DEM Aggregation and Smoothing Effects on Surface Runoff Modeling

A digital elevation model (DEM) may be used in modeling direct surface runoff. Topographic variability affects the apparent slope and flow path length extracted from DEMS. Smoothing of the digital elevation data is a common procedure for reducing or eliminating pits prior to watershed delineation using a DEM. Error is propagated in simulations of direct surface runoff if the apparent slope is flattened or flow path length is shortened due to smoothing or aggregation of the DEM. Model error due to cell‐size aggregation from 30‐ to 90‐, 150‐, and 210‐m resolution and due to smoothing using a $3\times 3$, $5\times 5$, and $7\times 7$ window is investigated. The magnitude of error propagation is measured using a finite element solution of direct surface runoff. The log of the hydrograph error scales linearly with the log of the relative entropy loss. Further, low rainfall intensities proportionally produce larger errors than higher intensities. The method of error analysis provides an a priori means of assessing the magnitude of the consequent error due to aggregation and smoothing.