Articles Currently Under Peer Review by the URISA Journal
WATER RESOURCE APPLICATIONS OF GIS
John P Wilson, Helena Mitasova, and Dawn Wright
INTRODUCTION
Water resource applications of GIS are concerned with the hydrologic cycle and related processes. They are multi-faceted because: (1) many of the problems involve interactions between the hydrosphere, atmosphere, lithosphere, and biosphere; (2) solutions must serve competing groups of users; and (3) many of the important hydrologic processes have local, regional, national, and global dimensions (Naiman et al 1997; National Research Council 1999). Moreover, it is sometimes difficult to translate research outcomes into management strategies because much of the fundamental hydrologic research is conducted at specific sites or on small plots and many of the management strategies are focused on watersheds and/or administrative jurisdictions (Figure 1).
Figure 1. The process, problem, and geographic domains that might be used to classify water resource applications of GIS.
Available in Hard Copy only!
The immediate challenges in the water resource domain are:
These challenges are substantial and a range of solutions will be required because of the dramatic change in watershed management that has occurred during the past 5-10 years. There has been a shift from large government-directed regulatory programs towards local initiatives with government providing some support. The main participants are land owners, often organized into associations, such as the Landcare programs in Australia and New Zealand or watershed associations here in the US (EPA has already over 4,000 such associations registered). This will have a profound impact on the GIS tools that are being developed for water resources management. The target is no longer large government organizations with professional staff and we will need tools for retrieving and analyzing watershed information that can be used by people who are not specialists and are located in many different places. That means that a wider range of different tools at different levels are needed, from complex and sophisticated to very simple ones. These tools will need to operate at the watershed level in the future as well. The National Research Council (1999), for example, recently argued that watersheds as geographic areas are the natural organizing units for dealing with the management of water and closely related problems.
MAJOR GISCIENCE CONTRIBUTIONS AND SIGNIFICANCE
Modern life as we know it depends on our ability to match the supply and demand of water of appropriate quality to specific communities and users at specific times or rates. Our cities, farms, parks, and recreation areas all require water and their success (i.e. sustainability) relies on natural and human water delivery systems. Large amounts of time and effort are invested in learning more about the spatial and temporal patterns and characteristics of individual hydrologic processes so we can anticipate, manage, and modify system behavior to sustain modern lifestyles and prevent shortages (droughts), surpluses (floods), and resource impairment (pollution). Concerns about numerous issues, such as population growth, point source pollution, soil degradation, food supply, and energy have eased somewhat over the past years with many positive trends. Several other water-related issues, notably those concerned with water supply, non-point source pollution, and surface and groundwater quality impairment are still issues of great concern globally.
Solving this second set of water resource problems will require an improved understanding of the fundamental physical, biological, economic and social processes, and a better knowledge of how all these components operate together within watersheds. For example, the National Research Council (1999, 2-8) recently identified five sets of improvements that will be required to improve our management of water resources:
Viewed this way, water resource assessment and management are inherently geographical activities. Some combination of GIS and simulation models will be required to improve our knowledge in these areas. GIS offers powerful new tools for the collection, storage, management, and display of map-related information, whereas simulation models can provide decision-makers with interactive tools for understanding the physical system and judging how management actions might affect that system (National Research Council 1999). The five subsections that follow illustrate some of the ways in which GIS has already been used to advance water resource management.
Management and Delivery of Data
The development of new satellite sensors, other data capture tools, new data delivery options has expanded the accessibility and reduced the cost of many hydrologic data sets. Many of these changes are linked to the World-Wide Web (WWW) and role of GIS in massive, far-reaching, and on-going information technology (IT) developments, such as digital libraries, data warehouses, data mining and universal networking, have greatly expanded hydrologic data accessibility (Openshaw 1997, Newton et al 1999). The University of Arizona has compiled a list of approximately 300 land-surface hydrology data links (see http://www.hwr.arizona.edu/ hydro_link.html for details). GIS has enabled government agencies and private organizations to extend the delivery of their data from tables (numbers) to maps, and to support various forms of spatial searches for relevant data. A good example of the latter is the EPA "Surf Your Watershed" site which allows the users to get water quality data in the form of maps and numbers (see http://www.epa.gov/surf/ for details). These types of capabilities have an enormous impact beyond research and management because they can influence, for example, the values of real estate or decisions on business locations.
These types of opportunities also elevate the importance of metadata (i.e. information about where and when and how the data were collected and by whom). The Federal Geographic Data Committee (1995) has proposed a national spatial data infrastructure (NSDI) and specified guidelines for describing the minimum metadata required for different types of GIS information. This metadata is required for users to decide if a particular data set is adequate for their particular purpose(s).
GIS and Hydrologic Modeling
GIS has influenced the development and implementation of hydrologic models at several different levels. The examples that follow also illustrate how GIS has been used to address water supply, water quality, and storm-water management problems in several different contexts.
First and foremost, GIS have provided new opportunities to develop and run fully distributed models efficiently. These models take into account and predict the values of studied phenomena at any point within the watershed (e.g. Vieux 1991; Julien et al 1995; Mitasova et al (1996); Vieux et al 1996; Mitas and Mitasova 1998). This is very important from the point of view of management, as it allows users, for example, to identify the location of possible sources of pollution.
Second, GIS has also allowed users to run more traditional lumped models more efficiently and to include at least some level of spatial effects by partitioning entire watersheds into smaller sub-watersheds. Hence, Shamsi (1996) combined a planning level GIS with the Penn State lumped parameter Runoff Model (PSRM) and used them to implement a watershed-wide storm-water management plan in one such application. Storm-water management aims to prevent or mitigate the adverse impacts related to conveyance of excessive rates and volumes of storm-water runoff. Watershed-wide approaches are required to avoid shifting the location and/or increasing the magnitude of the problem downstream. The GIS was used to estimate the physical site parameters required by the model. Both vector and raster systems were used depending on the size of the study area (watershed) and several of the inputs were derived from simple GIS overlays and lookup tables. PSRM is a single event simulation model that incorporates Soil Conservation Service (SCS) techniques for infiltration, the kinematic wave method for overland flow, and non-linear routing for storage. The model was calibrated with observed hydrograph data, and used to simulate runoff hydrographs for various durations and frequencies and to create peak flow presentation and release rate tables from the simulated hydrographs. The information summarized in these tables was then used to create a watershed release rate map that satisfied the requirements of the Stormwater Management Act of Pennsylvania (1978) and provided a practical tool for implementing storm-water management plans. The adoption of this approach in six of Pennsylvania’s 356 designated watersheds indicates that the PSRM and GIS integration offers cost-effective and technically sound solutions to Pennsylvania’s watershed-wide storm-water management problems. Djokic and Maidment (1991) used ARC/INFO to simulate the drainage system and assess whether or not the existing drainage system in a portion of the City of Asheville, North Carolina, can accommodate 10- and 25-year return period design flows. Their approach used the rational method to examine the contributions from surface terrain (i.e. overland flow), man-made structures (i.e. pipes and channels), and storm water intakes.This type of application is now very common and numerous lumped parameter models (HEC-1, HEC2, MODFLOW, SHE, SWAT, etc.) have been linked to GIS in these ways to predict surface and ground water flows. Orzol and McGrath (1992), for example, described how the structure of MODFLOW was altered to facilitate its integration with ARC/INFO and they demonstrated that the results were the same as if the model was run as a standalone product. Similarly, Hellweger and Maidment (1999) automated a procedure to define and connect hydrologic elements in ARC/INFO and ArcView and write the results to an ASCII file that is readable by the
Hydrologic Engineering Center's Hydrologic Modeling System (HEC-HMS). The modular structure and availability of source code have favored the use of the GRASS GIS in many of these environmental modeling applications (see Mitas et al 1996; Vieux and Gauer 1994; Vieux et al 1996; Mitas and Mitasova 1998 for additional examples). Watkins et al (1996) compared the advantages and disadvantages of different GIS/model interfaces and showed how the spatial analysis and visualization capabilities of GIS could be used to improve parameter estimation/determination, grid design and scale effects, and the sensitivity of model outputs to parameter uncertainty and model discretization. Wilson (1999a) reviewed many of the recent attempts to develop models inside GIS and geographic modeling systems. The latter aim to provide libraries of landscape simulation components from which watershed simulation models can be assembled to represent user-specified processes and problems in watersheds of interest (e.g. Peters 1995; Leavesley et al 1996a, b). The accomplishments of the Danish Hydraulic Institute are particularly noteworthy in this regard. They have implemented numerous modeling systems for river basins, urban drainage, sewer systems, rivers and channels, estuaries, and coastal waters during the past decade and since 1998 have embarked on an ambitious program to link their models with the ESRI family of GIS products. Many of their modeling systems now support GIS data transfer and one, MIKE BASIN which provides a versatile decision support system for integrated water resources planning and management, runs inside the ArcView GIS.Third, GIS has been used to transform what were originally site-specific models into spatially distributed models. Carbone et al (1996), for example, combined GIS and remote sensing technologies with the SOYGRO (Wilkerson et al 1983) physiological soybean growth model and used them to predict the spatial variability of yields in Orangeburg County, South Carolina. This model relates the major processes of soybean growth (photosynthesis, respiration, tissue synthesis, translocation of protein, senescence, etc) to environmental conditions. SOYGRO has been tested in a variety of environments and has proven reliable in estimating yield in well-managed conditions (Curry et al 1990). The ARC/INFO GIS was used to organize the meteorological, soil and crop management inputs, and the SOYGRO model was run for 40 combinations of weather and soil conditions over a six-year period (1986-91) in this instance. The results showed that the spatial variability in simulated county yield was large and linked to soil moisture availability. This soil property is a function of available water holding capacity and the timing and amount of precipitation, both of which varied greatly across space. Carbone et al (1996) concluded that the examination of spatial patterns of simulated yield improved county production estimates and identified vulnerable areas during droughts.
