Click for next page ( 43


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 42
4 Spatial and Temporal Scales of Recharge and Discharge Our groundwater resources are highly vulnerable, as each new drought reminds us. Expected in- creases in groundwater extraction, increases in societal value of ecosystem function, and changes induced by global change (Chapter 3) indicate that forecasting potential variations in groundwater recharge and dis- charge will become ever more important. However, in the context of forecasting recharge and discharge, what length of time is of interest? What area of land? Fluctuations in the Great Salt Lake (Box 3.1) and the Ogallala aquifer occur over time scales of decades, centuries, and even millennia. In contrast, major changes in flows and levels of hillside springs and shallow household wells may occur in days or weeks. Similarly, in terms of spatial scale, town- ships are concerned with recharge over square kilometers, and river basin planners over significant fractions of continents. We may suppose that the critical scales span weeks to centuries and from tenths of km to tens of thousands of kin. How can measurements and estimates for one spatial scale and time period be scaled to different geographical areas and different time spans? Issues related to organization of measurements, ob- servation networks, and modeling studies to address a host of scaling issues are the focus of this chapter. NEXUS OF TECHNOLOGY AND NEED Evaluations of groundwater fluxes often are based on observations at the scale needed for wetland delineation, seep and spring identification, recharge area identification for groundwater protection, or the identification of gaining and losing streams. Estimations of fluxes at larger scales are typically in the form of water budgets on a basin scale for river management. Now, the need to understand the effects of climate change has further expanded the range of scales of interest to continental (see Box 4-1) and global scales. However, published water-budget estimates at these scales (e.g., Oki, 1995; Dettinger and Diaz, 2000) rarely treat groundwater runoff separately from surface water runoff, nor do they separate discharge to the atmos- phere (e.g., by phreatophytes) from surface-water discharge. Further, the need to assess anticipated changes in climate, land use/land cover, availability of groundwater resources, and trends in water quality begs for a predictive capability on many scales in both space and time. Recent advances in technology provide both challenges and potential solutions to address these multi-scale needs. New data collection techniques have generated a plethora of data (much of it from re- mote sensing). Concurrent gains in computational power permit us to access and mine these very large data sets. Likewise, advances in simulation capability (e.g., resolution, processes modeled, and processor speed) 42

OCR for page 42
Spatial and Temporal Scales of Recharge and Discharge 43 BOX 4-1 The Role of Climate Models and Satellite Data (GRACE) in Estimating/Mapping Changes in Groundwater Storage Observing and modeling components of the terrestrial water budget at regional to continental scales is now feasible. Remote sensing of precipitation by satellite (Huffman and Bolvin, 2002) and radia- tion from geostationary satellite (e.g. Pinker and Laszlo, 1992) allows for remote sensing driven hydrologic modeling. As these models improve through large-scale validation, there is the potential for better under- standing of the spatial and temporal variation of terrestrial water storage (soil moisture and groundwater). For the results from a 50-year run of the Variable Infiltration Capacity (VIC) macroscale model (Liang et al, 1994; Cherkauer et al., 2003) over the Mississippi River basin (Figure 4-1), forced with "ridded precipi- tation, air temperature, and other derived forcings from NOAA Cooperative Observer stations (Maurer et al., 2002) shows the maximum range of model-predicted soil moisture over the 50-year period for each grid cell. Averaged over the Mississippi River basin, and weighted by mean annual precipitation, the range is 29.6 cm. The upper Mississippi River basin, the Red-Arkansas basin and the eastern side of the Rocky Mountain divide are areas of large interannual variation in water storage, and therefore expected large natu- ral variability in groundwater levels and recharge. As a comparison, early land surface modeling work (Manabe, 1969) used a global average soil moisture capacity of 15 cm, a number that has been widely used in the climate community. The implica- tion of the early climate models is a suppressed coupling between the terrestrial hydrosphere and the at- mosphere, resulting in decreased predictions of evaporation. .... (2~'.'-~ l ~~ it I ~ I. . Irk JA .~ _ In, hare'_\' Lit ~ en." .~. 0 1Q D 30 40 5O ~11 Ulaisture Rarige, cn1 FIGURE 4-1. Range in simulated soil moisture content over the Mississippi River basin from the VIC model simulations, 1950-2000. SOURCE: Reprinted, with permission, from Maurer et al. (2002~. ~ 2002 by American Meteorological Society.

