The application of field soil salinity map in classifying Landsat imagery

Soil degradation including soil salinity, erosion and waterlogging are among main factors of desertification. In this research, attempts are made to evaluate the efficiency of Landsat TM data dated 28 June 1987 and geographic information system for studying soil salinity and salt affected vegetation. The study area is a part of Qazvin plain, located in the Roudkhaneh Shour catchement. The optimum band combination was selected for image classification using optimum index factor, two dimensional feature space analyses and principal component analysis. The training classes were selected based on the type, density of vegetation and soil salinity classes. Then 18 spectral classes were classified by maximum likelihood method. These classes were regrouped in to three in formation classes and salt affected vegetation map was obtained. In order to calculate the correlation coefficient between the maps, the spectral and field soil salinity maps were crossed. The soil salinity map of the study area was elaborated in 1984. In order to map soil salinity, the salinity level of the soil was first measured for each horizon. Then a salinity rating is given for the layer: 0 - 50, 50 - 100 and 100 - 150 cm, by averaging the Electrical conductivity ( Ec) data of the horizons. The three soil salinity ratings of the horizons were then converted in to one final rating for the profile. The results have shown that the accuracy of spectral map obtained from remotely sensed data is about 8.4%, but its correlation with field soil salinity map is about 60% that is relatively low. The reason may be attributed to different approaches of mapping soil salinity and effect of some other soil physico-chemical properties the surface reflectance.