Comparison of Logistic Regression, Information Value, and Comprehensive Evaluating Model for Landslide Susceptibility Mapping

This study validated the robust performances of recently proposed comprehensive landslide susceptibility index model (CLSI) for mapping (LSM) by comparing it to logistic regression (LR) and analytical hierarchy process information value (AHPIV) model. Zhushan County in China, with 373 landslides identified, was used as area. Eight conditioning factors (lithology, slope structure, angle, altitude, distance river, stream power index, length, road) were acquired from digital elevation models (DEMs), field survey, remote sensing imagery, government documentary data. Results indicate that CLSI has highest accuracy best classification ability, although all three can produce reasonable (LS) maps. The performance is due its weight determination a back-propagation neural network (BPNN), which successfully captures nonlinear relationship between occurrence factors.