Investigating Potentiality of IRS-P6 Images for Soil Salinity Modeling

Saline soils are one of the agricultural problems in many parts of the world. Soil salinity knowledge is an essential factor for proper land planning and management. Preparing soil salinity map based on fieldwork, especially in large areas, is time-consuming and costly. One of the useful technologies for soil salinity mapping is remote sensing for which various satellite data have been evaluated. In this study, potentiality of IRS-P6 images for soil salinity modeling in 180000 ha of Mahabad county lands was investigated. For this purpose, 147 soil samples from a depth of 0-15 cm using a systematic grid 1000 × 1000 meter were collected, and also the saturated electrical conductivity was measured. On the other hand, after the geometric correction of images, synthetic bands and salinity indices were generated. DNs of corresponding points in all bands and indices were extracted. After data normalization, correlation coefficient value between DNs and salinity was extracted. Band 2 and Fusion index (Pan and B2) had the highest correlation values of 0.73 and 0.736 respectively. The stepwise regression analysis was performed and a salinity model was developed with 0.628 correlation value. Finally, the model validation was performed using 20% of the samples, and the values of RMSE, ME and correlation coefficients between actual and estimated salinity values were obtained. The results indicated that IRS-P6 images could explain about 0.39 of soil salinity variation which reflect the high capability of the images in soil salinity modeling, according to the situation of the study area.