My Account: Log In | Join | Renew
Table of Contents
Select All Chapters
Although many advanced information technologies are potentially of value to environmental impact assessment, this chapter focuses on those generally included under the term geographic information technologies, including geographic information systems (GIS), remote sensing, and the global positioning system (GPS). The chapter begins with a brief description of each of these, and a more detailed review of the state of development of GIS, its current capabilities, and impediments to greater use. Problems having to do with the integration of the geographic information technologies, and the raster and vector approaches to GIS, are discussed in the context of non-point-source pollution in the vadose zone, and attention is drawn to current research on advanced geographic data models, which may hold the key to further progress in integration.
Geographic information systems (GIS)-based modeling of solute transport at the regional scale involves linking three different kinds of activities, namely model building (to describe the physical and chemical processes), data collection (to provide values for model parameters and attributes), and GIS (to provide an organized data structure for the automated handling of the input data to the computations and for the display of results). The development of mathematical model building for specifying physical and chemical processes—from one- to four-dimensional—and increasing in complexity in space and time, is juxtaposed with (i) the collection of data at the regional scale and (ii) the design of current GIS. The basic assumptions and thought structures in present-day GIS, including objects and fields, overlay structures and relational databases and their suitability for hydrological modeling are discussed. The basic principles, advantages and problems of (i) linking data and models, and (ii) linking models and GIS (loose coupling, tight coupling, and embedded coupling where the model is written using GIS functionality) are presented and discussed. Special attention is paid to questions of uncertainty, spatial and temporal variation and upscaling, model calibration, validation and error propagation. The chapter presents practical suggestions for creating successful modeling tools within a specialized GIS environment.
Geographic information systems (GIS) provide software for capturing data and simple operations thereafter. These operations are limited to data handling (concatenation, classification, retrieval) and visualization involving little statistical interpretation and essentially no modeling of space or time dependence. Yet detection and modeling of space and/or time correlation is critical since connectivity in space of, e.g., extreme transmissivity and saturation values condition solute transport modeling and related remediation designs. Geostatistics offers tools specifically designed to handle space and time correlation without having to call for arbitrary, and somewhat counterproductive, hypotheses or models of data independence. In the last decade geostatistics has been extensively used for modeling petroleum reservoirs and aquifers in physical and data environments similar to those prevailing in soils and vadose zones. The corresponding experience and software could be transferred into an extended geographic information system adding interpretative capabilities to such a system. This chapter suggests such relevant additions useful for modeling spatial distributions in a data-sparse environment.
Stochastic modeling of solute transport in the vadose zone has been formulated both as a stochastic-convective (parallel soil column) and a stochastic-continuum process, the latter referring to transport that evolves macroscopically from a stochastic-convective to a convective-dispersive process. Of the two, a stochastic-convective formulation seems the most compatible with GIS, because it uses a relatively simple local process driven by parameters that might be associated with soil morphological features, and could be integrated up to a large scale by simple arithmetic averaging over the local sites. Such an approach has been undertaken in the past coupling a piston-flow local transport model with soil survey information. The challenge for this or any other approach will be to develop a reasonable local-scale model whose parameters can be related to identifiable local-scale features. Transfer functions may offer an unexplored opportunity for such a modeling effort in the future, by representing local features through a travel time distribution whose parameters are related to local conditions, and regional features through stochastic representation of rainfall or irrigation patterns. An example of the latter approach will be given in the presentation.
In recent years, worldwide attention has shifted from point source to non-point source (NPS) pollutants, particularly with regard to the pollution of surface and subsurface sources of drinking water. This is due to the widespread occurrence and potential chronic health effects of NPS pollutants. The ubiquitous nature of NPS pollutants poses a complex technical problem. The areal extent of their contamination increases the complexity and sheer volume of data required for assessment far beyond that of typical point source pollutants. The spatial nature of the NPS pollution problem necessitates the use of a geographic information system (GIS) to manipulate, retrieve, and display the large volumes of spatial data. This chapter provides an overview of the components (i.e., spatial variability, scale dependency, parameter-data estimation and measurement, uncertainty analysis, and others) required to successfully model NPS pollutants with GIS and a review of recent applications of GIS to the modeling of non-point source pollutants in the vadose zone with deterministic solute transport models. The compatibility, strengths, and weaknesses of coupling a GIS to deterministic one-dimensional transport models are discussed.