These types of assessments take many different forms and have been conducted for larger areas as well (Wilson 1999b). Corbett and Carter (1996), for example, showed how GIS can be used to: (1) synthesize and integrate many more data than in the pre-computer (pre-GIS?) era; and (2) shift the design of agro-ecological and agroclimatological studies towards user-specified classifications. Their analysis focused on Zimbabwe, a semi-arid country where a national agroecological classification and map, the Natural Regions scheme (Vincent and Thomas 1960), has been widely used in agricultural research and policy-making. This map used rainfall and temperature data to calculate effective rainfall and vegetation to interpolate this variable between stations. Corbett and Carter (1996) constructed seasonal rainfall surfaces for Zimbabwe using decadal (i.e. ten day) rainfall data (82-99 stations; 31 years of data), the African DEM (13,400 grid points) produced by Hutchinson et al (1996), and the ANUSPLIN climate interpolation procedures described by Hutchinson (1995a, b). They generated surfaces showing mean rainfall and annual rainfall anomalies to describe the main rainfall period (March-October) for Zimbabwe in terms of rainfall variability. They demonstrated that the natural regions experienced considerable spatial variability in terms of mean and inter-seasonal variability of rainfall. Corbett and Carter (1996) then combined these surfaces with those of Deichmann (1994) to show that only 19% of Zimbabwe’s population lives in areas that can expect to receive more than 600 mm of rainfall (which serves as an approximate threshold for maize cultivation in southern Africa) with 75% probability.Fourth, GIS is sometimes used to vary model inputs and compare model outputs with field data in hopes of improving the scientific basis of key water quality policies and management plans. Inskeep et al (1996), for example, compared several modeling approaches that might be applicable for classifying the USDA-NRCS County Soil Survey Geographic database (SSURGO; Bliss and Reybold 1989, Reybold and TeSelle 1989) soil map units according to their leaching potential. They also used detailed site-specific measurements in some of their model runs and they compared the model results with observed data collected at a field site in southwestern Montana. Data from a two-year field study of pentafluorobenzoic acid, 2,6-difluorobenzoic acid, and dicamba (3,6-dichloro-2-methoxybenzoic acid) transport in fallow and cropped systems under two water application levels were compared to simulations obtained using the Chemical Movement through Layered Soils (CMLS) and Leaching and Chemistry Estimation (LEACHM) models. CMLS is a 1-dimensional solute transport model that uses a piston flow approach to simulate the vertical movement of selected chemicals through the agricultural root zone on a layer by layer basis (Nofziger and Hornsby 1987). LEACHM is a 1-dimensional finite difference model designed to simulate the movement of water and solutes through layered soils that has been validated and used as a predictive tool at the plot and field scale (Wagenet and Hutson 1989; Wagenet et al 1989). Several attempts have been made to combine both of these models with GIS databases for regional scale assessments of leaching behavior (e.g. Petach et al 1991; Foussereau et al 1993; Hutson and Wagenet 1993; Wilson et al 1993, 1996).
Inskeep et al (1996) varied the resolution of model input parameters according to different sources of data. Model inputs were obtained primarily from detailed soil profile characterization and site-specific measurements of precipitation, irrigation, and pan evaporation for one run (Case 1). LEACHM predictions were also generated using estimated conductivity and retention functions from SSURGO textural data (Cases 2 and 3). CMLS predictions were generated using detailed site-specific measurements (Case 1), and volumetric water contents estimated from SSURGO textural data and daily water balance estimated from WGEN (Richardson and Wright 1984) and the MAPS (Nielsen et al 1990) climate database (Cases 2 and 3). Comparison of observed and simulated mean solute travel times produced the following results. First, both the LEACHM and CMLS performed adequately with high-resolution model inputs. Second, model performance declined when field conditions were conducive to preferential flow. Third, saturated hydraulic conductivity values estimated from regression equations based on textural data were problematic for generating adequate predictions using LEACHM. Fourth, the CMLS predictions were less sensitive to data input resolution, in part because the CMLS provides an oversimplified description of transport processes. These results demonstrate the importance of model validation and suggest why model predictions based on GIS-based model input data sets with low spatial resolution may not accurately reflect transport processes occurring in situ.The future is some way off, in part, because geographic information technologies are relatively new and still near the lower end of the growth curve in terms of: (1) applications, and (2) their influence as tools on the ways in which scientific inquiries and assessments are conducted (Goodchild 1996). Several additional challenges related to our knowledge of specific processes and scale effects that must be overcome to achieve this future are noted below as well. The National Research Council (1999, 139-63), for example, reviewed some of these same activities and concluded that many of our existing models are inadequate for watershed management. New models are required that are directly linked to geographic information and decision support systems, incorporate all facets of watershed management, and span a variety of scales for application. The National Research Council (1999) envisaged a future in which models were as easy to use as a typical word processor or spreadsheet in order to serve both those that need them and those that created them.
New GIS Data and Tools
The steady increase in the number and variety of functions incorporated in GIS that are suited to water resource applications during the past 5-10 years shows that some progress has been made. This trend is best exemplified by the GRASS GIS environment whose open architecture is particularly suited to the rapid prototyping of new functions in support of environmental modeling applications. The incorporation of several new terrain analysis tools, thin-plate splines, kriging, and related geostatistical techniques represent very important innovations in this respect (e.g. the inclusion of ANUDEM (Hutchinson 1989) elevation gridding procedure in ARC/INFO (Versions 7.0 and higher)). ANUDEM and TOPOGRID (as it is called in ARC/INFO) take irregular point or contour data and create square-grid DEMs. The procedure automatically removes spurious pits within user-defined tolerances, calculates stream and ridge lines from points of locally maximum curvature on contour lines, and (most importantly) incorporates a drainage enforcement algorithm to maintain fidelity with a catchment's drainage network (Hutchinson 1989). The increased availability of GPS-derived elevation data (Twigg 1998) and difficulty of using published USGS DEMs for hydrologic studies documented by Hammer et al (1994), Zhang and Montgomery (1994), Hodgson (1995) and Mitasova et al (1996) suggest an important role for these types of tools in the future.
