The Effect DEM Resolution Has on the Morphometric Measurements of Small Watersheds; A Case Study of the Evans Branch Watershed in Southeastern Michigan

Watershed delineations for hydrologic studies (i.e. BASINS) increasingly depend on digital elevation models (DEM), but few studies have quantified the influence grid cell size has on characterizing watershed variables (area, slope, and stream length). Consequently, the purpose of this study is to examine the effect DEM resolution has on morphometric and hydrologic studies of smaller sub-watersheds (<20 sq. mi), such as the Evans Branch watershed, located in Southeastern Michigan. A total of 80 DEM resolutions, ranging from 1m to 80m, are used to assess the impact resolution has on the physical characteristics of the Evans Branch watershed. The DEM is statistical analyzed using two methods, Statistical Process Control Charts (SPCC) and Revised Automated Proximity and Conformity Analysis (R-APCA). SPCC, which are commonly used in manufacturing and industry to statistically analyze production process in order to estimate deviations that may result in product rejection, set statistically significant upper and lower control limits for individual morphometric measurements (e.g. watershed area, sinuosity) as grid cell size increases. R-APCA, which is used to compare modeled output values against observed field values using statistical parameters (mean and standard deviation), quantifies the level of agreement between the modeled stream location for each DEM and the control stream. The results suggest there is an optimum resolution, defined as the coarsest resolution in which detail is not sacrificed, to conduct a hydrologic study on the Evans Branch watershed. As a result, grid cell sizes that are finer than the optimum resolution only produce an increase in data volume and processing time while coarser resolution influences calculations of watershed variables. These results can guide future hydrologic models by establishing grid cell size restrictions to generate more reliable models.