 |
|
|
||||
| Watershed Modeling |
| Home | People | Research Topics | Publications | Software | Training | Related Links | FAQs |
|
A Modular Approach to Integrating Environmental Modeling and GIS
Table of ContentsKeywords: MMS, GIS Weasel, GIS, Modeling
G.H. Leavesley, R.J. Viger, S.M. Markstrom, and M.S. Brewer
Keywords: MMS, GIS Weasel, GIS, ModelingABSTRACT INTRODUCTION In addition to the issues of model creation and selection are the associated issues of study area characterization and the estimation of parameters for these models. Guidelines for characterization and parameterization are normally few and users commonly develop their own methods to analyze available data and derive initial parameter estimates. The subjective nature of many such methods makes it difficult to compare model applications among different users and to effectively transfer methods to other study areas. Geographic information systems (GIS) provide the capability to develop more objective and reproducible methods of characterization and parameter estimation. A GIS tool set, termed the GIS Weasel, has been developed to use a wide variety of methodologies in support of lumped and distributed parameter models. Parameter estimation procedures specific to MMS modules are also provided by the GIS Weasel. The linking of modeling and GIS tools provides a common framework in which to focus multidisciplinary research and operational efforts to provide a flexible modeling system for developing, analyzing, and applying selected modeling approaches for a wide variety of environmental problems. The purpose of this paper is to provide an overview of MMS, the GIS Weasel, and the integration of these technologies. MODULAR MODELING SYSTEM The GIS Weasel supports MMS as a pre-processing component. However, it can also provide GIS post-processing capabilities for the visualization and analysis of spatial and temporal model output fields.
Figure 1. A schematic diagram showing the integration of the GIS Weasel with the
Pre-process ComponentPre-process component functions include all data preparation and analysis functions needed to meet the data and selected parameterization requirements of a user-created or user-selected model. Time-series data from existing databases as well as from field instrumentation are prepared for use in selected model applications. The spatial analysis needed to derive model parameters is accomplished using the GIS Weasel.
Model ComponentThe model component is the core of the system and includes the tools to selectively link process modules from the module library together into a model and to interact with this model to perform a variety of simulation and analysis tasks. The library can contain several modules for a given process, each representing an alternative conceptualization for simulating that process. The user, through an interactive model-building interface (Xmbuild), selects and links modules to create a specific model. Once a model has been built, it may be saved for future use without repeating the MBUILD step. This capability allows `canned' models to be provided to users.
Post-process ComponentTools to analyze model results are included in the post-processing component. These include a variety of visualization and statistical tools as well as the ability to interface with resource management models and decision support systems (DSS). Optimization (Opt) and sensitivity (Sens) tools are provided to analyze model parameters and evaluate the extent to which uncertainty in model parameters affects uncertainty in simulation results. A modified version of the National Weather Service's Extended Streamflow Prediction Program (ESP)2 provides forecasting capabilities using historic or synthesized meteorological data. THE GIS WEASEL The GIS Weasel provides tools to delineate, characterize, modify and parameterize an area and its associated "model response units" (MRUs) using a digital elevation model (DEM) and ancillary digital data (Fig. 2). MRUs are defined as areas delineated within a watershed, or region of interest, that reflect a model's treatment of spatially distributed characteristics, such as slope, aspect, elevation, soils, and vegetation. An MRU may represent an area whose character or composition is assumed to be homogenous with respect to one or more attributes or it may be heterogeneous with respect to all attributes. In addition to an MRU map and estimates of model parameters for each MRU, the GIS Weasel also delineates the drainage network and computes the connectivity of MRUs with the drainage network. The location of data collection sites can be overlaid on the MRU map and the connectivity relations among MRUs, the drainage network, and the data sites also can be computed. The history of previously completed processing steps is also recorded to enable the user to reenter the computational process at any point and use an alternative computational procedure. ----------------------------------------
Delineation PhaseIn this phase, the DEM, existing in an Arc/Info GRID format, is used to derive additional topographic surfaces, including the area of application, drainage network, and preliminary MRU map. The GIS Weasel will fill depressions in the DEM in order to insure that all paths of hydrologic flow will eventually reach the edge of the DEM. Flow direction and accumulation surfaces are created from this input. The area of interest within the DEM is then delineated from either a set of automatically extracted watersheds, a user-selected point on the flow accumulation surface, or a pre-existing Arc/INFO coverage or GRID.