Several recent projects have also compared tools and/or input data. In one such study, Bolstad and Stowe (1994) evaluated the accuracy of elevations, slopes, and aspects computed from USGS 30 m and SPOT-STX DEMs. Their results showed that the Gesalt Photomapper-derived USGS DEM provided a better representation of microtopography. Gao (1996) examined the impact of DEM resolution on the accuracy of terrain representation and slope gradients in three distinctive study areas. The results showed that representation accuracy decreased moderately at intermediate resolutions and sharply at coarse resolutions in all three types of terrain. Resolution changes also had a large impact on computed slope gradients. One would expect even larger impacts for topographic attributes that are calculated as second derivatives, such as plan and profile curvature (Moore 1996). Carrara et al (1997) defined a series of objective criteria for evaluation of the quality of digital terrain models derived from contour lines. These criteria were used to evaluate four different interpolation procedures: the r.surf.contour procedure in GRASS (Version 4.1), the MDIP procedure developed by Carla’ and Carrara (Carrara 1988; Carla’ et al 1997), the ArcTin procedure in ARC/INFO (Version 7.0), and the Terrain Modeler procedure in Intergraph’s MGE GIS (Version 5.0). The first two procedures generated square-grid DEMs and the last two procedures generated TINs. These methods were applied to three sample areas and the results showed that the MDIP and Terrain Modeler techniques performed best in that they produced terrain models that reflected the ground surface as expressed by the input contour lines.There has also been a gradual but steady increase in the spatial content of hydrologic data sets. Hutchinson et al (1996), for example, describe the development and distribution of a gridded topographic and mean monthly climate database for the African continent. The monthly mean precipitation and temperature grids were prepared by applying fitted thin plate splines to the new Africa DEM. The final surfaces interpolate monthly mean temperatures to within standard errors of about 0.5oC and monthly mean precipitation to within errors of about 10-30% (Hutchinson et al 1996). Similarly, Graham et al (1999) describe the development of a new data set of watersheds and river networks that can be used to route continental runoff to the appropriate coast (i.e. ocean or inland sea). This data set includes watershed and flow direction information, as well as supporting hydrologic data, at 5', 1/2o, and 1o resolutions globally. Both of these data sets will be useful in fully coupled land-ocean-atmosphere models, terrestrial ecosystem models, and macroscale hydrologic modeling studies.
The shift in conceptual paradigms of soil survey and mapping that has occurred during the past 30 years represents another important innovation (Burrough et al 1997). The early models, exemplified by the STATSGO and SSURGO databases, used crisp classes in attribute space linked to crisply delineated mapping units in geographical space. A series of recent models has utilized fuzzy classification and geostatistical interpolation for simultaneously handling continuous variation in both attributes and location (see McBratney and Odeh 1997 for a description of the basic strategy). These methods mean that the values of soil properties obtained when a GIS is queried are increasingly likely to be estimates derived by methods of spatial interpolation such as kriging from actual data stored in the GIS. These changes are likely to improve both the model inputs and the ways in which uncertainty and error in model inputs and outputs are handled (Davis and Keller 1997; Lark and Bolam 1997). These concepts and the accompanying tools have been applied most often to soil attributes but they are equally adept at describing other types of environmental variation (Burrough 1996b). The recent work of Bardossy and Disse (1993), Bardossy and Duckstein (1995), Mitasova et al (1996), Mitas et al (1996), and Mitas and Mitasova (1998) illustrate the potential benefits of using these types of innovations to develop spatially distributed hydrologic models.
Water Resource Decision Support Systems
Several efforts have been launched to develop and sustain water resource decision support systems. Some of these systems are aimed at research applications and others are designed to support specific watershed management goals. The examples described below are instructive on two counts: (1) they illustrate recent accomplishments and shortcomings; and (2) they indicate the types of training and skills that water resource specialists are likely to need in the 21st century. Two recent systems are reviewed here to illustrate the accomplishments and skills that are likely to be required to develop and use these systems.Paniconi et al (1999) reviewed of the strengths and weaknesses of GIS and explained why distributed hydrologic models typically rely on GIS, data visualization, and other software tools for pre- and post-processing, and as complementary components of decision support systems. They developed a decision support system to estimate soil moisture from satellite measurements and validate these estimates using ground truth measurement and catchment scale hydrologic modeling. Their initial integration efforts used standard data formats and the creation of graphic user interfaces for data and tool management and their more recent work has used CAD frameworks. These frameworks consist of software infrastructures that were developed to integrate uncooperative, often proprietary tools, in the world of computer aided design. The latter approach is based on a data flow paradigm through which the modular components of an application-specific system can be connected. Such an approach may dramatically reduce the time and effort devoted to tool and data integration although such systems may only be suited to projects involving small groups of research scientists and care must be taken to insure that these systems do not influence the direction of the research itself. Clark (1996) has observed the last problem in other water resource applications, and the potential problems may be compounded in situations where the science is very complicated and/or poorly understood (as illustrated in the next example).
Downs and Priestnall (1999) developed a fluvial geomorphology GIS to explore river channel adjustment processes and patterns and then tried to evaluate the advantages and disadvantages of this system. They thought that the system was useful in the sense that it had automated the estimation of several of the key parameters and that this would eventually allow them to test a series of specific hypotheses related to river channel adjustment. However, they also concluded that their system was impenetrable to non-GIS specialists (like many other highly customized applications of GIS) and that most users would be unable to extend or substantially modify the system by themselves. In addition, this particular system (in its current form) can only address some of the processes thought to control river channel adjustment along specific reaches of a river. This is a problem in this instance because the complex interaction of many factors over varying spatial and temporal scales may always preclude a deterministic understanding of river channel adjustment at the watershed scale (Howard 1996).
The above two systems are very specialized and yet limited in terms of both the scientific understanding incorporated in them and the numbers and types of users who can use them. Additional problems may arise if model limitations are glossed over when GIS-based modeling applications are developed and unskilled users fail to recognize the impact of these shortcomings on the results (Burrough 1996a). This state of affairs characterizes many of the recent attempts to implement GIS-based soil erosion models for example (Wilson and Lorang 1999). There is also the danger that fieldwork for model calibration, validation, and scientific investigation will be neglected if model building is too easy (Burrough 1996a).
Improved Visualization
Advances in computer hardware and software have greatly improved visualization during the past 5-10 years. Mitas et al (1997) used several case studies to illustrate the role of exploratory cartographic visualization in the development and presentation of models of landscape processes and patterns. Their approach integrates knowledge from GIS, computer cartography and scientific visualization, and supports advanced visual analysis of multivariate georeferenced data by displaying multiple surfaces and volumes in an appropriate projection of 3-D space together with point and vector data. These visualizations can be implemented on the WWW and animated to show change through time. Dynamic cartographic models are now used as either a process of research and discovery with visualizations feeding refinements of models, or as a method of communicating complex measured or modeled geographic phenomena, which is frequently encountered in water resource applications. Other examples of work of this type include Hibbard and Santek (1989), Fisher et al (1993), Rhyne et al (1993), Hibbard et al (1994), and Brown et al (1995). Another important development has involved the extension of interactive visualization capabilities to cartographic models accessible through the Internet using VRML. Experiments are being performed with the aim of developing tools to visualize and manipulate hydrologic data and models using Virtual Reality in ways that will allow users to directly interact with the landscape and models (in real time) (e.g. Johnston and Reez 1998).
LINKAGES TO UCGIS RESEARCH CHALLENGES
The application examples highlighted in the previous section identified some important research challenges in addition to recent accomplishments. The UCGIS recently described the GIScience research agenda as a series of fundamental topics and the discussion that follows identifies individual topics and/or areas within these topics that are particularly relevant to the water resource domain. Each of the ten research topics identified by the UCGIS intersects with the challenges and problems encountered in water resource applications of GIS.
