Using multivariate approach and GIS applications to predict slope instability hazard case study of Machrouha municipality, NE Algeria

Landslides have caused severe damages in the northeastern Algeria mountain region during the last decades. The present research use a raster-based GIS and statistical processing with a probabilistic method (multiple logistic regression approach, LR) for landslides hazard assessment and mapping in Machrouha municipality. Data included digitized geology, slopes, exposure, elevations, forests, roads, settlement, streams; precipitations and past events, etc. was manipulated under a GIS platform using ‘Arcgis 9.3’ software. Landslides location was defined from a stereoscopic interpretation of satellite images completed by a field survey. The distribution of 104 landslides events occurred between (1987 2012) was compared with 12 causal factors. LR relates predictor variables to the occurrence or nonoccurrence of landslides within geographic cells. This operation shows a strong correlation between theses dependent and independent variables. Different classes were assigned for each thematic layer; a corresponding weighting and rating values were attributed for each data class, subclasses and layer in this GIS processing. By applying LR approach, landslides hazard areas were assessed and mapped into four ranks using the correspondence between landslides occurrence and permanent factor maps. The accuracy of the probabilistic model was verified by the overlaying of a new collected landslides pack location with the final landslides hazard map.