Saline and alkaline soil mapping using ASTER data in the Qazvin plain

Salinity and alkalinity are two major phenomena leading to soil degradation in semiarid and arid areas. The main aim of this study was to evaluate the capability of ASTER data to provide soil salinity and alkalinity mapping in the selected parts of the Qazvin plain, which is known as an arid area. In this study, spectral classes were provided from sensed data, and with the help of field observation and soil analysis reorganized to have soil salinity and sodicity classes. Finally, soil salinity and sodicity maps were prepared. Soil sampling was implemented using stratified random sampling method, depending on landscape complexity and homogeneity, as well as on the representativity to ASTER data. Furthermore, at least one profile was studied in each soil map unit in order to examine subsoil salinity variation. Field samples from augur and profiles were analyzed in laboratory for Na+ , Ca2+ , Mg2+ cations, as well as soil texture , ECe and pH. We have analyzed additional data such as digital elevation model and slop that may improve the accuracy of classification. In addition, NDVI, SRVI, PVI, SAVI, SI, BI and NDSI indices, and PCA were analyzed. The results indicated that the combination of DEM with them ASTER bands would lead to highest accuracy. This study showed that thermal bands of ASTER increased the classification accuracy, and this illustrated its effective role to classify the soil salinity and sodicity. PCA had almost highest accuracy, among studied processing techniques. The indices had low accuracy in differentiating the saline soils. The optimum index factor had low overall accuracy. The sodicity map was less accurate as compared to the salinity map. The accuracy for moderate sodicity levels was less than the accuracy for low and high sodicity levels.