Multi-Sensor Data Fusion for Enhanced Persistent Scatterer Analysis: Improving Ground Deformation Monitoring

The growing availability of multi-sensor Persistent Scatterer (PS) data offers high-precision, multi-temporal measurements for monitoring ground displacements across large areas. The ability to detect ground deformation phenomena is largely determined by PS density, influenced by the sensor resolution and site-specific characteristics. This research proposes a robust data fusion method to integrate multi-band/ multi-sensor PS products, prioritizing spatial dimensions to enhance coverage and exploit complementary information from diverse sensors. The method employs a weighted least squares approach with an adaptive weighting scheme to generate synthetic measurement points that synthesize deformations along the East-West and Up-Down components, effectively addressing challenges posed by sparse and unevenly distributed data. The methodology, validated through Sentinel-1 (S1) and COSMO-SkyMed (CSK) data integration, employs quantitative metrics and field observations to assess the reliability of fusion predictions. Statistical analysis demonstrates superior performance, with R-squared values of 0.950 for Up-Down and 0.868 for East-West components. Spatial coverage expands from approximately 10 $\%$ (S1) and 22 $\%$ (CSK) to 67 $\%$ of the total area of interest. The enhanced detection capabilities enable comprehensive monitoring of ground deformation processes, uncovering patterns otherwise undetectable through single-sensor analysis and providing crucial information for hazard assessment.