Research data

OverviewThis dataset supports the research article "Revealing the topographic-hydrological governance of soil salinity in arid oases through an explainable machine learning framework integrating SAR polarimetric decomposition and multi-source data". It provides high-resolution (10 m) processed features and ground-truth observations for soil salinity mapping in the Keriya Oasis, Xinjiang, China.Data ContentThe repository includes the following components:Ground-Truth Samples: Soil Electrical Conductivity (EC) measurements from field surveys (Excel format).Optimized Feature Set: 59 candidate features including Sentinel-1 SAR polarimetric parameters (MBDP decomposition components), Sentinel-2 MSI spectral indices (SI, NDRE2, etc.), and SRTM-derived topographic variables.MLP Model Outputs: High-resolution (10 m) soil salinity distribution maps for the Keriya Oasis.Methodology SummaryThe dataset was generated using a synergy of radar polarimetry and machine learning. Sentinel-1 data underwent Model-Based Dual-Polarization (MBDP) decomposition to extract volume scattering components. Feature selection was performed via the Boruta algorithm, and spatial predictive modeling was executed using a Multi-Layer Perceptron (MLP) architecture.Data Usage & LicensingThis data is released under a Creative Commons Attribution 4.0 International (CC BY 4.0) license. Users are free to share and adapt the material for any purpose, provided appropriate credit is given to the original authors.