Derivation of vegetative variables from a landsat tm image for modelling soil erosion

A study was carried out to assess the potential use of satellite thematic mapper (TM) images to produce maps of vegetation‐related variables for erosion modelling. In a Mediterranean study area in southern France the (semi‐)natural vegetation was described at 33 field plots using four quantitative methods: the Fosberg structural classification system, the cover and management factor of the Universal Soil Loss Equation, the leaf area index and the total percentage cover. After radiometric correction of the image, the spectral TM bands were processed following three different methods. Each method aimed at combining the data of the six spectral TM bands into a single band in such a way that the resulting image displayed optimal information on green vegetation cover. The algorithms used comprise the normalized difference vegetation index, the conventional ‘tasselled cap’ transformation and a locally tuned tasselled cap transformation. Only slight differences were found between the different methods to calculate spectral vegetation indices for this particular case. Furthermore, the correlations between the field variables and image‐derived spectral indices are generally small. The largest correlations were found for the normalized vegetation index and the leaf area index (r + 0·71) and for the normalized vegetation index and Fosberg's structural vegetation classes (r + 0·76). However, Fosberg's method results in very general classes, which are of little use for soil erosion models. Furthermore, the spectral indices appeared to be sensitive for the vitality of the vegetation. Consequently, an area covered by a sensed, senescent vegetation will not yield a large value for the spectral index, but its soil is protected against splash erosion. This might lead to a misinterpretation of the soil protective cover when satellite images are used. A final conclusion is that a balance has to be found between the more accurate, but time‐consuming field surveys to gather information on erosion‐controlling factors and a certain loss of accuracy associated with the use of quick and easy remote sensing methods.