Spatial Data Acquisition and Integration
Several of the water resource applications described in the previous section have benefited from the explosive growth in automated data capture techniques, such as GPS, satellite imagery, and ground-based data acquisition systems. The new GPS opportunities, satellite sensors, and short-range remote sensing instruments that are likely to help with the determination of subsurface transport parameters and non-point source pollution levels are described by Twigg (1998), Wilkinson (1996) and Corwin (1996), respectively. Similarly, the recent deployment of the WSR-88D radar by the National Weather Service represents an important new data source for meteorological and hydrological projects (Crum and Alberty 1993; Vieux and Farajalla 1996). However, the use of these indirect measurements to estimate rainfall and runoff in severe storms has its own problems. Vieux and Bedient (1998) found that WSR-88D radar reflectivity could only be used to accurately estimate rainfall in operational flood forecasting when an appropriate reflectivity/rainfall rate relationship was used and rain gauge accumulations were available to calibrate the radar rainfall estimates for a severe storm in south Texas. The development of these tools offers new opportunities for many more people to participate in the data collection process and requires much better tools to integrate different types of geographic data and solve specific water resource problems. The increased interest in local environmental quality and advent of "field" GIS means that some of the integration will need to be performed in the field as well.Distributed Computing
The reliance on several different sources and types of data in most of the water resource applications described in the previous section indicates why the increasing availability and popularity of distributed computing will promote further GIS work in this application domain. The continued development of metadata concepts and tools will be required as well, and the overload at some map servers (especially those which serve maps of interest to large numbers of people, such as the EPA) demonstrates there is a strong need for high performance as well as distributed computing. High performance is required for processing the data and serving them over the Internet, and for running complex models and certain applications (e.g. flood prediction) in near-real time.
Extensions to Geographic Representations
Many of the water resource applications described in the previous section used traditional geographic data representations that are geared towards the representation of static situations on a planar surface at a specific scale because the data were derived from paper maps. Some of the applications have used fuzzy classification systems to represent data of varying exactness and degrees of reliability. Further work to refine these techniques and the methods used to convey this additional information to the user is required (e.g. De Gruijter et al 1997). There is also a need for more effective extensions to integrate GIS with dynamic modeling (e.g. Wesseling et al 1996). These extensions will have an especially large impact in this domain because different data representations are suited to different types of applications and most solutions will require several types of information drawn from varying sources.
Most of the modeling applications summarized in this paper incorporated precipitation, soil, topographic, and land cover information. Most precipitation data consist of point estimates (i.e. climate station measurements) although the WSR-88D weather radar and some of the new satellite sensors offer spatially distributed data. Topographic information may utilize the square grid, irregular point, contour, or triangulated irregular network models. Most of the soil and land cover data sets that are currently available consist of raster grids or polygons, and most river systems are presented as a series of links (stream segments) and nodes (stream junctions). There are many tricks involved in working effectively with these different data types (see Custer et al 1996; Inskeep et al 1996; Wilson et al 1996; Mackay and Band (1997); and Hellweger and Maidment 1999 for examples).
Kemp (1997a, b) recently advocated the design of a level of user interaction that would focus on the user’s concept of the field and hide lower level issues of field representation as far as possible. Kemp (1997a) proposed a series of rules to guide conversions between data models based on the number of spatial elements per unit area (i.e. the relative size or spacing of the spatial elements). Kemp (1997b) described several field variables whose values can be used to select appropriate conversion procedures when working with two or more spatial data models.
These ideas need to be developed further, since the choice of and conversion from one field model to another is fraught with difficulties (Heuvelink 1996). In a similar vein, better methods of spatio-temporal representation for multidimensional data are also required. Time is still not supported well enough and more sophisticated spatio-temporal analytical tools are needed (see Yuan (1999) and Renolen (2000) for extended discussion of current options and shortcomings). The increasing availability of 3-D data, especially for atmospheric and groundwater modeling, are likely to promote additional work concerned with the handling, analysis, and visualization of volumetric data and their change in time.
Cognition of Geographic Information
Some of the innovations noted towards the end of the previous section point to steady but sustained progress in terms of our cognition and presentation of objects. In addition, Mackay et al (1992) and Robinson and Mackay (1996) recently indicated how the disciplinary scientist and manager may be afforded the opportunity to work with landscape elements such as hillslopes, streams and valleys, and river reaches instead of fields, polygons, and pixels. These types of extensions, which rely on logic-based systems augmented with various forms of inexact reasoning, may be required to develop the types of easy to use models and decision support systems described earlier. Sustained progress in this area is likely to improve the effectiveness of digital libraries and water resource decision support systems as well as GIS.
Interoperability of GIS
Many water resource applications require multiple systems, data sources, and enormous quantities of time and effort are expended to integrate these components (e.g. Carbone et al 1996; Shamsi 1996). Some progress has been made with data sharing and both metadata concepts and tools are evolving quickly. However, the current strategies work best for information that was largely cartographic in origin and research is still required to formalize methods for representing other types of geographic phenomena and to develop standardized languages for describing operations. These types of innovations would make it easier to integrate GIS data into dynamic models and facilitate increased data sharing among the environmental modeling community (e.g. Paniconi et al 1999). The launch of several new local, state and federal data sharing programs, increased numbers of citizens interested in local water resource issues, and the continued growth in the popularity of distributed computing will increase the need for and benefits flowing from progress in this area.
Scale
This term refers to the level of detail at which information can be observed, represented, analyzed, and communicated. The development and evaluation of topographic and hydrologic databases that extend over large areas (regions) is an area of active research as illustrated by the following account of recent work exploring the characteristics of digital elevation models and their impact on hydrologic modeling.
Many recent studies, for example, have examined the sensitivity of computed topographic attributes to the choice of data source, structure, and/or cell size. In one such study, Hammer et al (1994) compared 30 m USGS DEMs with field data and found that they correctly predicted slope gradient at only 21 and 30% of the field sampling locations, respectively, in two 16 ha study sites in Atchison County, Missouri. Srinivasan and Engel (1991), Zhang and Montgomery (1994) and Mitasova et al (1996) found similar results, and numerous authors have argued that DEMs with spatial resolutions of 2-10 m are required to represent important hydrologic processes and patterns in many agricultural landscapes (Wilson 1999b). Numerous studies have also shown how the choice of data source and resolution can impact model predictions. Panuska et al (1991) and Vieux and Needham (1993) quantified the effects of data structure and cell size on Agricultural Non-Point Source (AGNPS) pollution model inputs and showed how the computed flowpath lengths and upslope contributing areas varied with element size. Vieux (1993) examined the sensitivity of a direct surface runoff model to the effects of cell size aggregation and smoothing using different sized windows. Moore et al (1993) examined the sensitivity of computed slope and steady state topographic wetness index values across 22 grid spacings for three moderately large (» 100 km2) catchments in southeastern Australia. Hodgson (1995) demonstrated that the slopes and aspects calculated from 30 m DEMs are representative of grid spacings two or three times larger than the original DEM grid spacing. Issacson and Ripple (1991) compared 1o USGS 3 arc-second and 7.5' USGS 30 m DEMs and Lagacherie et al (1996) examined the effect of DEM data source and sampling pattern on computed topographic attributes and the performance of a terrain-based hydrology model. Chairat and Delleur (1993) quantified the effects of DEM resolution and contour length on the distribution of the topographic wetness index as used by TOPMODEL and the model’s peak flow predictions. Wolock and Price (1994) and Zhang and Montgomery (1994) also examined the effects of DEM source scale and DEM cell spacing on the topographic wetness index and TOPMODEL watershed model predictions. Garbrecht and Martz (1994) examined the impact of DEM resolution on extracted drainage properties for an 84 km2 study area in Oklahoma using hypothetical drainage network configurations and DEMs of increasing size. They derived various quantitative relationships and concluded that the grid spacing must be selected relative to the size of the smallest drainage features that are considered important for the work at hand. Bates et al (1998) showed how high frequency information is lost at progressively larger grid spacings.More work of this type is required across a broad spectrum of data themes. The DEM results indicate the magnitude of this task and why only limited progress has been made with each of the original research tasks in this area listed by the UCGIS despite long-standing recognition of the implications of scale for geographic inference and decision-making. The gaps in our knowledge and lack of appropriate tools have important consequences for most of the water resource applications described in the previous section. Similarly, the advent of new, high resolution data sets for large areas will allow analysis and modeling to be performed at much greater detail than is done now and the handling of large sets in relation to scale is likely to emerge as a critical issue in the immediate future. See Wilson et al (1998) for an example of the type of research required here.
