breakthroughs rather than fine- tuning or developing graphical interfaces for codes like MODFLOW. Such work is already being done by the private sector (e.g., Visual MODFLOW, Groundwater Vistas). In the context of regional groundwater investigations, the USGS should continue to develop appropriate conceptual and numerical "framework" models covering large geographic areas, and it should de- velop the means for focusing or telescoping these models to smaller scales. The recent work on telescopic mesh refinement (Leake and CIaar, 1999) provides examples of such techniques. In addition, analyti- cal element (AK) models can be used for scaling from regional to local simulation. A far-field AE model can be used to develop boundary con- ditions for a local finite-difference model. Analytical element models have the added advantage of allowing exploration of a model's sensitiv- ity to boundary conditions, an important step that is rarely done (Hunt et al., 1998~. FACILITATING USE OF GROUNDWATER INFORMATION IN DECISION-MAKING Investigation of these regional issues must provide useful infor- mation to water resources managers and decision- or policy-makers. This section discusses three ways that the USGS can assist in this pro- cess: (1) by promoting the use of information from USGS studies in decision-making by quantifying and reducing uncertainty in predictions, (2) by scaling results of local studies to the regional level, and (3) by assisting in the development of decision-making and risk models that incorporate groundwater information. The WRD's mission statement clearly emphasizes the need to actively disseminate hydrogeologic data and reports to the public: The mission of USGS Water Resources Division (WRD) is "to pro-
Scientific Issues 95 vice reliable, impartial, timely information that is needed to understand the nation's water resources. WRD actively promotes the use of this information by decision-makers to · Minimize the loss of life and property as a result of water-related hazards, such as floods, droughts, and land movement. · Effectively manage groundwater and surface-water resources for domestic, agricultural, commercial, industrial, recreational, and ecologi- cal uses. · Protect and enhance water resources for human health, aquatic health, and environmental quality. · Contribute to the wise physical and economic development of the nation's resources for the benefit of present and future generations." (USGS, 1999c). Quantifying and Reducing Uncertainty in Predictions Predictions about groundwater systems are always subject to uncer- tainty as a result of spatial and temporal variability in subsurface prop- erties and processes. Additional uncertainty arises from attempts to characterize the subsurface based on limited and possibly imprecise measurements. Although uncertainty is an integral part of groundwater systems, past models, measurements, and predictions have not always explicitly identified the associated error. Future groundwater predictions should specifically include an asso- ciated quantitative error. One benefit of estimating error is an improve- ment in decision-making. Error estimates allow decision-makers and others to understand that hydrologic variables can take on a range of values, facilitating the development of options that will meet objectives under various scenarios. Thus, reporting errors in hydrologic variables should lead to more robust decisions. Associating uncertainties with predictions and measurements also provides a rational basis for future data collection efforts. Understand- ing uncertainty and its source allows development of sampling plans that will result in the greatest reductions in uncertainty subject to fiscal and other constraints. Parameter-estimation modeling provides a measure of uncertainty in predictions that is badly needed. Parameter-estimation modeling should
96 Investigating Groundwater Systems become standard practice, especially when models are used as a basis for water resources decisions. For example, what is the probability that monitoring will detect contamination? If the uncertainty in model re- sults is unacceptable, as it may well be, strategies are needed to diminish that uncertainty. Scaling Available Information to the Regional Level How can information that has already been collected at a variety of scales be used in regional-scale studies? Data from past studies are likely to be available on many different scales in new regions of interest to the USGS. Data from smaller-scale or local groundwater studies are likely to have been collected in the past by the Survey and others, and some regional-scare information may be available as well. For example, saturated hydraulic conductivity data may be available from permeame- ter tests on sediment samples, slug tests, and aquifer tests. Regional studies will require data collection on regional scales, since many hydrogeologic variables depend upon the measurement scale. However, it makes sense for a regional study to incorporate smaller- scale data previously collected within the region. In regions or parts of regions where hydrogeologic variables are statistically stationary, small- scaTe parameter values may