Integrating ICOPS Time-Series InSAR Measurement with the Convolutional Neural Network (CNN) and Optimized Hot Spot Analysis (OHSA) to Monitor Land Subsidence in Pekalongan, Indonesia

The condition of land subsidence in Pekalongan has worsened the area that prone to coastal inundation during high tide. Monitoring land subsidence in Pekalongan becomes vital to mitigate the other possible land subsidence occurrence area and the possible hazard caused by land subsidence. This study used Synthetic Aperture Radar (SAR) datasets from the Sentinel-1 radar satellite between 2017 and 2020 and processed using Improved Combined Scatterers Interferometry with Optimized Point Scatterers (ICOPS) with the integration of convolutional neural network (CNN) as the optimization algorithm and Optimized Hot Spot Analysis (OHSA) as the statistical clusterization method to identify significance measurement point between each data. The comparison of the time-series Interferometry SAR (InSAR) result with the GPS measurements shows a good correlation. Furthermore, this study uncovered a significant correlation between land subsidence, geological landforms, and land use within the study area.