Spatial Analysis in a GIS Environment
This topic is important because several of the innovations identified by the UCGIS would produce immediate benefits in the water resource application domain. Clearly, the increased availability of large, geographically referenced data sets and improved capabilities for visualization, rapid retrieval, and manipulation inside and outside of GIS will demand new methods of spatial analysis that are specifically tailored to this data rich environment (Wilkinson 1996; Gahegan 1999). Similarly, new methods that incorporate and exploit the benefits of geostatistics are required. These methods would provide more accurate descriptions of key variables and improved diagnostics for error assessments and accuracy (uncertainty) determinations. Increased knowledge of these properties can be expected to improve the ways in which many types of environmental data are collected, stored, analyzed and visualized in the future (see Burrough et 1997; De Gruijter et al 1997; and Lark and Bolam 1997 for examples of soil survey applications).Other innovations are required because many of the data sets used in the water resource applications reviewed in the previous section were derived inside GIS. Additional work is required to refine and/or document the consequences of using specific methods. The choice of flow routing method, for example, can have a large impact on computed terrain attributes (Wolock and McCabe 1995; Desmet and Govers 1996). The current options include the D8 (deterministic eight node; O’Callaghan and Mark 1984) and Rho8 (random eight node; Fairfield and Leymarie 1991) single flow algorithms, FD8 multiple flow algorithm (Freeman 1991, Quinn et al 1991), and the DEMON stream tube algorithm of Costa-Cabral and Burges (1994). However, this is an active area of research and more modified forms of the FD8 algorithm and a new grid-vector and grid-triangular multiple flow routing algorithms were recently proposed by Quinn et al (1995), Mitasova et al (1996) and Tarboton (1997). Additional work is now required to know which of these algorithms works best with different types of source data (square-grid DEMs, contours, GPS data sets, etc) in specific environments (Wilson 1999b). The best method for a particular application will be the one that simulates or mimics the runoff processes occurring in that particular landscape. In addition, different methods may be suited to different parts of a landscape, as Mackay and Band (1998) have demonstrated for a series of lake-dominated landscapes in Ontario, Canada. The results of this type of work and the inclusion of new tools in GIS software will have important implications for the successful deployment of GIS in water resource applications.
The Future of the Spatial Information Infrastructure
The increased interest in local communities and environmental issues at all levels of government will require technical and institutional programs to support the creation and sharing of local knowledge. The new data capture tools and advances in distributed computing noted earlier provide important new opportunities to identify gaps in existing data, collect new data, and correct errors in existing data. There is an immediate opportunity to promote the accelerated growth and utilization of geographic information resources in meeting society’s water resource needs in many communities. The development of spatial information infrastructure can have a dramatic impact on the role which spatial information plays in the life of every citizen in many areas, including water resources. The availability of water resources information will have an impact on planning at every level - from government, through business and farmers, to citizens buying their homes. Research will be required to identify the best approaches for customizing the same information for different users and/or purposes.
Uncertainty in Geographic Data and GIS-based Analyses
The increased numbers of users with very different backgrounds who will be using water resource data to make important decisions, coupled with the issues raised in the previous section, elevates the importance of finding reliable methodologies for estimating, visualizing, and using uncertainty for a wide range of applications. This is important for spatial data in general, but it is especially important for water resource data where a small local change may have a dramatic impact. Several of the research projects cited earlier have tried to evaluate the uncertainty inherent in various data sets and/or analytical methods. That uncertainty exists in every phase of the geographic data life cycle, from data collection to data representation, data analyses, and final results is well known. However, our knowledge of uncertainty in geographic data and its consequences for water resource decisions made using GIS is very incomplete. More work like that of Weih and Smith (1997), who traced the influence of cell slope computation algorithms through to a common forest management decision, is urgently needed in the water resource domain.
GIS and Society
This connection is obvious because our continued prosperity depends on effective water resource management. GIS can help with the collection, storage, analysis, and visualization of key information and thereby help with the development of effective water resource programs and practices. Not all water resource problems require GIS and simulation models (e.g. Lovejoy 1997); however, those that do require technologically sophisticated solutions are likely to benefit from additional research and education to ensure that the GIS/modeling results can be interpreted and used appropriately. The educational challenges are addressed next.
LINKAGES TO UCGIS EDUCATION CHALLENGES
The research applications and challenges give some indication as to the types of skills and backgrounds that will be required by the next generation of water resource specialists. The UCGIS recently described the GIScience education agenda as a series of fundamental topics (referenced below with underlined text) and the following discussion identifies individual topics and/or areas within these topics that are particularly relevant to the water resource domain. Many of our universities will need to improve their supporting infrastructure and modify their curricula to provide the two sets of improvements that are required.
The first involves the strengthening of the GIScience curriculum. The UCGIS Model Curriculum project that is now underway is an important innovation in this regard although its final impact will depend on the redeployment of resources and rates of adoption. The current draft specifies education and training goals and content for four levels of users: (1) informed users; (2) disciplinary analysts; (3) GIScience analysts; and (4) GIScience developers. Most of the existing academic programs are aimed at level 1 and many of these programs will need to be reorganized and expanded to serve the final three levels identified in the model curriculum. These changes can be articulated as part of a second and much broader set of educational goals as noted below.
The second set of improvements is tied to multidisciplinary education and the need to build stronger and more substantial links between the science, engineering, and policy programs that intersect with the water resource application domain. Deliberate planning and skilful identification and negotiation of tradeoffs will be required to foster these types of linkages inside universities. The rewards of such an approach will be substantial although their exact character can be expected to vary by discipline. Bouma (1997), for example, advocated a future in which soil scientists operate as "knowledge brokers" with skills that can support both general assessments (synthesis) and detailed investigations (new research). Wilson and Burrough (1999) have advocated adding fuzzy classification, geostatistics, and dynamic modeling to physical geography curricula. Geographers, in general, will need to strengthen their computer and quantitative methods skills if they are to retain their key role in GIScience education, research, and outreach. Similarly, the computer science and engineering participants would benefit from formal geographical training. These examples indicate that we should equip the next generation of scientists with broad as well as deep knowledge and skills and the ability to communicate the goals, methods, results, and utility of their research at varying levels of certainty to a variety of stakeholders. The growth in professional education programs and their use of emerging technologies to deliver GIScience education (e.g. Penn State, USC GIScience Graduate Certificate Programs; ESRI Virtual Campus) may improve access, help to facilitate these types of changes, and divert some of the focus from research-driven graduate GIS education. The multi-disciplinary character of the water resource application domain adds another level of complexity to the task of integrating curricula serving the GIScience and water resource application domain.
POLICY IMPLICATIONS
Water resource management is crucial as we search for ways to build environmentally and socially sustainable communities and lifestyles. In some instances, we need to find new methods for changing supply and/or demand for water resources. In other cases, we need to find faster and more effective methods to identify and manage sources of pollution and minimize surface and ground water contamination. At other sites, the contamination has already occurred and our primary focus is clean up. GIS may contribute to problem solving in each of these instances. At a more general level, GI technologies may help to guide the adoption of water resource policy and promote a more efficient and equitable allocation of natural and community resources as we strive to achieve the above goals.
PRIORITY AREAS FOR RESEARCH AND EDUCATION
The National Research Council (1999) recently advocated a watershed management future that aims to develop careful, long-term solutions to problems and provides sustainable access to resources. Four sets of innovations will be required to achieve these goals:
The assessment of the current status of water resource applications of GIS offered in section 2 suggests that most if not all of these innovations can be implemented with the assistance of or possibly inside GIS. The linkages to the UCGIS research challenges documented in section 3 suggest that additional research will be required for this to happen. Some of the research challenges identified by the UCGIS two years ago are driven as much by changes outside GIS and their significance to water resource applications of GIS is modest at best. The topics concerned with spatial data acquisition and integration, distributed computing, interoperability of GIS, the future of the spatial information infrastructure, and the implementation of and impacts of GIS on society might be classified this way. Progress on the remainder of the research challenges outlined by the UCGIS will require substantial contributions from GIScientists and/or special attention to the water resource domain. Advances in three broad areas are required:
Sustained progress on these GIScience research challenges and delivery of the types of simulation models and spatial decision support systems envisaged by the National Research Council (1999) has tremendous implications for education as well. Of the two sets of necessary improvements mentioned in section 4, the need to build stronger and more substantial multi-disciplinary links is not receiving as much attention as the specification of the GIScience model curriculum (which is currently being attended to by a Model Curriculum Task Force funded by ESRI). In considering multidisciplinary education for the water resource application domain, there are three categories of students to be considered: (1) those who know about something water resources but nothing about GIS; (2) those who know something about GIS but nothing about water resources; and (3) those who know neither topic. One of the most pressing problems in reaching all three categories of students is how best to insert GIS-related education and training into curricula that are already quite full, particularly in engineering and agricultural programs. To this end, the "Learning with GIS" education challenge of the UCGIS is especially pertinent. GIS is an excellent teaching tool for introducing and exploring many aspects of water resources, including resource monitoring, water storage and flow in rural and urban communities, stream flow monitoring, surface and groundwater hydrology, irrigation engineering, farming practices, wetlands ecology, water pollution, and many others. Two high priority recommendations in the context of "Learning with GIS" include:
With respect to professional education, a set of modules should be developed that treats water resources from the point of view of the manager (working for a water management board, water district, extension office, county, state, federal government, etc) or a farmer. Rather than a "plug in", these modules should form the basis of a one- or two-day short course that might be offered over the web, as a video conference, or in conjunction with a professional association’s meeting or a water resources conference.
