Quantitative Inversion of Salinity Based on Hyper-spectral Data in Songliao Plain

LU Chun;Liaoning Provincial College of Communications;

resources and environment satellite;;saline-alkali soil;;atmospheric correction;;quantitative inversion;;partial least squares regression

HSI sensor of HJ-1A is the first hyper-spectral image in China. It has high time resolution (96 hours) and fine spectrum (5 nm), which will be widely used in resources, environment, disaster monitoring, etc. However, as a new sensor, application of HSI data is still in the exploratory stage in China. To get surface reflectance of HSI images, atmospheric correction model of FLAASH was used to correct the HSI images. The reflectance after a variety of mathematical transformation was carried on partial least squares regression analysis with the measured values of soil salinity to retrieve the salinity in Songliao plain using the analysis method of GIS. The results show that, reflectance of HSI hyper-spectral images after the countdown and the first-order differential transform can significantly improve the correlation coefficient with salinity of saline-alkali soil. The correlation coefficient was 0.818 and 0.851. The Root-Mean-Square Error was 0.770 and 0.694. HSI images were used as new data source in quantitative inversion of soil salinity, which has provided significant reference for getting salinity precisely, quantitatively, quickly and salinization control in Songliao plain.