This article mainly study how to solve the common no data or data abnormality problem in ASTER GDEM .At first, detecting the section which no data or data abnormality in ASTER GDEM using slope analysis together with density divide method .Then using SRTM DEM according to regression fusing combine with interpolation to simulate and mend the ASTER GDEM data .At Last, Evaluating the results through using the mended ASTER GDEM data to set up 3D model and Hydrological Model . The study results indicate that regression fusing and interpolation method can produce a less error margin and advanced precision data .The mended data is primly matched with Remote Sensing data and the Hydrological Model is reasonable. Above all, this method mended no data and abnormality section in ASTER GDEM improving the data integrality and applicability.