OCR for page 42
44 Groundwater Fluxes Across Interfaces An alternate approach to estimating evapotranspiration over large areas is to compute an atmospheric moisture balance: V 'tqv+ ~ :l q= P - E where the first term on the left-hand side is the divergence of the atmospheric moisture field (usually com- puted from a column-average effective moisture and corresponding wind field), and the second term is the change in atmospheric moisture storage over the domain. One limitation of this approach is that accurate es- timates of the moisture flux convergence require spatial domains that are quite large, typically 106 km2 or greater. Ropelewski and Yarosh (1998) describe a moisture balance study of the central U.S. (the approxi- mate domain of the Mississippi River basin). They used the atmospheric moisture balance equation with ob- served precipitation and convergence computed from radiosonde observations to solve for evapotranspiration. Figure 4-2 shows the accumulated departure from the mean for precipitation, computed evapotranspiration, and an inference of storage change that would have been required to satisfy the accumulated balance. The difference between the largest and smallest value of the normalized storage is an estimate of the minimum subsurface moisture storage capacity. This value is about 45 cm, or around 1.5 times the value estimated by the VIC model. Also, the period of analysis for the atmospheric budget (1973-1992) is less than half as long as that used in the surface modeling approach; the discrepancy would certainly be larger if the periods had been compatible. Comparison ofthese two estimates raises several interesting questions: 1. Does the surface modeling approach, which ignores groundwater interactions, tend to bias the estimated moisture storage excursions downward? Restated, do groundwater-surface water interactions exert a significant influence on variations in subsurface storage, and hence evapotranspiration, over areas as large as the Mississippi River basin. If they do, how can they be estimated? 2. A second possibility that cannot be excluded, is that a significant part of the apparent subsurface storage "requirement" based on the atmospheric balance is attributable to "noise" in the atmospheric conver- gence estimate, which is effectively integrated in estimate of required storage (Gene Rasmussen, personal communication, 2002~. 3. The new observing strategies, like the recently launched Gravity Recovery and Climate Experi- ment (GRACE) mission, and planned GRACE follow-on may provide some insights into issues addressed in 1 and 2, above. Given potential capabilities (at least with GRACE follow-on) to estimate total moisture varia- tions (atmospheric plus surface and subsurface) to within a few cm over spatial scales as small as 100 km, can the contributions of the various moisture storage terms be deconvolved to sufficient accuracies so as to allow better understanding of continental scale subsurface moisture dynamics, and their relation to continental scale evapotranspiration? Figure 4-3 (from Rodell and Famiglietti, 2002) shows the mean annual cycles of monthly changes in terrestrial water storage and its components over Illinois. Their analysis indicates that these sea- sonal changes would be detectable if they occurred over regions greater that 500,000 sq. km. The potential for understanding the seasonal and inter-annual variation in terrestrial water storage at continental scales has greatly increased with current modeling and remote sensing capabilities. The challenge to the hydrology community is to integrate into groundwater flux studies the results from macroscale hydro- logic modeling, modeling and observations of moisture fluxes in the atmosphere, and terrestrial water storage observations to provide further insights into the spatial and temporal variability of recharge.

