A Model For The Retrieval Of Soil Electrical Conductivity In Ugan-kuqa Delta Oasis

Soil salinization is one of the soil degradation form in arid and semi-arid areas and it always occurs in high evaporotion , low precipitation and high ground water table area. Soil salinization affect directly or indrectly human life, agricultural productivity and development. For management purposes, quentifying both the extent and distribution of salinization is extrimly important.But with development of quantitative Remote Sensing monitoring, requirement for the quantitative extraction of salinized information is getting stricted and pixel scale data did not satisfy our demand.Domestic and overseas study shows that, in soil salinization monitoring study, using single salinized soil spectrum can not reflect salinization informatiom correctly. Therefore, modifying or advancing the image identifying algorithm is new desire for the extraction of regional soil salinization.This study starting with purification of mixed pixels to reduce the uncertainty of information as a result of mixed pixels widely contained in the Remote Sensing image because of the complexity of spectrum taken from the field and limit spatial resolution of sensor .Therefore, we used Spectral Angle Mapper (SAM) as a mapping method to generated soil electrical conductivity map and calculated Normalized Diffrence Vegetation Index(NDVI) togather with extracted the Soil Moisture Content (SM) through model of soil moisture monitoring by remote sensing (PDI).At last, we applied nonlinear regression model to identifying the relations among Electrical Conductivity(EC) ,Normalized Diffrence Vegetation Index(NDVI) and Soil Moisture Content(SM) .Result shows that, we used SAM to obtain soil salinization (EC) map and integrated with field data which consist of measurement of electrical conductivity (EC) are obtained by the combination of geophysical methods to validate the extended regional data. Accuracy is 0.7584, 0. 7388 respectivly and can preferably express the salinization degree.Correlation coefficient between dependent and indepentdent variables is 0.78, corresponding value is 0.001in nonlinear regression model that we used. Result refers that the function we gained is distinct on level0.05. At the same time ,we used F inspection method to inspect the distinctness of regression model, F=15.384 explains that this model has been collaborated very well.