ACKNOWLEDGEMENTS
This manuscript was revised several times to incorporate the suggestions offered by delegates at the 1999 UCGIS Summer Assembly and several anonymous reviewers. We greatly appreciate their assistance and the opportunity afforded by the 1999 UCGIS Summer Assembly Program Committee to prepare this review and assessment of water resource applications of GIS.
REFERENCES CITED
Bardossy A and M Disse (1993) Fuzzy, rule-based models for infiltration. Water Resources Research 29: 373-82
Bardossy A and L Duckstein (1995) Fuzzy Rule-Based Modeling with Applications to Geophysical, Biological, and Engineering Systems. New York, NY, CRC Press
Bates P D, Anderson M G, and M Horrit (1998) Terrain information in geomorphological models: Stability, resolution, and sensitivity. In S N Lane, K S Richards, and J H Chandler (eds) Landform Monitoring, Modeling, and Analysis. New York, NY, John Wiley & Sons: 279-310
Bliss N B and W U Reybold (1989) Small-scale digital soil maps for interpreting natural resources. Journal of Soil and Water Conservation 44: 30-4
Bolstad P V and T Stowe (1994) An evaluation of DEM accuracy: Elevation, slope, and aspect. Photogrammetric Engineering and Remote Sensing 60: 1327-32
Bouma J (1997) The role of quantitative approaches in soil science when interacting with stakeholders. Geoderma 78: 1-12
Brown W M, Astley M, Baker T, and H Mitasova (1995) GRASS as an integrated GIS and visualization environment for spatio-temporal modeling. In Proceedings of Auto-Carto 12, Charlotte, North Carolina: 89-99
Burrough P A (1996a) Opportunities and limitations of GIS-based modeling of solute transport at the regional scale. In D L Corwin and K Loague (eds) Application of GIS to the Modeling of Non-point Source Pollutants in the Vadose Zone. Madison, WI, Soil Science Society of America: 19-38
Burrough P A (1996b) Natural objects with indeterminate boundaries. In P A Burrough and A U Frank (eds) Geographic Objects with Indeterminate Boundaries. London, Taylor & Francis: 3-28
Burrough P A, van Gaans P F M, and R Hootsmans (1997) Continuous classification in soil survey: Spatial correlation, confusion, and boundaries. Geoderma 77: 115-35
Carbone G J, Narumalani S, and M King (1996) Application of remote sensing and GIS technologies with physiological crop models. Photogrammetric Engineering and Remote Sensing 62: 171-9
Carla’ R, Carrara A, and G Federici (1987) Generazione di modelli digitali del terreno ad alto precisiione. Firenze, Quaderni Dip Ingegeria Civile
Carrara A (1988) Drainage and divide networks derived from high-fidelity digital terrain models. In C F Chung, A G Fabbri, and P Larsen (eds) Quantitative Analysis of Mineral and Energy Resources. Dordrecht, Reidel: 581-97
Carrara A, Bitelli G, and R Carla’ (1997) Comparison of techniques for generating digital terrain models from contour lines. International Journal of Geographical Information Science 11: 451-73
Chairat S and J W Delleur (1993) Effects of the topographic index distribution on predicted runoff using GRASS. Water Resources Bulletin 29: 1029-34
Clark M J (1996) Professional integrity and the social role of hydro-GIS. In K Kovar and H P Nachtnebel (eds) Application of Geographic Information Systems in Hydrology and Water Resources Management. Wallingford, International Association of Hydrological Sciences Publication No 235: 279-87
Corbett J D and S E Carter (1996) Using GIS to enhance agricultural planning: The example of inter-seasonal rainfall variability in Zimbabwe. Transactions in GIS 1: 207-18
Corwin D L (1996) GIS applications of deterministic solute transport models for regional-scale assessment of non-point source pollutants in the vadose zone. In D L Corwin and K Loague (eds) Application of GIS to the Modeling of Non-point Source Pollutants in the Vadose Zone. Madison, WI, Soil Science Society of America: 69-100
Costa-Cabral M C and S J Burges (1994) Digital elevation model networks (DEMON): A model of flow over hillslopes for computation of contributing and dispersal areas. Water Resources Research 30: 1681-92
Crum T D and R L Alberty R L (1993) The WSR-88D and the WSR-88D operational support facility. Bulletin of the American Meteorological Society 74: 1669-87
Curry R B, Peart R M, Jones J W, Boote K J, and L H Allen (1990) Response of crop yield to predicted changes in climate and atmospheric CO2 using simulation. Transactions of the American Society of Agricultural Engineers 33: 1383-90
Custer S G, Farnes P, Wilson J P, and R D Snyder (1996) A comparison of hand- and spline-drawn precipitation maps for mountainous Montana. Water Resources Bulletin 32: 393-405
Davis T J and C P Keller (1997) Modeling uncertainty in natural resource analysis using fuzzy sets and Monte Carlo simulation: Slope stability analysis. International Journal of Geographical Information Science 11: 409-34
De Gruijter J J, Walvoort D J J, and P F M van Gaans (1997) Continuous soil maps: A fuzzy set approach to bridge the gap between aggregation levels of process and distribution models. Geoderma 77: 169-95
Deichmann U (1994) A Medium Resolution Population Database for Africa. Santa Barbara, CA, National Center for Geographic Information and Analysis, University of California
Desmet P J J and G Govers (1996) Comparison of routing algorithms for digital elevation models and their implications for predicting ephemeral gullies. International Journal of Geographical Information Systems 10: 311-31
Djokic D and D R Maidment (1991) Terrain analysis for urban storm water modeling. Hydrological Processes 5: 115-24
Doe W W, Saghafian B, and P Y Julien (1996) Land use impact on watershed response: The integration of two-dimensional hydrological modeling and geographical information systems. Hydrological Processes 10: 1503-11
Downs P W and G Priestnall (1999) System design for catchment-scale approaches to studying river channel adjustments using a GIS. International Journal of Geographical Information Science 13: 247-66
Fairfield J and P Leymarie (1994) Drainage networks from grid digital elevation models. Water Resources Research 27: 709-717, 2809
Federal Geographic Data Committee (1995) Content Standards for Digital Geographic Metadata. Washington DC, Federal Geographic Data Committee
Fisher P, Dykes J, and J Wood (1993) Map design and visualization. Cartographic Journal 30: 136-42
Foussereau X, Hornsby A G, and R B Brown (1993) Accounting for the variability within map units when linking a pesticide fate model to soil survey. Geoderma 60: 257-76
Freeman G T (1991) Calculating catchment area with divergent flow based on a regular grid. Computers and Geosciences 17: 413-22
Gao J (1997) Resolution and accuracy of terrain representation by grid DEMs at a micro-scale. International Journal of Geographical Information Science 11: 199-212
Gahegan M (1999) What is Geocomputation? Transactions in GIS 3: 203-6
Garbrecht J and L W Martz (1994) Grid size dependency of parameters extracted from digital elevation models. Computers and Geosciences 20: 85-7
Goodchild M F (1996) The application of advanced technology in assessing environmental impacts. In D L Corwin and K Loague (eds) Application of GIS to the Modeling of Non-point Source Pollutants in the Vadose Zone. Madison, WI, Soil Science Society of America: 1-17
Graham S T, Famiglietti J S, and D R Maidment (1999) Five-minute, 1/2o, and 1o data sets of continental watersheds and river networks for use in regional and global hydrologic and climate system modeling studies. Water Resources Research 35: 583-7
Hammer R D, Young F J, Wollenhaupt N C, Barney T L, and T W Haithcoate (1994) Slope class maps from soil survey and digital elevation models. Soil Science Society of America 59: 509-19