OCR for page 42
Spatial and Temporal Scales of Recharge and Discharge 30 20 - o lab :5 -1 0 ~ E 1 ~ - -20 _ 1, O- r _ . -30- [~ p) ~ \ ,, ' (.. I__ a_ ~ f ~--_\ - c. 5 - S / ~ \ OCR for page 42
46 Groundwater Fluxes Across Interfaces make it possible to test our understanding in new ways. Finally, the last ten years have brought forth a host of low-cost sensors that can collect data with unprecedented temporal resolution. These data provide a rich source of "ground truth" for the more spatially complete remotely sensed data. To take advantage of this nexus of need and technology we must use measurements made at many disparate spatial and time scales to address questions posed at another scale. Further, we must understand how our process descriptions vary with scale and that coupling of processes occulting at different scales. These scaling relationships are a function of geologic and climatic regimes, and precipitation intensity. Identifying these relationships wall allow quantification of recharge and discharge at the local scale, where they affect urban planning and eco- logical function; prediction of and preparation for the effects of climate change; and most fundamentally, better understanding of recharge and discharge mechanisms. One caveat is in order. Much of the new technology to address spatial scaling issues relates to near- surface measurements. Such measurements may provide ways to estimate net infiltration, by measuring surface evaporation and plant consumption, not by measuring net recharge directly. However, more accu- rate recharge estimates may be made not at or near the soil surface but deeper underground where flow is ... . . .. ., .. ~ . . . ~~ . . 1 ~ . ~ . . .. . . . . ~ .. . ~ . .. .. . . . . less coupled to other mass and energy transfer processes than IS 11OW In the so1l-plant-atmosphere system. At depth, the effects of evapotranspiration, and spatial and temporal fluctuations in infiltration are attenu- ated, rendering deeper subsurface flow more uniform and therefore easier to quantify (Figure 4-4~. (Subsur- face information that relates directly to fluid flow is typically limited to measurements taken in boreholes, which may be relatively far apart. Though such boreholes cannot provide information directly about lateral scale processes smaller than the distance between them, such detail is less important in the subsurface than it is in the near surface due to the slower and more uniform character of flow at depth.) DEFINITION OF SCALING Because the term "scaring" is not always clearly defined, there often is some confusion surrounding discussions of problems and issues. The term "scaling", to many, is veiledl in a nimbus of exciting mystery. At a basic level, part of the mystery simply comes from confusion of two connotations of the word - meaning either scale invariance (i.e., processes behaving similarly at small and large scales) and upscaling/downscaling (i.e., aggregating/-disaggregating data (Bloschl, 2001~. ~ this report, we use the term "scaring" in the sense of aggregation and disaggregation of estimates and data. Upscaling refers to taking measurements made at a series of points or small scales and determin- ing how to use them to estimate quantities or rates over a larger scale. In downscaling, measurements at a large scale are disaggregated to a finer scale using statistical methods. A brief discussion of typical arid settings helps frame the challenge of the scale in recharge. In arid regions, recharge tends to occur at high elevations or along mountain fronts or focused along streams or swaths of irrigated land, whereas in humid regions, recharge is more diffuse over wider geographical areas; discharge tends to occur at springs, wetlands, and playas, and along the shorelines of oceans and estuaries, streams, and lakes. Except for focused spring discharge, these are neither points nor broad geographic areas. As an example, consider precipitation falling over basin floors in the playa landscape of the semiarid Southwest (Figure 2-5~. Over vast areas of basin floor there is no net recharge. The geothermal gradient

OCR for page 42