The objectives of a solute transport study determine the form of the subsequent modeling and experimental efforts. In addition, to achieve the study objectives, it is essential that the modeling approach and experimental design are consistent with each other. For example, the efficacy of model discrimination and parameter estimation strategies is strongly contingent on experimental design. Likewise, development of an effective experimental design is generally best achieved on the basis of hypothesis testing regarding specific candidate transport descriptions. The connection between experimental design and parameter estimation is illustrated using sensitivities, which are variations in predicted solute concentrations resulting from variations in model parameters. These sensitivities can be used to develop robust experimental designs with respect to parameter estimation. Criteria for model discrimination are presented. Finally, field studies of the spatial structure of transport variabilities are reviewed and the role of variability structure in determining the appropriate scale for a transport description is discussed. The influence of measurement scale on model discrimination and parameter estimation is examined using data from recent field studies. These examples illustrate that an understanding of the structure of variabilities can greatly reduce sample requirements and enhance process identification and model calibration efforts. Possible areas of future research also are discussed.
Assessments of non-point source pollution, with mathematical models designed to produce multicolored maps, are now being used in the decision management arena. This has been possible primarily because of the marriage of solute transport models to geographic information systems that add a geo-referenced dimension to transport models. Albert Einstein said that “everything must be made as simple as possible, but not simpler.” The utility of relatively simple vulnerability maps, which have been produced at regional scales with geographic information system technology, is undermined by significant uncertainties related to model and data errors. In this chapter, the three most commonly used methods for characterizing simulation uncertainties are discussed: sensitivity analysis, first-order analysis, and Monte Carlo analysis. Examples of each method are presented.
Many soil properties that are used as parameters in solute transport models vary in space. Their use as parameters is assumed to be valid provided the assembly of values remains constant with time. Nonlinearity in models interacts with this variance giving possibly misleading results when a model is used, and care is needed with both simple and spatial averaging procedures. Averaging or interpolating the parameter before running a nonlinear model does not give the same result as running the model and then averaging or interpolating the results. This difference could be important for some uses to which solute transport models are put, particularly for large areas. A simple test for nonlinearity is suggested for use as part of sensitivity analysis. It is important to know how the variance changes as the area over which a model is used increases. This makes the variogram of the property used as a parameter a relevant part of sensitivity analysis. Land use may affect both the properties of the soil that determine solute transport and the generation of nitrate within the soil; however, sensitivity analysis of models describing land use effects may be difficult, because it is hard to see how a system for analyzing such sensitivity would differ from a system used in practice to provide decision support. The results of the study support the conclusion that capacity-type models will be the most useful for simulation of solute transport in large areas.
A GIS hydrologic model couples a digital description of the geographic features of the flow environment with a physical or statistical description of the mechanisms governing the flow and transport processes taking place in that environment. An 8-step procedure is presented for using a combination of GIS and hydrologic models to describe the transport of agricultural pollutants in the vadose zone. The State Soil Geographic (STATSGO) database of the USA provides a soil description suitable for regional studies. A spatial statistical analysis of the occurrence of nitrate in Texas groundwaters shows that the character of the hydrogeologic unit is the dominant variable governing the regional pattern of contamination, overshadowing spatial patterns of nitrate fertilizer application. Nitrate accumulation in groundwater appears to occur more in arid west Texas and the High Plains than in humid east Texas, thus questioning the axiom that contamination is greater where recharge is greater.
The application of GIS to environmental modeling depends on the quality and availability of environmental data. Environmental data include soil and soil-landscape attributes such as organic matter content and percentage of slope, respectively. These data are available in hardcopy as soil survey reports or in the following digital formats: SSURGO (Soil Survey Geographic), STATSGO (State Soil Geographic), and NATSGO (National Soil Geographic) databases; however, these databases do not contain many of the soil inputs needed to predict pollutant transport or fate. The USDA-NRCS (Natural Resources Conservation Service)-NSSC (National Soil Survey Center) is attacking this problem on two fronts. The first is the implementation of NASIS (National Soil Information System), which will improve delivery of soil and soil landscape-related data to the public and private sectors. NASIS will provide the users of GIS and other modeling techniques with primary soil data, such as transect and pedon observations. NASIS also will provide users with normalized primary soil survey data, such as aggregated field data, that are specific to the geographic extent of any given soil mapping unit within a soil survey area. The second is improving the compatibility between soil attribute data and computerized models. This issue is being addressed by making existing soil database attributes and structures available to researchers and modelers during model development, and by collecting or estimating model-sensitive inputs that are not readily available in current soil databases.
