Separating Non - Linear Deformation and Atmospheric Phase Screen ( Aps ) for Insar Time Series Analysis Using Least - Squares Collocation

We present a new method for separating ground deformation from atmospheric phase screen (APS) based on PSInSAR. By stochastic modeling of ground deformation and APS via their variance-covariance functions we can not only estimate the signals with the best accuracy but also assess the estimation accuracy using least-squares collocation [5]. We evaluate the APS estimated by our method and the APS obtained from a commonly used window-based filtering method [6] by comparing them to repeat-pass interferograms over ground surfaces outside the subsiding region of Mexico City. The comparison shows that our method results in a better estimation of APS than the filtering method which ignores the temporal variability of APS variance. Our method is desired when there are temporal gaps in a SAR time series. In such a case, the filtering method needs a large temporal window to suppress APS, which may lead to leakage from ground deformation to APS.