SOIL EROSION SUSCEPTIBILITY MAPPING USING SUPPORT VECTOR MACHINE AND REMOTE SENSING DATA IN SEMI-ARID ENVIRONMENT: SOUTH HIGH ATLAS OF MARRAKECH (MOROCCO)Zouhair Ourhzif Ahmed Algouti Abdellah Algouti Mohamed Ait Mlouka and Ali AyddaAbstract The aim of this study is to characterize the different kinds of water erosion and to detect their change from the satellite data of Landsat ETM + and OLI between 2000 and 2014. The approach is to identify the soils erosion from available spectral indices such as those describing the form of soil spectra (IF) color (IC) and hematite (IR). Soils can be characterised also by intensity (IW), hue (Tw) and saturation (S) and they were selected as indicators of soil condition. The results obtained allowed for two combinations of soil degradation. To improve these results, we have applied a supervised classification (SVM), this algorithm gives good results: has a low number of data and parameter of field. The overall accuracy Obtained for the soil classification is 65.13% (ETM+) and 95.21% (OLI). Validation against field reality demonstrated the potential of the method used for assessment to the state of soil resources in semi-arid environment. |