be representative of larger-scare effective values (Neuzil, 1994~. For example, researchers have observed similar values for effective flow parameters on multiple scales at the Mirror Lake site in New Hampshire (Hsieh, 1998~. However, as scale changes, new geologic features (fractures, stratigraphic changes) may become important, resulting in regional effective properties that differ from those observed in smaller-scare studies. At some sites, very large changes in permeability have been seen with observation scale (e.g., Bredehoeft et al., 1983~. More research is needed to determine if there are situations in which upscaTing (i.e., using data collected on smaller scales to derive informa- tion at larger scales) is possible and to develop upscaling methods. Many studies have collected hydrogeologic data on multiple scales; however, researchers may not have taken the further step of developing relationships between the scales. General methods for upscaling have not been established. Indeed, researchers will probably need different methods of upscaling depending on the region's characteristics (homoge-
Scientific Issues neons, stationary, trend/pattern, etc.~. 97 Upscaling parameter estimates may not be possible at sites with markedly nonstationary parameter fields unless an observable trend exists (e.g., a linear decrease in perme- ability with depth). When a pattern of variability is observed at a number of small-scare studies, researchers sometimes assume that pattern for the larger study. If a number of subregional studies have been conducted in a region, the small-scare studies have clear value as indicators of subregional vari- ability. This information may be particularly important for regional transport studies. where larae-scale dispersion is dependent upon small- scaTe permeability variation. "7 . . . . . . . We recommend that the USGS incorporate into its regional model- ing efforts relevant and reliable data collected during previous studies within the regions. Using data from previous studies is particularly im- portant in the groundwater field because of the spatial and temporal variability inherent in subsurface data sets. Because subsurface proper- ties and processes vary in space and time, it may be useful to character- ize modeled variables as random or stochastic. Given the impossibility of collecting data everywhere at all times, the properties and processes of interest are always uncertain. In this context, every additional piece of information is valuable in reducing uncertainty in modeling efforts. High data collection costs for the subsurface further increase the value of data available from past studies. Developing Decision-Making and Risk Models for Groundwater Use As noted earlier, the WRD should be involved not only in collecting data on water supply, but also in facilitating the use of this information by decision-makers, who have to contend with competing uses (domes- tic, agricultural, commercial, industrial, recreational, and ecological). Water-use allocation takes into consideration not only scientific knowI- edge about water resources, but also public policy options, and can be accomplished with the help of models that integrate the two areas ~C, 1991a). The WRD may, by working in partnership with other national or regional agencies, have a role in analyzing how various policies or laws affect water use regionally. Models could, for example, explore system behavior in response to changes in management or policy, where
98 Investigating Ground water Systems variables might include cost of pumping water, cost of crop production, income from crops, tax revenues, etc. CONCLUSIONS Numerous important advances in hydrogeology have occurred in the past two decades, but serious challenges remain. As part of the Ground- Water Resources Program (GWRP) and associated programs, the USGS WRD should investigate groundwater occurrence and movement in complex hydrogeologic environments such as fractured rock and karst and in heterogeneous media. Advances in theory should be supported by the creative application of field methods and should lead to more reaTis- tic models backed by sensitivity and uncertainty analysis. Surficial aquifers and their boundaries should also receive consider- able attention, even in regional studies. Aside from being vulnerable to contamination, shallow aquifers are the focus of research on the spatial and temporal distribution of recharge and discharge and on interactions of groundwater and ecosystems. Many scientific disciplines, including ecology, limnology, chemistry, hydrology, and meteorology, have some- thing to contribute to such groundwater investigations. Regional groundwater studies thus provide ideal opportunities for collaboration within WRD programs and with other USGS divisions and external or- ganizations. Collaboration should also facilitate the development of water-management models, which incorporate legal, economic, ecologi- cal, and other constraints. Finally, changing technology is creating opportunities for innovative approaches to the dissemination of the groundwater information and re- sults generated by such projects. Chapter 5 is devoted to these data is- sues.
