The Monitoring Of Soil Salinization On Cotton Field Based On Remote Sensing Technology In North Xinjiang Oases

【Objective】To acquire soil salinization physical and chemical characteristics informations timely,accurately, in large-scale and dynamically in cotton field is important for effective governance and controlof soil salinization. In this study, through analyzing characteristics of hyperspectrum and each band ofsatellite image of soil salinization in cotton field, we built spectrum-salt monitoring indexes and indirectmonitoring indexes, and then set up models of monitoring indexes and soil salt content for realizingdynamic monitoring of cotton field soil salinization by using remote sensing technology.【Method】The data which were from near ground hyper-spectrum, Landsat8satellite image and fieldresearch during cotton growing period in2013were analyzed using original reflectance, logarithmic,reciprocal, first derivative spectrum and anova analysis. Soil salinity characteristic bands were selectedout.Meanwhile, the estimating models for soil salinization were established based on correlation andregression analysis, and the accuracy of estimating models were assessed using RMSE (Root Mean SquarError), RE (Relative Error) and r (relatation coeffectance). We used ENVI5.0SP3Landsat8images ofexperimental zone to calibrate the radiation, atmospheric correction of pretreatment and analysis andestablished models.【Result】Through monitoring soil salinization in cotton field by using spectral technology, the followingresults were obtained:（1）The hyperspectrum characteristics of soil salinization in cotton field was illustrated.In400-2500nm band, salinization and non-salinization soil had similar spectral reflectance curves, andsalinization reflectivity value was greater than non-salinization, and with the severity of soil salinization thereflectance values became greater.（2）The estimating models for soil salinization were established.It indicated that the original spectral reflectance of1350-1850nm was selected out as sensitive band region.The bands of707nm,526nm,531nm were selected out as characteristic bands of the first derivativespectral. Then the salt indexes were built and, based on these indexes models were built and tested. R2ofSI1and BI was high, RMSE and RE were small. These showed that test results were ideal and it could berecommended as the best model to monitor soil salinity in cotton field.（3）Realization of monitoring soil salinization in cotton field by using satellite remote sensingThe results showed that each growth index of cotton including height, fruiting branches, dry matteraccumulation, leaf area had significant negative correlations with soil salinity. So using cotton dry matteraccumulation and leaf area index as the indirect monitoring index of cotton under salt stress was feasible.B5was the best wave band for monitoring cotton under salt stress. Blooming stage was the best period forsalinization monitoring. The monitoring index were built. Through regression analysis and test with soilsalt content, DVI-SI index monitoring model y=-3531.1X2+6976.3X-3444.1, R2=0.905**, high precision87.5%was good for monitoring soil salinity conditions in the key growing periods of cotton.