Next, the drainage network is extracted using a user-provided threshold value that represents the minimum upslope area to be associated with the initiation of a first-order channel3. The resulting drainage network is generated using the flow accumulation surface. The magnitude of the threshold effects the density of the drainage network by constraining the minimum area of a subwatershed. The user-defined first order drainage configuration may or may not reflect the actual configuration depending on the level of spatial aggregation desired. The interface allows the user to experiment with different thresholds until an acceptable value is found. The last step in the delineation phase of the GIS Weasel is the selection of a methodology for sub-dividing the application area into MRUs. Currently, there are seven different methods available: 1-3) three types of irregular (traditional flow-plane based) polygons; 4) a fixed-interval lattice of square MRUs; 5) non-contiguous or "pixelated" MRUs (highlighting statistical similarities rather than spatial contiguity); 6) contributing areas associated with a coverage of points (e.g.: stream gages); and 7) use of a pre-existing Arc/Info coverage or GRID. A preliminary MRU map is manufactured based on the method selected. The delineation phase is a highly automated sequence of processing steps with minimal user input. The characterization, modification, and parameterization phases, however, are more heuristic in nature and the options available are dependent on the type of delineation method selected. These three phases are not a fixed sequence of processing steps, but rather an open editing session where a user can iteratively evaluate, modify and parameterize the MRU map.
Characterization PhaseIn this phase, a set of tools allows the user to explore the nature of the MRU map and any ancillary thematic information that may be associated with the area of application. Data derived from the original elevation grid (e.g.; slope, aspect) or other user-supplied grids of attribute data (e.g.; vegetation, soils) can be examined on the basis of individual or groups of MRUs, the entire watershed, or individual grid cells using various queries and displays of attribute distribution. User-defined boolean-based logical queries can be to used to display and evaluate various combinations of data attributes. Histograms of the distribution of selected attributes can displayed.
Modification PhaseThis phase enables the user to make revisions to the MRU map based on information obtained during the characterization phase or on the basis of potential changes in the characteristics of the area of interest. An MRU map reflecting current conditions may be modified to reflect anticipated changes in land use as the result of anthropogenic or natural forces. MRUs can be grouped, divided, created or eliminated. The GIS Weasel provides version control and documentation to track modifications of MRU maps.
Parameterization PhaseThis phase generates estimates of model parameters using MRU attributes and statistical measures of these attributes. Parameter estimates can be defined as measures of central tendency or measures of variance. Some parameters related to geographic and topographic characteristics, such as area, elevation, and slope, are common among many models and can be derived from the DEM used in the delineation phase. More module-specific parameters may require the use of multiple digital data coverages and the development of specific GIS procedures to compute parameter estimates. Developed procedures for selected modules in the MMS module library can be selected and applied using menu lists of modules and parameter options. Parameter estimates can be written directly into MMS parameter files for application in the selected model INTEGRATED SYSTEMS APPLICATION A second application is the evaluation of the effects of alternative watershed and ecosystem management strategies5. Landscape-generation models provide spatial representations of alternative land-use patterns associated with alternative forest-management scenarios. Management scenarios might range from maximum timber yield to maximum habitat protection. Scenario outputs from the landscape-generation models can be input to the GIS Weasel and used to characterize and parameterize each scenario for application with a variety of watershed and ecosystem process models in MMS. The effects of these management strategies on issues such as streamflow, sediment, and selected chemical and biological processes can be evaluated. While the GIS Weasel provides the mechanisms for delineating, characterizing, and estimating parameters for watersheds and MRUs, much work remains to be done on the development, testing, and identification of the specific parameterization procedures that are most appropriate for each of the environmental processes that may be simulated using MMS. The ability to recreate sets of spatial parameters for an application area is crucial to isolating and comparing the physical response predictions of different models. In addition, the development of consistent methodologies for extracting spatial information from different application areas enables the comparison of physical responses of the different areas, as well as the extension of calibrated model settings from application areas with observed data to application areas lacking relevant observations. As these procedures are defined for selected modules, they will be linked to the module. When a model is built or selected for execution, the parameterization procedures linked with each module will be incorporated in the GIS Weasel for automated execution if the needed spatial data bases are available. Continued advances in physical and biological sciences, GIS technology, computer technology, and data resources will expand the need for a dynamic set of tools to apply these advances to a wide range of interdisciplinary research and operational environmental issues. The GIS Weasel and MMS provide a flexible framework in which to integrate and apply these tools.
REFERENCES
|
| AccessibilityFOIAPrivacyPolicies and Notices | |
 |
|
