Improved remote sensing detection of soil salinity from a semi-arid climate in Northeast Brazil

Abstract

Remote sensing techniques are being increasingly applied to investigate soil characteristics. Here we propose an approach that allows the detection of salt-affected soils in arid and semi-arid environments. We test the procedure in Northeast Brazil through a combination of remote sensing and geochemical ground-based measurements. Spectral indices were used to characterize soil salinization features and patterns. The Linear Spectral Unmixing technique (LSU) is applied in this study to improve the prediction of soil salinity. Eighteen indices were extracted from the MODIS Terra data. A moderate correlation was found between electrical conductivity and the spectral indices. An improvement occurs in most of the correlations after applying the LSU method. To generate a predicted salinity map, a multiple linear regression, based on the best correlated indices is conducted. The standard error of the estimate is about 12.1 μS cm −1.