Preparing Digtal Maps of Some Soil Chemical Properties Using Satellite Data of landsat ETM+

The images taken by the TM Landsat 7 since September 2001 were used to prepare digital maps of certain soil chemical properties in northwestern Isfahan area. Geometric and Radiometric corrections were made on the images and then the images were recorded with 0.58 pixel accuracy. The images underwent processing, including image enhancement and PCA. Total 77 samples were collected randomly from the soil surface and analyzed in laboratory. Salinity, acidity, lime content and alkalinity were measured for each sample. Overlaying soil sample map on satellite image, the digital number (DN) of each sample point was extracted. Correlation and regression coefficients between DN and measured parameters were then estimated. The results showed no significant correlation between DN and Soil parameters. It was also showed that the correlation coefficients are higher for ETM+ 4, ETM+ 7 and ETM+ 1. It was clear that we cannot fit a regression model to the parameters that may be able to help us prepare soil map from satellite images. Therefore, it is suggested to use supervised classification methods to prepare soil maps.