There is a growing interest in investigating the accuracy of digital elevation model (DEM). However people usually have an unbalanced view on DEM errors. They emphasize DEM sampling errors, but ignore the impact of DEM resolution and terrain roughness on the accuracy of terrain representation. This research puts forward the concept of DEM terrain representation error (Et) and then investigates the generation, factors, measurement and simulation of DEM terrain representation errors. A multi-resolution and multi-relief comparative approach is used as the major methodology in this research. The experiment reveals a quantitative relationship between the error and the variation of resolution and terrain roughness at a global level. Root mean square error (RMS Et) is regressed against surface profile curvature (V) and DEM resolution (R) at 10 resolution levels. It is found that the RMS Et may be expressed as RMS Et = (0.0061 ( V+ 0.0052) ( R - 0.022 ( V + 0.2415. This result may be very useful in forecasting DEM accuracy, as well as in determining the DEM resolution related to the accuracy requirement of particular application.
展开▼
