An improvement of offset tracking for cross hair (CH) and patch like (PL) elimination and reliability estimation for large deformation monitoring with SAR data

Abstract

SAR based offset tracking (OT) is an efficient tool for ground deformation observation, and signal to noise ratio (SNR) is its common error indicator. However, ground feature variations often weaken the accuracy of OT. In addition, SNR shows the signal reliability instead of result accuracy. Based on amplitude selection, cubic spline interpolation and double offset detecting, an improved OT method has been proposed in this paper. Subsidence caused by mining and GPS data have been employed to evaluate the performance of this scheme with TerraSAR-X data. The results indicate that patch like (PL) and cross hair(CH) are reduced efficiently, and the error estimated by the proposed method has a higher correlation with real error than SNR in the mountainous area.