Hellweger F L and D R Maidment (1999) Definition and connection of hydrologic elements using geographic data. Journal of Hydraulic Engineering 4: 10-8
Heuvelink G B M (1996) Identification of field attribute error under different models of spatial variation. International Journal of Geographical Information Systems 10: 921-35
Hibbard W L and D A Santek (1989) Visualizing large data sets in the earth sciences. Computer 22: 53-7
Hibbard W L, Paul B E, Battaiola A L, Santek D A, Voidrot-Martinez M-F, and C R Dyer (1994) Interactive visualization of earth and space science computations. Computer 27: 65-72
Hodgson M E (1995) What cell size does the computed slope/aspect angle represent? Photogrammetric Engineering and Remote Sensing 61: 513-7
Howard A D (1996) Modeling channel evolution and floodplain morphology. In M G Anderson, D E Walling, and P D Bates (eds) Floodplain Processes. Chichester, John Wiley & Sons: 15-62
Hutchinson M F (1989) A procedure for gridding elevation and stream line data with automatic removal of spurious pits. Journal of Hydrology 106: 211-32
Hutchinson M F (1995a) Interpolating mean rainfall using thin plate smoothing splines. International Journal of Geographical Information Systems 9: 385-403
Hutchinson M F (1995b) Stochastic space-time weather models from ground-based data. Agricultural and Forest Meteorology 73: 237-64
Hutchinson M F, Nix H A, McMahon J P, and K D Ord (1996) The development of a topographic and climate database for Africa. In Proceedings of the Third International Conference Integrating GIS and Environmental Modeling, Santa Fe, New Mexico, 21-25 January. Santa Barbara, CA, National Center for Geographic Information and Analysis, University of California
Hutson J L and R J Wagenet (1993) A pragmatic field-scale approach for modeling pesticides. Journal of Environmental Quality 22: 494-9
Inskeep W P, Wraith J M, Wilson J P, Snyder R D, and R E Macur (1996) Input parameter and model resolution effects on predictions of solute transport. Journal of Environmental Quality 25: 453-62
Issacson D L and W J Ripple (1991) Comparison of 7.5 minute and 1 degree digital elevation models. Photogrammetric Engineering and Remote Sensing 56: 1523-7
Johnston D and W Reez (1998) Visualization of geographic data using the CAVE. WWW document, http://www.gis.uiuc.edu/hpgis/visualization_htm
Julien P Y, Saghafian B, and F L Ogden (1995) Raster-based hydrologic modeling of spatially varied surface runoff. Water Resources Bulletin 31: 523-36
Kemp K K (1997a) Fields as a framework for integrating GIS and environmental process models: Part 1, Representing spatial continuity. Transactions in GIS 1: 219-34
Kemp K K (1997b) Fields as a framework for integrating GIS and environmental process models: Part 2, Specifying field variables. Transactions in GIS 1: 235-46
Lagacherie P, Moussa R, Cormary D, and J Molenat (1996) Effects of DEM data source and sampling pattern on topographical parameters and on a topography-based hydrological model. In K Kovar and H P Nachtnebel (eds) Application of Geographic Information Systems in Hydrology and Water Resources Management. Wallingford, International Association of Hydrological Sciences Publication No 235: 191-9
Lark R M and H C Bolam (1997) Uncertainty in prediction and interpretation of spatially variable data on soils. Geoderma 77: 85-113
Leavesley G H, Restrepo P J, Stannard L G, Frankoski L A, and A M Sautins (1996a) MMS: A modeling framework for multidisciplinary research and operational applications. In M F Goodchild, L T Steyaert, B O Parks, C Johnston, D R Maidment, M P Crane, and S Glendinning (eds) GIS and Environmental Modeling: Progress and Research Issues. Fort Collins, CO, GIS World Books: 155-8
Leavesley G H, Viger R J, Markstrom S L, and M S Brewer (1996b) Estimation and evaluation of spatially distributed model parameters using the Modular Modeling System (MMS). In Proceedings of the Third International Conference Integrating GIS and Environmental Modeling, Santa Fe, New Mexico, 21-25 January. Santa Barbara, CA, National Center for Geographic Information and Analysis, University of California
Lovejoy S B (1997) Watershed management for water quality protection: Are GIS and simulation THE answer? Journal of Soil and Water Conservation 52: 103
Mackay D S and L E Band (1997) Forest ecosystem processes at the watershed scale: Dynamic coupling of distributed hydrology and canopy growth. Hydrological Processes 11: 1197-1217
Mackay D S and L E Band (1998) Extraction and representation of nested catchment areas from digital elevation models in lake-dominated topography. Water Resources Research 34: 897-901
Mackay D S, Robinson V B, and L E Band (1992) Classification of higher order topographic objects on digital terrain data. Computers, Environment, and Urban Systems 16: 473-96
McBratney A B and I O A Odeh (1997) Application of fuzzy sets in soil science: Fuzzy logic, fuzzy measurements, and fuzzy decisions. Geoderma 77: 85-113
Mitas L and H Mitasova (1998) Distributed soil erosion simulation for effective erosion prevention. Water Resources Research 34: 505-16
Mitas L, Mitasova H, Brown W M, and M Astley (1996) Interacting fields approach for evolving spatial phenomena: Application to erosion simulation for optimized land use. In Proceedings of the Third International Conference Integrating GIS and Environmental Modeling, Santa Fe, New Mexico, 21-25 January. Santa Barbara, CA, National Center for Geographic Information and Analysis, University of California
Mitas L, Brown W M, and H Mitasova (1997) Role of dynamic cartography in simulations of landscape processes based on multivariate fields. Computers and Geosciences 23: 437-46
Mitasova H, Hofierka J, Zlocha M, and L R Iverson (1996) Modeling topographic potential for erosion and deposition using GIS. International Journal of Geographical Information Systems 10: 629-41
Moore I D, Lewis A, and J C Gallant (1993) Terrain attributes: Estimation methods and scale effects. In A J Jakeman, M B Beck, and M J McAleer (eds) Modeling Change in Environmental Systems. New York, NY, John Wiley and Sons: 189-214
Moore I D (1996) Hydrologic modeling and GIS. In M F Goodchild, L T Steyaert, B O Parks, C Johnston, D R Maidment, M P Crane, and S Glendinning (eds) GIS and Environmental Modeling: Progress and Research Issues. Fort Collins, CO, GIS World Books: 143-8
Naiman R J, Bisson P A, Lee R G, and M G Turner (1997) Approaches to management at the watershed scale. In K A Kohm and J A Franklin (eds) Creating a Forestry for the 21st Century: The Science of Ecosystem Management. Washington DC, Island Press: 239-53
National Research Council (1999) New Strategies for America’s Watersheds. Washington DC, National Academy Press
Newton P W, Taylor K L, Trinidad G S, Ackland R G, and D J Abel (1999) GIS networks and the marketplace. Transactions in GIS 3: 91-5
Nielsen G A, Caprio J M, McDaniel P A, Snyder R D, and C Montagne (1990) MAPS: A GIS for land resource management in Montana. Journal of Soil and Water Conservation 45: 450-3
Nofziger D L and A G Hornsby (1987) CMLS: Chemical Movement through Layered Soils Model User’s Manual. Gainesville, FL, University of Florida
O’Callaghan J F and D M Mark (1984) The extraction of drainage networks from digital elevation data. Computer Vision, Graphics and Image Processing 28: 323-44
Openshaw S (1997) The truth about Ground Truth. Transactions in GIS 2: 7-24
Orzol L L and T S McGrath (1992) Modifications to the U.S. Geological Survey Modular Finite-difference Ground-water Flow Model to Read and Write Geographic Information System Files. Portland, OR, United States Geological Survey Open File Report No 92-50
Paniconi C, Kleinfeldt S, Deckmyn J, and A Giacomelli (1999) Integrating GIS and data visualization tools for distributed hydrologic modeling. Transactions in GIS 3: 97-118
