1-Modeling Better Decisions

Modeling Better Decisions
Land evaluation and site assessment enhanced with GIS
By Brian D. Lee and Collin D. Linebach

Summary

An established framework for making land-use decisions was implemented in ModelBuilder to provide a more holistic approach to community planning.

Geospatial data and tools provide valuable support for the land-use decision-making process. A working example of a GIS-enhanced and raster-based version of the United States Department of Agriculture (USDA) Natural Resources Conservation Service (NRCS) Land Evaluation and Site Assessment (LESA) framework was created using ModelBuilder and publicly available data for seven central Kentucky counties.

The study region, located in the central part of Kentucky, is known as the "Horse Capital of the World."

The methods used in the model were outlined by James R. Pease and Robert F. Coughlin in Land Evaluation and Site Assessment: A Guidebook for Rating Agricultural Lands, Second Edition. Because of its relatively low computing power requirements, this model can be run on a laptop computer in the field. With a GPS-enabled computer, georeferenced imagery from the National Agricultural Imagery Program (NAIP) can be combined with data and scenario results. This capability can help a community's land planning process.

Acquiring Data

The study area for this example is approximately 1,927 square miles (or 4,991 square kilometers). Most of the input data was obtained from the Kentucky Geography Network (kygeonet.ky.gov), a geospatial data clearinghouse. Soil data for the study came from the NRCS Soil Data Mart (soildatamart.nrcs.usda.gov/). The datasets obtained also included soils polygon and tabular data for each of the seven counties, statewide digital elevation model (DEM) data, corporate boundary polygons, road centerlines, sewered areas, impervious surfaces, U.S. Census 2000 block data, National Land Cover Data, and GAP Stewardship Lands data.

Hardware Requirements

A major benefit of the approach described in this article is that different scenarios can be visualized in under 10 minutes per run using a Dell Latitude D800 laptop computer with a 1.7 GHz processor and 512 MB RAM running Microsoft Windows XP Professional, ArcGIS 9.2 (with Service Pack 5), and Spatial Analyst.

Performing Land Evaluation

Soils and DEM data was used to assess four characteristics in the land evaluation (LE) section of the ELESA model: soil agricultural importance, soil erodibility, land slope, and solar aspect. The soil agricultural importance map was converted to a GRID from the original shapefile based on the agrarian uses in the attribute table. The more valuable agricultural soils received a higher ordinal rating while less agriculturally important soils received a lower rating on a 1- to 10-point scale.

The study area, outlined in red, is shown on The Commonwealth Map, Kentucky's ArcGIS Server basemap implementation available from the Kentucky Geography Network (kygeonet.ky.gov).

The soil erodibility GRID was also derived from using the Feature to Raster tool (in the Conversion Toolbox of ArcToolbox) based on attributes. Less erodible soils were reclassified to a higher rating (10) and more erodible soils received a lower rating (1).

Using DEM data, slope (as a percent) and solar aspect GRIDs were both preprocessed using the ArcGIS Spatial Analyst extension prior to running the ELESA model. Rating and valuing operations were accomplished in the ELESA model. Slopes were valued based on steepness. Slopes of less than 3 percent were given the highest value (10 points). As slope values increased, point values decreased.

The aspect GRID depicted the direction the slope faced in degrees from north (moving in a clockwise direction). This GRID was reclassified and higher ratings were assigned to the southeast, south, and southwest slopes because these slopes are potentially more productive. Aspects facing a less southerly direction were given lower ratings.

Site Assessment

Site assessment normally considers nonsoil factors, such as development pressure and infrastructure, when rating the suitability of land for agriculture. This model used proximity to stewardship lands, imperviousness, population density, pasture/hay/crop conglomeration, proximity to urban development infrastructure, and proximity to scenic roadways as site factors. The model processed the GRID data using the geoprocessing tools in ArcGIS.

The Euclidian Distance tool in ArcGIS Spatial Analyst was employed to determine proximity to GAP stewardship lands. Then the GRID data was reclassified using values between 0 and 10 with lands closer to protected lands receiving a higher rating.

Data from the National Land Cover Dataset 2001 for the seven-county study region

The original imperviousness GRID was downloaded as a percent cover GRID with zero indicating no impervious cover and 100 indicating complete impervious cover. The GRID was reclassified based on equal intervals with higher imperviousness receiving a lower rating.

The population density was determined from the 2000 U.S. Census block data using the Feature to Raster and Reclassification tools. A 0- to 10-point GRID was created that assigned lower values to areas with higher density.

Because this model's purpose was to identify valuable agricultural land in the study area, a pasture/hay/crop conglomeration GRID was generated. Using the 2001 National Land Cover Dataset, areas of pasture, hay, and crop were identified. Large conglomerations received higher ratings, and small conglomerations received lower ratings. Areas were identified using the Region Group, Zonal Geometry, and Reclassification tools.

The proximity to urban development infrastructure GRID required the most complicated subanalysis because it involved combining multiple operations for determining the proximity of interstate exits, sewered areas, federal highways, and arterial highways. The Euclidian Distance tool was used on all four GRIDs (exits, sewers, and federal and arterial highways). GRIDs were reclassified to an appropriate 1- to 10-point value. The closer the proximity to these features, the lower the point value assigned. The GRIDs were combined to create a single 0-40 GRID that was rescaled to a 1�10 GRID.

ModelBuilder was essential for developing different scenarios quickly.

The viewshed GRID was created using designated scenic roadways with observer stations one-eighth of a mile apart and the observer level set at five feet along with the DEM. This GRID of visible and nonvisible cells was then reclassified to 10 (visible) or 0 (nonvisible). Because this model component required some time to perform the analysis, the Viewshed tool was run prior to the ELESA model.

Continued on page 2

Evaluating Parcels Systematically

Land-use decision making is an increasingly important capability for many local governments. The Land Evaluation and Site Assessment (LESA) framework, developed by the United States Department of Agriculture Natural Resources Conservation Service, has been used by hundreds of communities across the country to help make decisions, particularly in the context of farmland protection programs.

Land evaluation (LE) and site assessment (SA) are the two aspects of this approach. The LESA framework aids land-use decision makers by rating lands based on land evaluation and site assessment rating factors. The factor rating system is determined by a committee composed of local people who are familiar with the landscape and analysis objectives. A key benefit of using LESA is that it provides a clear and consistent decision tree-type framework for land rating.

LE primarily rates soils. In the agricultural example described in the accompanying article, the best soils were given a high ordinal rating (10), while less agriculturally rated soils were assigned a lower rating (1). The SA section is often more complex and can consist of three general categories: nonsoil agricultural factors, development pressures (e.g., road and sewer proximity), and other public values (e.g., historic, scenic). Like the LE portion, each SA factor described in this article was rated relative to agricultural generalization to demonstrate the functionality of this approach.