A study of aster GDEM denoising based on hydrological analysis

In Aster GDEM (Version 1), there exist some exceptional values which are much lower or higher than surrounding elevations in the area of high mountains and valleys. Due to the characters of noise in Aster GDEM, such as local and discrete form, irregular distribution and different noise intensity, this paper proposes a denoising method based on hydrological analysis. The shape and location of noises are extracted by using depress filling and peak clipping in hydrological analysis firstly. Then for the noise part, a new DEM is generated through interpolating contours. The experiment in North Shaanxi of China shows that this denoising method can remove most noises effectively without the premise of reducing DEM accuracy. Meanwhile the DEM denoised can maintain topographic characteristics very well.