Spatial and Temporal Scales of Recharge and Discharge Surface entry Net infiltration Water table I response q..~.. Time or space 47 FIGURE 44 Schematic illustration of the smoothing of the recharge signal as water enters the subsurface as infiltra- don and some component of this arnves at the water table after percolating Trough the vadose zone. This smoothing simplifies measurement of the recharge flux at the water table, reducing the need for high resolution spatial and tempo- ral mon~tonug. drives water vapor upward, taking all the liquid water that reaches depth and returning it to the surface. Dur- ing rainfall events that activate surface flow, small stream channels begin to collect and carry away surface runoff, and focus infiltration. Some streams are wide enough and flow long enough to allow recharge to the water table. So at the scale of a playa landscape (say 500 m), recharge occurs not at all or perhaps once a century, while at intermediate locations, relatively high rates of recharge occur episodically. Measurements taken in between stream channels would detect no recharge, while those in large stream channels would show that there is relatively high recharge. It follows that the average recharge wall depend on where the measurement is centered, over what period of time the measurements are made, and over what scale the average is computed (see Box 4-1, for example). In the example of the playa landscape, 99.9 percent of the basin floor might have no recharge whereas 0.] percent of the area consisting of ephemeral stream channels that promote infiltration during 0. Ipercent of the time generate essentially all of the Tong-term regional recharge. As the averaging area in- creased from a point on the basin floor, segments of stream channel would be included along with a larger Faction of basin floor. When averaged together a 'scale effect' would be evident, at least with respect to when a meaningful average value could be defined. The same applies to the temporal scale of averaging. A graph of flux rate vs. spatial or temporal averaging scale might be drawn similar to the classic diagram of the representative elemental volume (REV) used to show the scaling behavior of hydrogeologic parameters such as porosity (Figure 4-SA) or the scaling of correlation length in heterogeneous porous media (Figure 4- SB). By examining the relative magnitude and variability of groundwater fluxes at increasing spatial and

OCR for page 42
48 hA'rro~.onin ~ Mo:~rQSCoo1C In As o CL o Groundwater Fluxes Across Interfaces , Heterogeneous c~ - Homogeneous O 9, 92 93 ~ A Volume y mm m B Vs ~ inter basin aquifer In K alluvial basin Fluvial . aqul er core - ,' ~ Earth 1 1 1 1 1 1\ ,, ? km Mm Separation FIGURE 4-5 A) Scaling of porosity of a porous medium as it might be measured on samples of increasing volume (V- 1, V2, V3...) taken at a random point within the system. A plateau starts at V3, known as the representative elemen- tary volume. SOURCE: Reprinted, with permission, from Freeze and Cherry (1979). (a) 1979 by Prentice Hall. Adapted from Hubbert (1956) and Bear (1972). B) The semivariance fly) of the natural log of hydraulic conductivity as a function of the correlation length in heterogeneous porous material. In this case there is a series of plateaus as one moves, for example, from a single sand bed to an interbedded sand-clay package to stacked packages to an entire basin. SOURCE: Reprinted, with permission, from Gelhar (1986). ~ 1986 by American Geophysical Union.

OCR for page 42
Spatial and Temporal Scales of Recharge and Discharge 49 temporal scales, we may discover thresholds of continuity or uniformity that correlate with practical scales of measurement and/or application. For the example of the playa landscape, plateaus in scaling might occur at spatial scales related to the scale of the drainage network. Of course, for the scaling proc- ess to make sense knowledge of the underlying physical processes is needed; the averaging would not be simple statistical interpolation as is the case for many upscaling problems. EXPECTED SCALING BEHAVIOR OF RECHARGE/DISCHARGE FLUXES Understanding and measuring groundwater fluxes across interfaces along a continuum of temporal and spatial scales is important for determining water and solute budgets. For the purposes of this discussion we define increasing spatial scales as moving from smaller to larger areas of the landscape or volumes of material and increasing temporal scales as moving from short to longer time frames. Scaling issues related to recharge and discharge have been the focus of studies in a variety of re- gions (e.g., Dyck et al. 2003; Desbarats et al. 2001; Delin et al. 2000; Lin and Anderson, 2003; Flint et al., 2002, see Box 4-2; Stoertz and Bradbury, 1989; Jorgensen et al., 1989; also see the review article by de Vries and Simmers, 2002~. Many of these studies are, by their nature, specific to a place and do not neces- sarily address general processes. To our knowledge, no comprehensive studies of flux scaling exist. We can, however, make predictions about the expected scaling behavior of these processes. As we move from smaller to larger scales the variability of flux values will change (Figure 4-5A). These changes with scale are fundamentally related to the spatial and temporal heterogeneity of natural systems, and include both the physical properties of the system and the hydrodynamic parameters imposed by the larger envi- ronment. For example, for a small groundwater basin, recharge and discharge fluxes are controlled by the geology and topography (physical properties of hydraulic conductivity, porosity, surface drainage, etc.) and by the imposed stresses (climate, precipitation intensity, land cover, pumping, etc.