The use of mechanistic models at regional scales requires parameters that represent the soil hydraulic properties of the area and their variability. Because of the large number of samples required to characterize an extensive area such as a watershed, methods to estimate soil hydraulic properties from simpler data or from a small number of measurements are alternatives to making extensive field measurements. Readily available data that can be used to obtain hydraulic parameters include soil texture, bulk density, porosity, and 33 kPa water content. The choice of estimation method depends on the type of data available and the possibility of making field measurements. Examples of estimating certain important parameters are presented. Finally, a method of assessing the quality of the estimations must be chosen. The correlation of measured properties with estimated is not always the best method. Because a simulation model integrates the effects of soil hydraulic properties with climate variables and, if simulated, with plant factors, the targeted output should be the basis of comparisons. The results of simulations carried out using a soybean model, GLYCIM, indicate that estimated soil properties may only be useful if long term averages of soybean yield are desired.
Geographical information systems (GIS) are able to store and manipulate vast amounts of spatial data and are ideally suited to the investigation of non-point source pollution of water; however to be effective, a GIS needs accurate databases (both in parameter magnitude and spatial resolution) for the key soil, climate, and management attributes that affect agrichemical fate and transport. Collection of these data is the most costly and time consuming step in applying GIS. The quality of these databases is often the limiting factor in the success of the application of GIS to non-point source pollution. In recent years several new approaches have been developed and tested for inexpensive and rapid data collection of soil attributes that are important in determining agrichemical fate and transport. Of these methods, ground-penetrating radar and electromagnetic induction appear particularly well suited for rapid site characterization. These methods have been used for reconnaissance mapping to guide subsequent intensive sampling, have been used to measure surrogate variables in place of more difficult-to-measure variables, and to collect ancillary data used to improve the spatial mapping of soil attributes. A brief description of these two methods is given and an illustration is presented to show how geophysical methods can be used to improve a spatial database of soil organic carbon.
As geographical information systems (GIS) are increasingly being applied to surface and subsurface flow and transport modeling issues, it becomes important to more clearly define potential advantages and achievable objectives with this technology. This chapter describes an integrated conceptual framework for predicting basin-scale solute loading rates through and from the vadose zone. The approach conceptually couples the ARC/INFO geographical information system with a deterministic variably-saturated flow and transport model (HYDRUS), an unsaturated soil hydraulic property database (UNSODA), a digital soil database (STATSGO) in conjunction with pedo-transfer functions (PTFs), and a geostatistical software package (GEOPACK). Suggestions are made on how best to integrate currently available or future knowledge of surface hydrology, vadose zone hydrology, and groundwater hydrology so as to more effectively address specific nonpoint source pollution problems. Whereas computations by the different components within the proposed integrated architecture can be made to run interactively, individual components will keep their identities at previously defined positions. The resulting integrated approach involving loosely-coupled independent technologies should provide new possibilities for addressing non-point source subsurface pollution problems.
The one-dimensional version of the Unsatchem model was interfaced to a geographic information system (GIS). Unsatchem is a finite-element model of water flow in the vadose zone coupled with multicomponent solute transport, production and transport of CO2, and heat in addition to a major ion model of chemical speciation and kinetics. The new version, Unsatchemgeo, runs in a geographic context in association with the GIS, such that data required by the model are obtained directly from database tables and computed results are written to other database tables. This direct method for input and output of data is faster by a factor of 4.7 than an earlier approach utilizing the native macro language of the GIS. Support programs were written for initial generation of the INFO database and input of data in text form. To investigate the advantages and limitations of the coupled approach, Unsatchemgeo simulated water movement in a tile-drained agricultural field. Observed cumulative drainage flow from the tile drain system is greater than calculated drainage through the lower boundary of the model, by a factor in the range 1.2 to 2.1, for all times and at all locations. Also, the onset of increased tile drain flow occurs prior to the start of irrigations. These results and observations indicate the existence of water sources for drainage other than water supplied during irrigation of this field.
