How to select a common master image is one of the key points in multi-temporal InSAR processing. In this paper, three
main decorrelation factors, spatial baseline, temporal baseline, Doppler centroid frequency difference, are focused on to
establish a correlation model of SAR interferometry, especially, a seasonal factor is presented to reflect correlation
change in a cycle of seasons, then the model for optimum selection of common master image based on a criterion called
mean total pre-coherence was presented, followed by the corresponding computing algorithm. The experiments of master
image selecting are performed using the 22 ERS-2 SAR images of Nanjing from 1996 to 2000, and the results
demonstrate the effectiveness of the algorithm.