~. in general, at small spatial and temporal scales, we expect flux variability to be large, and to be highly correlated to small-scale variability in physical parameters. At larger scales we expect this variability to decrease as spatial and temporal changes are averaged. The ease or difficulty of upscaling of point estimates depends on how variable the point esti- mates are. Variability itself is, of course, variable. That is, in some instances, recharge flux variations, even on a small scale, may be only mildly variable (e.g., Dyck et al., 2003) whereas in other instances it may vary over orders of magnitude from point to point (e.g., Cook and Kilty, ~ 992~. STRATEGY TO ADDRESS THE ISSUE OF SCALE Below we sketch a framework or strategy to address the issue of scale as related to estimation of groundwater recharge and discharge. It is our belief that remote sensing will play an important role in the strategy. Although a case has been made for using various remote sensing tools in recharge and discharge studies (e.g., Cook and Kilty 1992; Salama et al. 1994; Meijerink 1996; Jackson 2002), and especially in scaling up from point measurements, there is much work to be done to develop the meth- ods to a point where they are truly useful. Of critical importance is the interplay between remote sens- ing, land-based measurement and numerical modeling. Conceptual Model. The conceptual model is a basic tool for visualizing the hydrogeologic system where direct measurements of fluxes and other hydrologic processes may be lacking but simi- larity of form and process are evident (Figure 1-3~. It provides a way of looking across multiple scales;

OCR for page 42
50 Grour~dwater Fluxes Across Interfaces BOX 4-2 Recharge Mapping at Yucca Mountain, Nevada -- The Scale Effect Flint et al. (2002) evaluated various methods to estimate net infiltration and recharge at Yucca Mountain. A summary of the methods, i.e. general approach, scale of application, and strengths and limitations, is was presented earlier in this report (Table 1-2~. These methods produce estimates of flux that reflect different spatial and temporal scales; have different data requirements, strengths and limitations; and have varying sensitivity to water flux in fractures. Recharge varies spatially owing to variations in precipitation, surface microclimates, thickness of alluvial deposits, faults and fractures, and thickness and hydrologic properties of geologic strata in the unsaturated zone. Two methods applied to measured data at different temporal and spatial scales may yield different fluxes, yet both could be correct. at the scale of measurement. For example, a water-balance model reflecting a conceptual model of shallow infiltration at Yucca Mountain was used to estimate the temporal and spatial variability of net infiltration (Figure 4-6a) for the upper boundary conditions of the unsaturated-zone flow model (Bodvarsson and Bandurraga, 1996) (Figure 4-6b). Although the average flux for both models are the same, this figure illustrates the differences between surface flux and flux at the water table due to the redistribution of percolating water that occurs in the unsaturated zone over varying spatial and temporal scales. Measurements of water content in shallow neutron-access boreholes used to calculate near- surface fluxes over the last 15 years may correctly represent average conditions over that period, yet be in apparent disagreement with fluxes estimated by applying the chloride mass-balance method to pore waters extracted from drill cores because the latter method may be representing fluxes that were in effect hundreds or even thousands of years ago. Recharge-estimation methods based on deeper measurements integrate, or average out, the excursions caused by local, near-surface processes. The values in Figure 4-7 are results from the various methods used to estimate recharge at Yucca Mountain. The methods generally are arranged in order of the integrated depth or temporal scale that the method addresses. Of) ~ D LL 8 LL o ':z o -I ~ - - I 0' 'I 2 go is. is. C o 54S,000 550,000 . . UTM EASTING, IN METERS 0-1 mmlyr Lo. o o Lo `. ._ - -: 547,000 552 pOO UTM EASTING, IN METERS FIGURE 4-6 (a) Spatial distribution of shallow infiltration at Yucca Mountain using a water-balance model, compared with (b) percolation flux (recharge) at the saturated zone. SOllRCE: Modified from Flint et al. (2002~.

