Submitted to: Transactions of the ASAE
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 10/5/2004
Publication Date: 2/1/2005
Citation: Cochrane, T., Flanagan, D.C. 2005. Effect of DEM resolutions in the runoff and soil loss predictions of the wepp watershed model. Transactions of the ASAE. 48(1):109-120.
Interpretive Summary: This paper explains the results of tests of the effects of using different quality of elevation data on predictions of runoff and soil loss with the Water Erosion Prediction Project (WEPP) model. In order to make quick estimates of the amount of soil erosion that may occur in a small watershed or on a farm field, conservation specialists use erosion prediction computer simulation models. One of the most important inputs to these models is the description of the field's topography - the land area, shape of the land and the slopes. Newer models use digital elevation data, that is basically the exact location of a piece of land (x and y coordinates) and height of the land (z coordinate) in a computer data file. Here, we studied the effect of having better detailed x, y, and z coordinates on predictions of runoff and soil loss, and how those predictions compared to measured runoff and sediment loss at 6 research watersheds. We expected to find that model predictions would be better when using the most detailed (finest resolution) data, but that was not always the case. In fact, there was no clear trend in the quality of the erosion predictions across the range of data and watersheds studied. This means that it may not be necessary to go to the trouble and expense of obtaining very detailed elevation data, particularly if someone is only interested in estimating runoff and sediment yield at a watershed outlet. For more detailed predictions along slope profiles and/or for smaller storm runoff events, it may be necessary to obtain more detailed elevation data. Overall results did indicate that it was more probable to have better predictions of soil loss from large storm events using high quality data compared to predictions of smaller events using lower quality elevation data. This work impacts scientists, field conservation personnel, and others who may utilize digital elevation data to create slope inputs to an erosion model. The ability to obtain acceptable model simulation results with less than the best data mean that erosion estimates may be able to be made without spending large amounts of time and money to develop more detailed Digital Elevation Models.
Technical Abstract: Erosion prediction utilizing Digital Elevation Models (DEMs) is a logical advancement for automating the simulation process for models such as the Water Erosion Prediction Project (WEPP). The effects of using different DEM resolutions on watershed simulations and the ability to accurately predict sediment yield and runoff from different event sizes were studied using three application methods and data from six research watersheds. Simulating watersheds with a range of resolutions can address the problem of having less than ideal topographic data for model simulations. The three application methods studied here were: 1.) Hillslope - Chanleng, 2.) Hillslope - Calcleng, and 3.) Flowpath. The Hillslope-Chanleng method uses a representative slope profile for watershed sub-basins derived from all water flowpaths determined with DEM data, and then sets the length of the profile by dividing the hillslope area by the adjacent channel length. The Hillslope-Calcleng method creates similar representative slope profiles, except the length is set with a weighting procedure using flowpath lengths and drainage areas. The Flowpath method utilizes all of the individual flowpaths as model input slope profiles, and then uses the output to provide very detailed sediment delivery to channels and spatial soil loss information on the watershed sub-basins. Statistical analyses for all methods, resolutions, and event sizes were performed by comparing predicted vs. measured runoff and sediment yield from the watershed outlets on an event-by-event basis. Analyses of runoff and sediment loss from hillslopes within the watersheds were also conducted as well as soil loss along the hillslope profiles. Results from the analyses of these watershed outlets suggest that, for a given range of DEM resolutions with valid input data, the two Hillslope methods can be used to facilitate the application of WEPP for simple watersheds and that the predictions are comparable to measured data. Interactions between resolution and erosion event size were also evident and showed coarser resolutions can be used to predict larger events, but finer resolutions may be needed to predict smaller events. The results from the application of these methods with different resolution DEMs also help us further understand important factors in watershed erosion modeling and point out limitations in our current model concepts.