Soil salinity in arid and semi-arid lands is one of the most important limiting factors which changes vegetation types and biomass due to natural resources production reduction. The Landsat 8 satellite images (2014) were used in this research to select the best satellite indices for soil salinity evaluation. All soil samples were conducted in September 2014. Based on 77 points of measurement the distribution maps of sodium, magnesium, potassium, calcium, electrical conductivity and soil acidity were prepared by Kriging interpolation method which was developed in ArcGIS®9.3 software. After that, the correlations between the produced maps and ten remote sensing indices have been investigated by use of spatial regression. Maps of the distribution of sodium, potassium, magnesium, calcium, soil conductivity, acidity, salinity and alkalinity also prepared and proper regression models were presented. The results show that for the detection of distribution of electrical conductivity and sodium, according to correlation coefficient, the Salinity Index and Malayer Salinity Index were suitable indices. In order to detect the distribution of magnesium, calcium and potassium in the study area due to the high correlation coefficient (0.88), the normalized difference salinity index can be used. Due to the not significant difference of spatial regression of soil alkality, it could not be used. The results showed that the normalized difference salinity index can be used for general measurement of all soil elements. According to the regression equation derived between indices and prepared maps of field studies, the optimal models for soil salinity mapping of the study area were determined and calibrated. Based on satellite data, obtained models of this study have suitable estimates of study elements because their coefficient of correlation is acceptable. With the completion, expansion and development of findings of this study, zonation lands without the need for sampling could be done. This method, while providing greater precision can also minimize the sample costs.
Manuscript profile
