A Comparison of Manual and Computer-Assisted Drainage Delineation Methods for Hydrologic-Unit Map Development
暂无分享,去创建一个
*Gregory P. Stanton, U.S. Geological Survey, 401 Hardin Road, Little Rock, Arkansas 72211 USA. gstanton@usgs.gov ABSTRACT Drainage basin or hydrologic-unit maps are necessary components of many natural-resource studies such as flood assessments, water-quality sampling, water-use reporting, watershed protection, conservation planning, and resource management. Watersheds are identified by 11-digit codes ranging in size from 40,000 to 250,000 acres (about 60 to 400 square miles). Most States are working to further delineate watersheds into 14-digit subwatersheds, which typically are 10,000 to 40,000 acres (about 16 to 60 square miles) in size. Historically, watershed delineation has been accomplished by manually marking drainage divides on 1:24,000-scale topographic quadrangles; however, this process is very time-consuming. Through the advancement of geographical information system (GIS) technologies, computer-generated maps such as digital raster graphic (DRG) images and digital elevation models (DEM) of topographic quadrangles have been made available for most of the United States. Using computer-assisted methods, the U.S. Geological Survey (USGS) has successfully delineated 14-digit-level subwatersheds in the Illinois River subbasin in northwestern Arkansas using DRGs and DEMs at two spatial scales. The resulting automated delineations were then compared to manual delineations from 1:24,000scale topographic quadrangles. The computer-assisted methods were applied to two separate elevation data sets: one comprising an elevation grid derived from 1:100,000scale USGS digital line graph (DLG) hypsography, and the other consisting of mosaicked USGS 1:24,000-scale level-2 DEMs. The computer-assisted method using the mosaicked 1:24,000-scale level-2 DEMs produced satisfactory results whereas the method using 1:100,000-scale DLG elevation data did not. The computer-assisted 14-digit subwatershed delineations based on 1:24,000-scale level2 DEMs were visually and statistically compared to manual delineations of the same subwatersheds. The computer-assisted delineation compared very well to the manual delineations, generally following drainage divides; however, some computer-assisted subwatershed boundaries required editing in small, low relief areas such as stream confluence floodplains. Statistically, the absolute value percent difference of the computerassisted and manually derived subwatershed areas averaged about 1.5 percent and ranged from 0.06 to 5.98 percent. The “common area” of the manual-delineated subwatershed that was included in the computerdelineated subwatershed averaged about 97.2 percent and ranged from about 90 to 99.75 percent. The computer-delineated area that extended beyond the manual drainage divide or “overestimated” the subwatersheds averaged about 2.3 percent and ranged from 0.02 to 6 percent. Labor costs (including data acquisition, preprocessing and editing) were reduced by about 30 percent by using the DEM computer-assisted delineating method. Additional labor savings are possible as available tools and data are enhanced. The computerassisted delineation method has been used by the USGS on adjacent watersheds in Arkansas.