Spectral model of soil salinity in Xinjiang of China

Soil salinity is a major environmental hazard. For mapping and monitoring the spatial distribution of it using remote sensing data, a thorough understanding of their spectral behaviour is of paramount important. This paper presents an approach to estimating soil salinity through correlation and regression analysis between the EC with R (reflectance), A(log(1/R)),and their derivatives respectively. The field spectral data is obtained by using ASD. We got 28 points in Aibi lake in Xinjiang,China, in July, 2004. The results show that A' yield the highest performance in the calibration stage. The selected wave bands are 0.45-0.51μm,2.295-2.365μm. SEC is 37.3141, R
²
is 0.48 and a of constant, band1,band2 is 0.056,0.008,0.013. On the whole, we can accept that these two bands are the sensitive bands to soil salinity, and the regression procedure of A' can give a quantitative model for estimating EC. From previous theories and experiment results of other people, soil salinity has a high reflectance in blue band and a carbonate absorption festure is also reported. Our outcome is accord with it. It should be noted that we also need more data to validate it.