GMTSAR SBAS Sensitivity for Landslide Mapping in a Mountain Test Site

Interferometric synthetic aperture radar (InSAR) is widely used to measure surface deformation associated with processes such as slow-moving hillslope instability. Although InSAR time series can reach millimetre-level precision under favourable conditions, the practical reliability of deformation maps is often difficult to assess without calibration and validation using in situ displacement measurements, which are rarely available in the remote mountainous settings where landslides commonly occur. This limitation means that key processing choices in operational workflows are frequently set based on user preference and computational constraints, with limited quantitative insight into how they influence the final deformation products and the resulting interpretation. Here we evaluate the sensitivity of Sentinel-1 time-series deformation results produced with the GMTSAR workflow for a study area in northern Pakistan where slow-moving landslides have been reported in the literature. We systematically vary controlling parameters including temporal and perpendicular baseline thresholds, multilooking factors, reference area selection, coherence thresholds, and the length of the time stack. For each configuration, we apply an identical post-processing procedure to detect hillslope deformation anomalies and delineate candidate slow-moving landslides, enabling a consistent comparison of the resulting inventories. We show that these processing choices can substantially affect the mapped landslide population and inferred spatial extent, while also changing the processing effort required to reach a stable solution. The outcomes provide practical guidance for selecting InSAR processing parameters for landslide mapping in data-sparse regions where ground calibration is not feasible.