Monitoring Soil Salinity Using Machine Learning and the Polarimetric Scattering Features of PALSAR-2 Data

Soil salinization is one of the major problems affecting arid regions, restricting sustainable development agriculture and ecological protection in Keriya Oasis Xinjiang, China. This study aims to capture distribution soil salinity with polarimetric parameters various classification methods based on Advanced Land Observing Satellite-2(ALOS-2) Phased Array Type L-Band Synthetic Aperture Radar-2 (PALSAR-2) Landsat-8 OLI (OLI) images Oasis. Eleven polarization target decomposition were employed extract scattering features. Furthermore, features highest signal-to-noise ratio value used combined optimal components form a comprehensive dataset named + PALSAR2. Next, two machine learning algorithms, Support Vector Machine (SVM) Random Forest, applied classify surface characteristics. The results showed that better outcomes achieved SVM classifier for PALSAR2 data, overall accuracy, Kappa coefficient, F1 scores being 91.57%, 0.89, 0.94, respectively. indicate potential using PALSAR-2 data coupled monitor different degrees