THE ROLE OF LOGISTIC REGRESSION AND GIS FOR ANALYSIS OF ENVIRONMENTAL HAZARDS (CASE STUDY: SYAHDARE WATERSHED, 2007-2009)

摘要

Understanding where the landslides are most likely to occur is crucial in reducing property damage and loss of life in future landslides. In this research logistic regression analysis as it requires less statistical assumption compared to other multiple statistical models and creating the best relationship between presence and absence of landslides and asset of causative factors to supply landslide susceptibility map in syahdare basin at first based on field surveys, local interview and review of previous works in similar region, ten primary causative factors on landslide occurrence such as elevation, slop gradient, aspect, rainfall, distance from fault, distance from drainage, distance from road, land use and lithology in study area recognized and their information layers has been created in GIS by using ARC GIS 9.2 soft wares based on photograph interpretation and field surveys. Seventy five landslides were recognized and also another 75 non landslides were selected randomly all over the basin after overlaying all pointsdandslides and non-landslides) with causative factors layers 1 and 0 codes belonged to presence and absence of landslides respectively. After entering independents variables including all coded classes and dependent variables including 150 landslides and non-landslides in to SPSS 12 and selecting forward stepwise method , data analysis were performed. Interpretation of coefficients obtained of logistic regression function analysis indicates that aspect and lithology, being miscorrelated with landslide occurrence by more than 0.05 significance are deleted from the model. At last, statistical model was performed based on the most effective factors on landslide occurrence including slope, elevation, rainfall, distance from drainage, distance from fault, land use and distance from road respectively. After transmitting this model to ARC GIS9.2 soft ware, landslide susceptibility map of syahdare basin was performed whit four classes. Therefore 51.94 residual area is located in high hazard regions. Model and then susceptibility map Verity was then susceptibility map Verity was assessed using -2LL, and Snell R2, Nagelkerk R2, occurrence ratio comparison and considering the deference percentage between landslide observed density and predicted probability and it was reliable.