Fully Automatic Mapping for Vegetation and Soil Erosion Monitoring Using Remote Sensing, Mathematic Modelling and GIS Techniques

摘要

Traditional mapping for vegetation and soil erosion monitoring was usually done by visual airphoto or image interpretation or by computer-aided classification and interpretation. These methods are time-consuming and less accurate. Here a new method is presented, which only uses landsat TM images, mathematic models and GIS techniques without using of any visual interpretation and computer-aided mapping. This research also designed and experimented on a new atmospheric correction method for vegetation index calculation. This method depends upon one or few correct reference values, which were extracted from TM image itself or derived from ordinary geographic statistics. For example, by TM images classification or dynamic color composition, the average vegetation coverage value for a study area (only one value for one scene of image) can be obtained, with which the due average vegetation index value can be deduced. Continually, the due correct parameters for atmospheric correction can be obtained. Using these correction parameters, the correct vegetation index values and maps (all pixels have their values) was calculated and generated. By mathematic models, the vegetation coverage maps and vegetation coverage grades maps were calculated and generated. Thereupon the vegetation change monitoring was realized. By vegetation coverage grades maps and slope grades map from DTM, the soil erosion grades maps were calculated and produced and the dynamic soil erosion monitoring was also realized.