Abstract: The traditional landslide depth detection methods only determine the depth in some sparse points,and cannot accurately reflect the depth distribution of the whole landslide. The cost is high. In this paper,we develop a landslide depth inversion method based on the ascending and descending InSAR deformation field. And the main process includes: first,the ascending and descending InSAR deformation fields are extracted based on the time series InSAR technique. Second,according to the spatial geometric relationship between the satellite line-of-sight and the landslide,the two-dimensional deformation fields along the slope and normal directions of the landslide surface are calculated. Third,a landslide depth inversion model is constructed based on two-dimensional slope deformation under the constraint of mass conservation criterion. Finally,we estimate the landslide depth based on the developed landslide depth inversion model. We conduct a case study of an ancient landslide in Taoping Village,Li County in Sichuan Province. The InSAR results show that the moving area of the landslide is about 2.9 km². When the rheological parameter is set as 0.7,the estimated landslide depth is 9~33m,and the landslide volume is about 3.49×10⁷ m³. The inversion result is consistent with the result of previous study,which proves the reliability of the theory and method developed in this paper. The proposed method can obtain continuous depth over the landslide,and can more directly reflect the risk area of the landslide. It can provide data support for the landslide disaster impact analysis and prevention.