2015 GSA Annual Meeting in Baltimore, Maryland, USA (1-4 November 2015)

Paper No. 207-2
Presentation Time: 9:00 AM-6:30 PM

COMPARING THE EFFECT OF DEM RESOLUTION ON THE ACCURACY OF DIGITAL TERRAIN ANALYSIS-BASED SEDIMENT EROSION PREDICTION


VOSS, Josh, Southeastern Minnesota Water Resources Center, Geoscience Department, Winona State University, PO Box 5838, Winona, MN 55987 and DOGWILER, Toby, Geography, Geology, and Planning Department, Missouri State University, 901 S. National Ave, Springfield, MO 65897, jcv15@live.missouristate.edu

Digital Terrain Analysis (DTA) is a technique that uses Digital Elevation Models (DEMs) to predict landscape processes, such as erosion. Using DTA can help minimize the amount of time and money spent completing fieldwork. One problem with DTA is choosing an appropriate DEM resolution that will minimize computer processing overhead and still provide accurate predictions of erosion. This study compares this balance with the goal of identifying the ideal DEM resolution for predicting erosion using DTA based on DEMs. We started with a 1-meter resolution DEM of the Root River watershed that was aggregated to a 3-meter, 10-meter, 15-meter and 30-meter resolution DEMs in ArcGIS. Using different spatial analyst extensions, the Stream Power Index (SPI) was calculated at each resolution. Other studies have shown that areas of the landscape with high SPI values are likely to experience erosion. Fieldwork was conducted to verify the SPI predictions. Based on the field verification a threshold percentile was determined above which the SPI was a statistically good predictor of erosion. Our analysis showed that the finer resolution DEMs could accurately predict small scale erosion, but as the resolution coarsened there was an increase in the number of sites with low SPI scores where erosion was observed during field verification. Thus, at coarser resolutions the SPI metric under-predicts which sites are likely to erode. We attribute this under-prediction to the scale of the erosional features observed in the field (decimeter-scale), which are not adequately resolved by lower-resolution DEMs. We concluded that a 3-meter DEM is the ideal resolution.