Documentation and analysis of a geographic information system application for combining data layers, using nonpoint-source pollution as an example

Abstract

A geographical information system application has been developed that allows scientists to combine multiple data layers into a single data layer. This application provides an effective tool for identifying areas where the potential effect of the combination of data layers may be greater than any single data layer. Such a tool is useful in studying an activity that cannot be measured directly. Scientists wanting to identify areas where an activity may have the greatest effect can identify factors that directly or indirectly reflect the effect of the activity being studied. When combined, these factors would identify areas where the potential for the activity to have an effect are greatest. The data layers used to develop the single data layer determine the activity addressed by the application?the application was developed to identify areas where the potential for nonpoint-source pollution to affect areas of Indiana is greater relative to other areas in Indiana. To evaluate the potential in other states or areas, data layers for those states or areas would be used. To address a different activity, even activities not related to water resources, data layers that directly or indirectly reflect the effects of the activity being studied would be used in the application. The application was developed using Environmental System Research Institute's ArcView geographical information system and the ArcView extension, Spatial Analyst. To use the application, the user selects data layers related to the activity being studied, describes the variability of a data element between geographic areas on each data layer, assigns a relative importance factor to each data layer, and focuses the output by identifying a watershed of interest. The application then assigns a rank value to the features within each data layer on the basis of the spatial variation of the data layer. The data layers are converted to grids. The values of the grid cells are the product of the rank values and the importance factors. The cells for each data layer are combined to form a single grid. The combination is summarized to describe the spatial variation for the watershed of interest. The combined data layer does not show the actual potential effect of the selected factors but rather the relative difference in the potential effect among areas when all data layers are considered. An analysis of the application indicates that the selected data layers to be combined should be at the greatest spatial resolution possible; however, all data layers do not have to be at the same spatial resolution. The spatial variation of the data layers should be adequately defined. The size of each grid cell should be small enough to maintain the spatial definition of smaller features within the data layers. The most accurate results are shown to occur when the values for the grid cells representing the individual data layers are summed and the mean of the summed grid-cell values is used to describe the watershed of interest.