Soil Salt Content Estimation Through Integrated Analysis of ZiYuan1 Hyperspectral and Sentinel‐1 Synthetic Aperture Radar Images

Soil salinization adversely affects soil health and poses a threat to crop growth. Accurately estimating soil salt content (SSC) is of critical importance for achieving sustainable agricultural development. Hyperspectral remote sensing provides abundant spectral information that reflects the spectral reflectance characteristics of soil salinity, while synthetic aperture radar (SAR) data offer backscatter coefficient features associated with soil salinity. This study presents a method for estimating SSC using a combination of ZiYuan1 (ZY1) hyperspectral and Sentinel‐1 SAR data. This approach determines the spectral bands characteristic of soil salinity and extracts them using continuum removal. On this basis, fractional order derivative is applied to enhance spectral features (soil salinity spectral bands and spectral indices), which are then combined with SAR features (backscatter coefficients and radar indices) to estimate SSC using the extremely randomized trees algorithm. Validation of the proposed method was carried out using 84 soil samples and satellite images collected from Zhaoyuan County, Heilongjiang Province, China. Analysis of the results suggests that extracting soil salinity spectral bands reduces data redundancy, enhances the mechanism of the estimation process, and improves estimation accuracy. Compared to using the full spectral range (400–2400 nm), the proposed method increased the coefficient of determination (R2) from 0.56 to 0.69 and the residual predictive deviation (RPD) from 1.51 to 1.80. The incorporation of spectral indices further increased R2 to 0.74, and the combination of spectral features with SAR features improved the estimation accuracy even further, with R2 and RPD increasing to 0.84 and 2.57, respectively. In this work, we develop a novel approach for SSC estimation by combining hyperspectral and SAR data for soil salinization monitoring and evaluation.