Land management practices can affect shallow aquifers. Methods for assessing the sensitivity of shallow aquifers to pollution have relied more on the use of watershed simulation models, and on defining the hydrogeologic units and less on quantitative measures of aquifer properties or behavior. The objectives of this study were to develop a baseflow days map for the SWAT watershed model, and to demonstrate the value of including base-flow information in rating the sensitivity of shallow aquifers. Baseflow days and base-flow/total flow (baseflow sensitivity) were used to map the sensitivity of regional aquifers. Comparisons of baseflow measures against independent local estimates showed good agreement. Examples are given to show that inclusion of baseflow within the framework of regional vulnerability models such as DRASTIC would give more accurate assessment of pollution potential to shallow aquifer systems.
In recent years concern about pollution of groundwater by pesticides has grown. Threats are posed to groundwater, both with respect to the magnitude of the leaching, and the total area involved. A one-dimensional pesticide leaching model, PESTRAS, in combination with a geographical information system (GIS) was used to calculate the leaching potential of pesticides into the groundwater. Calculations were performed for unique combinations of soil texture, organic matter content, groundwater-depth-class, land-use, and climate. Model-inputs were derived from these basic spatially-distributed parameters using transfer functions. Spatial patterns of the pesticide leaching potential were obtained by combining the calculated results with geographic information. The number of unique combinations for which the model had to be run to get spatial patterns for the Netherlands could be reduced from 93 000 (which is the total number of relevant 500 × 500 m2 grid-cells in the Netherlands) to a manageable 897. With a realistic dose of 1 kg ha−1, atrazine (2-chloro-4-(ethylamino)-6- (isopropylamino)-s-triazine) concentrations in soil leachate were above the European Union (EU) drinking-water standard (0.1 µg L−1) in 78% of the agricultural area, for bentazone (3- isopropyl-1H-2,1,3-benzothiadizin-4-(3H)-one 2,2-dioxide) this figure amounted to 89%. In general, peat soils were invulnerable to pesticide leaching, whereas sandy and loamy soils with low organic matter contents were very vulnerable (concentrations above 1 µg L−1).
Improved simulation-based methodology is needed to help identify broad geographical areas where potential NO3-N leaching may be occurring from agriculture and suggest management alternatives that minimize the problem. The Nitrate Leaching and Economic Analysis Package (NLEAP) model was applied to estimate regional NO3-N leaching in eastern Colorado. Results show that a combined NLEAP/GIS technology can be used to identify potential NO3-N hot spots in shallow alluvial aquifers under irrigated agriculture. The NLEAP NO3-N Leached (NL) index provided the most promising single index followed by NO3-N Available for Leaching (NAL). The same combined technology also shows promise in identifying Best Management Practice (BMP) methods that help minimize NO3-N leaching in vulnerable areas. Future plans call for linkage of the NLEAP/GIS procedures with groundwater modeling to establish a mechanistic analysis of agriculture-aquifer interactions at a regional scale.
An overview is presented of previously published work by Corwin and his colleagues concerning the application of a geographic information system (GIS) to non-point source (NPS) pollutants in the vadose zone. Two different GIS-based approaches are described for the prediction of the areal distribution of a NPS pollutant, specifically salinity, at a basin scale. The first approach couples a regression model of salinity development to a GIS of soil salinity development factors (i.e., permeability, leaching fraction, and groundwater electrical conductivity) for the Wellton-Mohawk Irrigation District near Yuma, AZ, during the study period 1968 to 1973. The regression model predicts the composite salinity of the root zone (i.e., top 60 cm.). Areas of low, medium, and high salinization potential are delineated for the entire 44 000 ha irrigation district. Measured salinity data verified that 86% of the predicted salinity categories were accurately predicted. The second approach loosely coupled the one-dimensional, transient-state solute transport model, TETrans, to the geographic information system ARC/INFO. Slightly less than 2400 ha of the Broadview Water District located on the west side of central California's San Joaquin Valley are being used as the test site to evaluate the integrated GIS/transport model during the study period 1991 to 1996. TETrans uses the GIS as a spatial database from which to draw its input data. Preliminary simulations are presented for the main growing season of 1991. Display maps show spatial distributions of soil salinity profiles to a depth of 1.2 m, irrigation efficiencies, drainage amounts, and salt loading to groundwater over the 2396 ha study area. These maps provide a visual tool for making irrigation management decisions to minimize the environmental impact of salinity on soil and groundwater. The first approach is best suited for areas where steady-state conditions are approximated, while the second approach can be used under transient-state conditions.