DERIVING LAND USE AND CANOPY COVER FACTOR FROM REMOTE SENSING AND FIELD DATA IN INACCESSIBLE MOUNTAINOUS TERRAIN FOR USE IN SOIL EROSION MODELLING

摘要

Soil erosion is one of the severe land degradation problems in many parts of the world. It not only affects on decreasing agricultural productivity but also it causes disasters such as siltation of reservoirs and flooding in low lying areas in case of high rainfall events. To predict soil losses various models are available the results of which can be used for formulating soil conservation planning. In order to run the predictive models, one of the crucial data required is the land cover/land use and the canopy cover information. While land cover data can be used to derive information on effective soil hydrological depths and to estimate kinetic energy of leaf drainage, vegetation canopy cover will give information on rain interception factor, both of which are important input in calculating runoff and soil loss. These data may not be easily available in many mountainous areas because of inaccessibility problem. In this situation remote sensing data becomes very important. For classification of remote sensing data, however, one has to keep into account the problem due to illumination variations induced by variations in topography in mountainous areas, which may not lead to normal distribution of training samples, an assumption required by maximum likelihood classification. To solve this problem, illumination variation was removed using different techniques. The resulting data was classified to generate land use map. The result shows improvement in classification accuracy. The cover factor or the C factor is generally estimated using normalized difference vegetation index (NDVI) assuming a linear relationship which may not be the case always. The surface cover includes not only vegetation canopy cover but also plant residue and soil surface cover. In this study C-factor is derived by using in first place field assessment and then followed by calculating normalized difference vegetation index (NDVI). Correlation is carried out and a curve is fitted. The estimated C-factor represents the percent ground cover for each land cover type, as well as the presence of plant residue. The result was tested for its reliability and estimation error using field data, which shows that the coefficient of efficiency was higher (0.77) and the root mean square error was 0.03. The study was applied in a watershed in northern Thailand