Author(s)

J. Earls & B. Dixon

Abstract

Resolution is a sensitive issue in environmental modeling and geo-spatial analysis. This study utilizes the SWAT (Soil and Water Assessment Tool) model integrated with ArcView. The overall goal of this research was to determine how sensitive the SWAT model was to the resolution of input data. For example, input data layers that originated at 250m, 125m, 30m and 3m were resampled to 30m, 90m, 120m and 240m. Data was collected for the Alafia River watershed in the Tampa Bay Estuary in West Central Florida. Initial input layers to SWAT were: Digital Elevation Models (DEMs), soils and landuse; all analyzed for sensitivity at four different resolutions; viz. 30m, 90m, 120m and 240m. Landuse data photo-interpreted from 1:12,000 color-infrared digital ortho-quarter quadrangles (DOQQs) were obtained from the Southwest Florida Water Management District (SWFWMD). GIRAS landuse data was obtained from the EPA BASINS website at 125m resolution. Soil data at 30m resolution was obtained from Soil Survey Geographic Database (SSURGO) and 250m resolution was obtained from State Soil Survey Database (STATSGO). DEMs were obtained from the EPA BASINS website at 90m resolution and from United States Geological Survey (USGS) at 30m resolution. This study makes no assumption that original resolution was in any way improved by resampling to higher resolution. Output variables tested were: model flow out, sediment out, total phosphorous (TP) out and NO3 - N out. From analysis of these outputs, it is evident that model outputs were, in fact, sensitive to the variation in input resolutions, some outputs more than others. Results indicate coarser landuse, soils and DEMs all tend to under-predict the higher resolution soils, DEMs and landuse at annual temporal resolution. Keywords: remote sensing, SWAT, GIS, soil, streamflow, sensitivity, resolution, landuse.

Keywords

remote sensing, SWAT, GIS, soil, streamflow, sensitivity, resolution, landuse.