Integration of InSAR coherence and SAR backscatter increases accuracy of LULC mapping in tropical high-mountain ecosystems

Abstract

Páramos, neotropical high-mountain biomes of the northern Andes, are increasingly threatened by human land use and climate change. Accurate land use/land cover (LULC) mapping is critical for monitoring these pressures. However, persistent cloud cover poses substantial challenges for optical satellite imagery. This study leverages multitemporal Interferometry SAR (InSAR) coherence and dual polarimetric backscatter intensity to improve LULC mapping accuracy in páramos ecosystems. Using a 1-year Sentinel-1 time series from January to December 2019, 17 LULC classes were mapped across 17,000 hectares. Results demonstrate that integrating backscatter and coherence increased F1-Score and overall accuracy by 9% and 11%, respectively, compared to backscatter alone. This emphasizes the critical role of multitemporal SAR for LULC mapping in cloud-prone regions. Additionally, this study underscores the utility of cloud-based platforms for efficient SAR data processing, offering a scalable approach for mapping high-mountain ecosystems worldwide, especially in inaccessible and heavy precipitation regions.