The fluctuations of atmospheric water vapor content in space and time, which will lead to an uncertainty propagation delay of the radar signal, is one of the main factors affecting the accuracy of InSAR. As a new sounding instrument, the foundation GPS can provide in real time continuous, all-weather, high precision precipitation water vapor values, which can be used for atmospheric correction of InSAR, nevertheless, it needs interpolation in space for using GPS water vapor observations to correct atmospheric delay in InSAR. Inappropriate interpolation methods will lead to distortion of the spatial distribution of water vapor. In conventional interpolation methods, such as inverse distance weighted (IDW) interpolation and Kriging, due to the failure to take account of the influence of topography on atmospheric delay, the interpolation results are not ideal. Therefore, this paper proposes an improved Kriging (IKriging) method, which can take into account the impact of elevation and distance on the atmospheric interpolation. The experimental results in Hong Kong show that IKriging has obvious advantages over the Kriging method. This method can be further promoted to establish the multi-factors fitting interpolation model.