OCR for page 42
Spatial and Temporal Scales of Recharge and Discharge at o IL A o 6 en o o in: IL Q U] LU C) CO UJ CO nit o IL I UJ Neutron moisture logs for 80 boreholes Watershed model using neutron moisture data Tntium peak in channel alluvium Darcy flux from borehole saturation profiles Borehole temperature profiles Carbon-14 in gas phase Chlonde in porewaters Chlorine-36 bomb pulse and natural variations Chloride in perched waters Chloride in local ground- water Modified Maxey-Eakin model PTn lateral ~ flow. PTn Shallow , . . Present-day (Holocene) \~ ~ Late Pleistocene <0.1 0.1 1 10 100 1000 Percolation flux, mm/yr FIGURE 4-7 Comparison of percolation fluxes estimated by various methods. A bar represents the range of estimates using a given method in different topographic settings, whereas each point represents a single estimate. SOURCE: Modified from Flint et al. (2002~. Shallow point measurements address surface processes generally acting on a yearly to decadal scale, or reflect processes in a single topographic feature. Measurements made deeper in the mountain or integrating deep boreholes reflect an integration of time and space owing to unsaturated flow processes, stratigraphic influences, and differences in fracture/matrix interaction within the various hydrostratigraphic units. Some methods, such as those using perched-water chemistry, incorporate various percentages of water that may be thousands of years old. In summary, because recharge rates are variable in space and time, considerable differences in estimated flux values using different methods do not necessarily indicate that one of them is wrong. What is important is to match the appropriate method to the spatial and temporal scale of the problem. Regional or long-term average rates may require one method; local or recent rates, another. 51

OCR for page 42
52 Groundwater Fluxes Across Interfaces a good conceptual mode! is essential to the design of new observing networks. The conceptual model also guides the interpretation of historical records (e.g., of streamflow and groundwater levels) and is a first step in evaluating the effect of climate forcing on recharge/discharge processes. Prototype and Lab-Scale Experiments. Benchtop experiments and new instrumentation packages will be an important step in creating new observing systems in field applications. Many of the fundamental processes that give rise to variability in space and time are manifest at small scales: e.g., crack-flow, unstable wetting, flow through random media packings. Given the intrinsically greater control and measurability of laboratory simulations, these experiments will be an integral component of gains in understanding of the underlying processes that give rise to scare-dependence in hydrologic processes. Field Experiments. The design of field experiments to capture the multiple spatial and tem- poral scales of recharge/discharge is fundamental to the scaling issue (Chapter 2~. With the advent of low-cost sensor technology, meso-scaTe geophysical methods (mostly advances in interpretation), high- resolution elevation and remotely sensed land cover, and the computational capacity to digest this information, we now have the potential to enhance predictions from regional mean behavior to catch- ment, hilisiope and plot scales. At the same time we must anticipate global climatic changes that will likely alter the patterns of these processes so critical to life. The challenge is to measure the nature, pattern and magnitude of fluxes with coordinated observations of the atmosphere, soil moisture regime, and groundwater-stream system. The Information SYstem. Although we have vast data sets. and are noised to obtain even richer observational results, unless we can synthesize across these diverse sources of information we are fundamentally limited in our ability to compare observations at a range of scales. Efficient infor- mation management and the cataloging of compatible formats of hydroclimatic data that are collected at disparate spatial and temporal scales will be crucial for seeing the scale behavior in the data, and thereby identifying new constitutive relations. and validation of theories or process at multiple scales. Modeling. Finally, numerical modeling is fundamental to identification, estimation, simu- lation and prediction of recharge and discharge, and provides a fundamental tool to address coupling and scale interaction of such processes. Through model calibration the magnitude of average fluxes occurring at the watershed scale (>10,000 km2) can be estimated reasonably well provided discharges are measured accurately. At local scales (i.e., the small to intermediate scale of 0.1 to 10 km2), how- ever, flux values as predicted by a model and measured in the field at pairs of proximal points (e.g., in a spring and ~ meter away from the spring) may be completely different and have large associated error. Since fluxes at local scales are those most relevant to ecosystems and to human habitation, there is great advantage to advancing our understanding to