Development of Agricultural Field DEM Using Repeated GPS Measurements from Field Operations: Effects of Sampling Intensity and Pattern
Widespread use of GPS systems in agricultural vehicle allows farmers to collect elevation data repeatedly to develop field-level DEM. Accuracy of these DEMs can be improved by understanding how errors are introduced from sampling procedures. In this research, a 120 m by120 m test field was modeled using a 10-m-grid USGS DEM of Winneshiek County, Iowa. Multiple sets of vehicle-based GPS elevation measurements from four filed operations (tillage, planting, spraying and harvesting) with different swath width and speed level were simulated using inverse distance weighting (IDW) interpolation from the test field. GPS errors were modeled using Gauss-Markov process and added to the simulated measurements. Then DEMs were created using a method proposed by Aziz et al. (2005). Results show that RMSE gradually decreased as the number of measurement sets used increased and leveled out after approximately 12 measurement sets unless an increase in input resolution of the elevation data were introduced to improve the RMSE of the resulted DEM. For the widest swath width, as the speed level (distance between data points along a track) decreased, the RMSE decreased from 0.23 m to 0.16 m. Track patterns on the other hand, had significant effects on the topographic maps if very small grid size is used to generate the DEM.