Statistically robust detection and evaluation of errors in DTMs

Digital Terrain Models (DTMs) have been an important topic in the study of ground surface landform, therefore precise evaluation of errors in DTMs production is a critical factor to assess the quality of DTM. In this paper the attribute of errors in DTMs are characterized and robust statistical methods are proposed as accuracy measure. A method based on robust statistical estimation is presented to detect gross errors in DTMs. For practical example a region in Catalonia, Spain, including city areas (Terrassa) as well as forest steep mountainous terrain (La Mola) is selected to evaluate the performance of DTM generation algorithm and to analyze the significance of errors for World view-1 satellite images.