enable accurate predictions of fluxes at ever smaller dimensions. In addition to prediction of fluxes at various scales, models allow for testing hypotheses of scale interaction, for "what-if?" scenario investigations for water management and allocation. A third application of modeling to scaling issues is the problem of data assimilation (NRC, 2002a) and how the observing system will continuously update the model and model forecasts. The issues of scale relative to recharge and discharge fluxes involve understanding the interde- pendence between variability in fluxes and the physical and chemical characteristics of a site. While both recharge and discharge occur as diffuse and focused fluxes, it is implicit that recharge and dis- charge processes must be studied independently. Furthermore, each process is expected to scale in dis- tinct ways. Up to the scale of perhaps 100 km2, scaling studies can be done at the benchmark sites pro- posed in Chapter 2. Each experimental setting should be studied at nested scales with nested monitor- ing sites within the basin and measurement/sampling/monitoring stations on a range of scales across the

OCR for page 42
Spatial and Temporal Scales of Recharge and Discharge 53 basin. Methods of assessing uncertainty will be required for each of the issues as well as the develop- ment and verification of related scaling theories. The development, application, and evaluation of mul- tiple assessment methods specific to each issue and the development, application and evaluation of conceptual and numerical models, and their comparison and independent validation are also required. RELATION OF INFILTRATION TO GROUNDWATER RECHARGE Most of the newly available technologies that may allow us to monitor variables at a variety of scales pertain to the near-surface environment (Smith, R. et al., 20021. Except at the scale of 100's of km via identifying changes in the water table through satellite gravimetry (Box 4-1), it is not possible to monitor groundwater recharge and discharge remotely. Since infiltration and recharge may vary in magnitude, timing, and distribution (e.g., Box 4-2), it is imperative for the scaling problem that we de- velop procedures that relate recharge and discharge to landscape features that we can observe directly with high spatial resolution. These relationships must be developed through tests of models with ground-based measurements that will yield estimates of infiltration (with associated uncertainties) cou- pled with the variety of methods that we have at our disposal to estimate recharge. The development of these relationships will itself involve attacking a scaling problem because the temporal and spatial vari- ability regimes of infiltration and recharge are different (Figure 4-~. Ground-based measurements of infiltration will require elucidation of soil-plant-atmosphere controls on near-surface soil fluxes. Estimation of recharge from infiltration can also be done by estimating evapotranspiration; re- charge then is calculated as the difference between infiltration and evapotranspiration. Unfortunately, direct estimates of evapotranspiration are only possible at relatively small spatial scales - e.g., via flux towers using eddy correlation or Bowen ratio methods (Baldocchi et al., 2001), which typically have footprints of at most a few square km. Remote sensing using LIDAR can result in high-resolution im- ages of water vapor that can be especially important in areas of high flux variability such as riparian zones (e.g., Cooper et al., 2000~. Indirect methods, such as catchment water balance methods, are ap- plicable to Tong (multi-year) time periods where subsurface storage changes can be averaged out, or alternatively require a modeling strategy to estimate subsurface storage change. Comparative studies of some of these methods have shown promise (e.g., Wilson et al., 2001~. Hydrologic and/or macroscaTe land surface models typically produce estimates of evapotranspiration (and its space-time distribution) following calibration of model parameters to produce a match with observed streamflow. Given that they are forced with observed precipitation, they arguably can produce usable estimates of evapotran- spiration provided that observed streamflow is matched reasonably well. Nonetheless, confidence is greater in their long-term average predictions than the time sequencing, due to difficulties in verifying that the dynamics of subsurface storage are properly represented. With respect to the representation of subsurface storage, essentially all surface hydrology mod- els (e.g., those intended for flood and drought forecasting) and land surface models (e.g., those intended to represent the role of the land surface in climate prediction models) represent the subsurface as one or more soil "slabs", with finest vertical resolution of depth typically a meter or two. Land surface models can be run in so-called "off-line" mode, that is, forced with observed precipitation and other surface atmospheric variables (e.g., downward radiation, wind, surface air temperature, vapor pressure deficit). Implemented in this way, they behave essentially like continuous watershed simulation models, al- though at much larger scales. As indicated above, if streamflow is simulated reasonably well, these models can offer insights into the dynamics of evapotranspiration, as well as changes in subsurface moisture content. Many important sources of error and uncertainty affect the available estimates of infiltration rates. These include Tow space-time resolution of storm events; neglect of surface runoff; relatively or under-

