Towards a revised framework of modified time series InSAR for mapping land deformation

Interferometric synthetic aperture radar (InSAR) time series technique has been widely applied to map the ground deformation over the past two decades. One of the key steps of InSAR time series approach is to select appropriate measurement scatterer (MS) pixels to build the triangulated irregular network (TIN). In order to build a stable TIN, there are four criteria need to be satisfied: (1) working from whole to the part, (2) hierarchical network, step-by-step control, (3) with sufficient accuracy, and (4) with appropriate spatial density. The effectiveness of Geodesy and Earth Observing Systems-PSI (GEOS-PSI) and GEOS-Advance Time Series Analysis (GEOS-ATSA) approaches have met the first three criteria above. Nevertheless, it is noteworthy to mention that higher sparseness of the TIN may lead to unwrapping failures, while having more evenly distributed of MS offers a higher chance to eliminate such effect and include more additional pixels during the adaptive processing. In this work, we proposed a revised MS pixels selection approach to satisfy the fourth criterion, in combination with the GEOS-PSI and GEOS-ATSA processing chain: the modified GEOS-ATSA (GEOS-MATSA). As compared to the conventional MS pixels selection criterion, the proposed approach has three advantages: (1) the MS pixels are added into the main network with control through an adaptive estimation strategy, (2) isolated sub-networks have better chances of including into the main network, and (3) the computing workload can be decreased by reducing the iteration times. The case study area is selected in Beijing–Langfang region, while the ideal threshold settings for the initial reference network are discussed for both persistent scatterer and distributed scatterer pixels in this work.