Quantifying soil erosion rates in tropical drylands: An innovative SBAS InSAR and statistical clustering approach in La Tatacoa, Colombia

Quantifying soil erosion in vulnerable tropical drylands is critical for desertification management but remains methodologically challenging due to the spatial limitations of field-based methods. This study introduces a novel framework integrating Small Baseline Subset Interferometric Synthetic Aperture Radar (SBAS-InSAR) with unsupervised statistical clustering to quantify erosion rates and validate spatial heterogeneity in the La Tatacoa dry forest, Colombia. Using 123 Sentinel-1 radar images (from 2018 to 2023), we generated high-resolution displacement velocity maps, which were then integrated with Normalized Difference Vegetation Index (NDVI) data from Sentinel-2 and geomorphometric attributes derived from the ALOS PALSAR Digital Elevation Model. Study-wide mean erosion rates ranged from 0.43 to 0.54 mm/year; however, these averages masked significant localized instability. Time-series K-means clustering of 100 randomized sampling points identified four distinct deformation regimes, statistically confirming that high-velocity zones are structural realities rather than measurement noise. Specifically, Cluster 4 identified critical hotspots where erosion rates reached 30 mm/year exceeding the United States Department of Agriculture (USDA) soil tolerance threshold (0.93 mm/year) by a factor of 32. These hotspots are systematically confined to dissected terrain with slopes > 9 ° and NDVI < 0.15 . By rigorously distinguishing genuine deformation regimes from background variability, this framework provides a scalable, evidence-based tool for prioritizing conservation interventions in data-limited dryland ecosystems.