Research on Optimal Analysis Window Size and Resampling Method According to DEM Analysis Scale

The scale issue is one of the most important theory problems in digital terrain analysis and geo-science analysis. The accurate extraction of DEM terrain parameters according to right DEM analysis scale and the acquisition of the experienced threshold of DEM modeling rely on the selection of the optimal analysis window size. Current research indicates that the analysis window size is hard to accurately be selected in DEM based on different analysis scales. This paper presents a new algorithm of selecting the right analysis window size in original DEM analysis scale basing on an improved local variance measurement. Using statistics characteristics of local variance fitting curves, it can be concluded that the Bilinear Interpolation method is the optimal DEM resampling model of DEM scale analysis and the Regular Resample method can not effectively construct series of DEM that contains scale characteristics the same as original DEM data. The conclusions are validated by the results of different terrain test areas.