Panuska J C, Moore I D, and L A Kramer (1991) Terrain analysis: Integration into the agriculture non-point source (AGNPS) pollution model. Journal of Soil and Water Conservation 46: 59-64
Renolen A (2000) Modeling the real world: Conceptual modeling in spatiotemporal information system design. Transactions in GIS 4: 23-42
Petach M C, Wagenet R J, and S D DeGloria (1991) Regional water flow and pesticide leaching using simulations with spatially distributed data. Geoderma 48: 245-69
Peters J (1995) The HEC Hydrologic Modeling System. Davis, CA, U.S. Army Corps of Engineers, Hydrologic Engineering Center, Technical Paper No 150
Quinn P F, Beven K J, Chevallier P, and O Planchon (1991) The prediction of hillslope flow paths for distributed hydrological modeling using digital terrain models. Hydrological Processes 5: 59-79
Quinn P F, Beven K J, and R Lamb (1995) The ln(a/tanb ) index: How to calculate it and how to use it within the TOPMODEL framework. Hydrological Processes 9: 161-82
Reybold W U and G W TeSelle (1989) Soil geographic databases. Journal of Soil and Water Conservation 44: 28-9
Richardson C W and D A Wright (1984) WGEN: A Model for Generating Daily Weather Variables. Washington DC, United States Department of Agriculture, Agricultural Research Service Report No 8
Rhyne T, Bolstad M, Rheingans P, Petterson L, and W Shackleford (1993) Visualizing environmental data at the EPA. IEEE Computer Graphics and Applications 13: 34-8
Robinson V B and D S Mackay (1996) Semantic modeling for the integration of geographic information and regional hydroecological simulation management. Computers, Environment, and Urban Systems 19: 321-39
Srinivasan R and B A Engel (1991) Effect of slope prediction methods on slope and erosion estimates. Applied Engineering in Agriculture 7: 779-83
Shamsi U M (1996) Storm-water management implementation through modeling and GIS. Journal of Water Resources Planning and Management 122: 114-27
Tarboton D G (1997) A new method for the determination of flow directions and upslope areas in grid digital elevation models. Water Resources Research 33: 309-19
Twigg D R (1998) The Global Positioning System and its use for terrain mapping and monitoring. In S N Lane, K S Richards, and J H Chandler (eds) Landform Monitoring, Modelling, and Analysis. New York, NY, John Wiley and Sons: 37-62
Vieux B E (1991) Geographic information systems and non-point source water quality and quantity modeling. Hydrological Processes 5: 101-13
Vieux B E (1993) DEM aggregation and smoothing effects on surface runoff modeling. Journal of Computing in Civil Engineering 7: 310-38
Vieux B E and P B Bedient (1998) Estimation of rainfall for flood prediction from WSR-88D reflectivity: A case study, 17-18 October 1994. Weather and Forecasting 13: 407-15
Vieux B E and N S Farajalla (1996) Temporal and spatial aggregation of NEXRAD rainfall estimates on distributed hydrologic modeling. In M F Goodchild, L T Steyaert, B O Parks, C Johnston, D R Maidment, M P Crane, and S Glendinning (eds) GIS and Environmental Modeling: Progress and Research Issues. Fort Collins, CO, GIS World Books: 199-204
Vieux B E and N Gauer (1994) Finite-element modeling of storm water runoff using the GRASS GIS. Microcomputers in Civil Engineering 9: 263-70
Vieux B E and S Needham (1993) Nonpoint-pollution model sensitivity to grid-cell size. Journal of Water Resources Planning and Management 119: 141-57
Vieux B E, Farajalla N S, and N Gaur (1996) Integrated GIS and distributed storm water runoff modeling. In M F Goodchild, L T Steyaert, B O Parks, C Johnston, D R Maidment, M P Crane, and S Glendinning (eds) GIS and Environmental Modeling: Progress and Research Issues. Fort Collins, CO, GIS World Books: 199-205
Vincent V and R G Thomas (1960) An Agroecological Survey of Southern Rhodesia: Part I, Agroecological Survey. Salisbury, Government Printer
Wagenet R J and J L Hutson (1989) LEACHM (Leaching Estimation And CHemistry Model): A Process Based Model of Water and Solute Movement, Transformations, Plant Uptake, and Chemical Reactions in the Unsaturated Zone. Itacha, NY, Water Resources Institute, Cornell University
Wagenet R J, Hutson J L, and J W Biggar (1993) Simulating the fate of a volatile pesticide in unsaturated soils: A case study. Journal of Environmental Quality 18: 78-87
Watkins D W, McKinney D C, and D R Maidment (1996) Use of geographic information systems in ground-water flow modeling. Journal of Water Resources Planning and Management 122: 88-96
Weih R C and J L Smith (1997) The influence of cell slope computation algorithms on a common forest management decision. In M J Kraak and M Molenaar (eds) Advances in GIS Research II: Proceedings of the Seventh International Symposium on Spatial Data Handling. London, Taylor and Francis: 857-75
Wesseling C G, Karssenberg D J, Burrough P A, and W P A van Deursen (1996) Integrating dynamic environmental models in GIS: The development of a dynamic modeling language. Transactions in GIS 1: 40-8
Wilkerson G G, Jones J W, Boote K J, Ingram K T, and J W Mishoe (1983) Modeling soybean growth for management. Transactions of the American Society of Agricultural Engineers 26: 63-73
Wilkinson G G (1996) A review of current issues in the integration of GIS and remote sensing. International Journal of Geographical Information Systems 10: 85-101
Wilson J P (1999a) Current and future trends in the development of integrated methodologies for assessing non-point source pollutants. In D L Corwin, K Loague, and T W Ellsworth (eds) Assessment of Non-Point Source Pollution in the Vadose Zone. Washington DC, American Geophysical Union: 343-61
Wilson J P (1999b) Local, national, and global applications of GIS in agriculture. In P A Longley, M F Goodchild, D J Maguire, and D W Rhind (eds) Geographical Information Systems: Principles, Techniques, Management, and Applications. New York, NY, John Wiley and Sons: 981-98Wilson J P and P A Burrough (1999) GIS, fuzzy classification, geostatistics, and dynamic modeling: New sneakers for a new geography? Annals of the Association of American Geographers 89: 736-46
Wilson J P and M S Lorang (1999) Spatial models of soil erosion and GIS. In M Wegener and A S Fotheringham (eds) GIS and Spatial Models: New Potential for New Models? London, Taylor & Francis: 83-108
Wilson J P, Inskeep W P, Rubright P R, Cooksey D, Jacobsen J S, and R D Snyder (1993) Coupling geographic information systems and models for weed control and groundwater protection. Weed Technology 6: 255-64
Wilson J P, Inskeep W P, Wraith J M, and R D Snyder (1996) GIS-based solute transport modeling applications: Scale effects of soil and climate data input. Journal of Environmental Quality 25: 445-53
Wilson J P, Spangrud D S, Nielsen G A, Jacobsen J S, and D A Tyler (1999) Effects of sampling intensity and pattern on GPS-derived contour maps and estimated soil terrain attributes. Soil Science Society of America Journal 62: 1410-7
Wolock D M and G J McCabe (1995) Comparison of single and multiple flow direction algorithms for computing topographic parameters in TOPMODEL. Water Resources Research 31: 1315-24
Wolock D M and C V Price (1994) Effects of digital elevation model and map scale and data resolution on a topography-based watershed model. Water Resources Research 30: 3041-52
Yuan M (1999) Use of a three-domain representation to enhance GIS support for complex spatiotemporal queries. Transactions in GIS 4: 137-60
Zhang W and D R Montgomery (1994) Digital elevation model grid size, landscape representation, and hydrologic simulations. Water Resources Research 30: 1019-28
Copyright © 2000, Urban and Regional Information Systems Association
Comments/Questions: info@urisa.org