OCR for page 42
54 Groundwater Fluxes Across Ir~te~aces standing of processes that control water uptake by plants from soils and fractures; poor definition of processes and properties that contribute toward the generation of focused infiltration and corresponding fast flow paths; and the transient nature of infiltration processes. VARIATION OF FOCUSED AND DIFFUSE DISCHARGE WITH MEASUREMENT SCALE Understanding and measuring groundwater discharge at various temporal and spatial scales is es- sential for completing any groundwater budget, and is a critical part of most groundwater flow models, yet most groundwater discharge measurements are highly uncertain. Moreover, the uncertainty is poorly char- acterized and probably most often underestimated. The main discharge points for groundwater (in addition to pumping welis) are springs and seeps, streams, lakes, wetlands, and oceans. Flow measurements in each of these environments present unique conceptual and measurement challenges. In addition, the common practice of estimating recharge using numerical flow models is subject to an implied scaling due to the in- abiTity of models to reproduce all surface-water features. Springs and seeps- As point locations of groundwater discharge, springs are often attractive Toca- tions for flow measurements, yet the proportion of groundwater discharge that a particular point spring represents is usually unmown, and may be small. Seeps associated with wetlands can be complex, dy- nam~c, and difficult to measure, but may exert a profound influence on the water balance of a basin. Can we develop ways to predict the relative proportion of diffuse to focused flow at different spatial and temporal scales? Streams- Hydrogeologists commonly use low-flow, or baseflow, conditions in streams and rivers as a measure of basin-we groundwater discharge, yet there is great uncertainty in such measurements (e.g., Halford and Meyer, 2000~. We commonly describe the time series of low-flow measurements in a stream using flow-duration curves, but it is unclear what statistical measure derived from such curves (e.g., Q60, Q8o, Q7, 10) appropriately represents groundwater discharge to the stream over various time scales. Our best low-flow measurements and statistics come from long-term USGS gaging stations with well-defined rating curves, yet such stations are usually only installed on major streams, and nationally the number of gaged sites is decreasing. Furthermore, even if we are convinced that a streamflow measurement under baseflow conditions represents groundwater discharge, how do we disaggregate the complex patterns of inflow, out- flow, and interflow that occur higher in the watershed? Groundwaterflow models- Commonly-used finite-difference groundwater flow models (e.g., MOD- FI-OW) include numerical methods for linldng groundwater to surface water through head-dependent boundary conditions (such as the MODFLOW river, lake, drain, and streamflow routing packages). Due to model and or mesh limitations such models must always ignore surface-water features smaller than a given size, and the usual rationalization for ignoring these features is that they are unimportant to the larger-scare problem. However, ignoring such small-scale features is likely to cause model-based recharge estimates to be smaller than field estimates. Continued effort into sub-gr~dscaTe parameterization schemes is essential and has proven to be a powerful avenue to address this issue intrinsic to spatially explicit modeling.

OCR for page 42
Spatial and Temporal Scales of Recharge and Discharge USE OF REMOTELY SENSED/MAPPED PARAMETERS AS SURROGATES FOR RECTIARGE/DISCHARGE 55 Remote sensing and high-resolution digital elevation models wall be fundamental to the extrapola- tion over landscape scales from points of estimation. The essential next step is for the highly variable fea- tures that control discharge and recharge (e.g., permeability distnbution), to be associated with the remotely measured surrogate values. One example of this is the HOST (Hydrology of Soil Types) classification (Lilly et al., 1998), originally developed to predict river flows for ungaged catchrnents and later used within other models for predicting contaminant concentrations in runoff. HOST sem~-quantitatively relates soil classification (e.g., loamy sand) to soil hydrologic characteristics, and thence to stream response to precipi- tation. Similarly, vegetation patterns (e.g., Rosenberry et al., 2000) have shown promise as surrogates for discharge.