With the acceleration of the industrialization process, land resources are facing the serious problem of salt alkali degradation. This study chose Chongming Dongtan in Shanghai as the research object. We derived band reflectance, the salt index SI, SI1 (Salinity Index), NDVI (Normalized Difference Vegetation Index), NDSI (Normalized Differential Salinity Index), and CRSI (Canopy Response Salinity Index) based on Landsat remote sensing data. Based on these remote sensing indicators and the sampling data, we used the Multivariate Adaptive Regression Splines model model (MARS) and Partial Least Squares Regression (PLSR) approach to establish the regression model of soil salinity and study the spatial pattern of regional saline alkali soil distribution.
Results/Conclusions
(1) The soil salinity had obvious absorption in the infrared band, and it showed high correlation with the fourth, fifth band reflectance and NDSI; 2) MARS model had better performance in monitoring soil salinity than PLSR (R2 = 0.74 and 0.63, respectively); 3) There was high spatial heterogeneity in Chongming Dongtan area. The coastal area showed high risk of saltwater intrusion. This paper aims to provide a fashion for the regional monitoring of soil salinization in coastal areas, to prevent the further saline deterioration of coastal soil and protect the ecological construction of the islands.