Inventario de deslizamientos y modelado de la susceptibilidad mediante regresión logística múltiple en la cuenca del río La Carbonera, flanco SE del volcán Pico de Orizaba, México

Abstract

The Río La Carbonera watershed (71.9 Nm2), located on the southeastern flanN of Pico de Orizaba volcano (5675 masl), has been selected as a case study to characterize areas that are prone to slope instability in volcanic and sedimentary terrains. The oc- currence of landslides was determined through a landslide inventory map created from multi-temporal aerial photographs, field worN, and with an adaptation of the Landslide Hazard Zonation Protocol of the Washington State, USA, Department of Natural Resources. Two hundred and thirty-five gravitational processes of 6 types were recognized: shallow, debris-avalanche, deep seated, debris and earth flows, me- ander cutting, and rocN fall. The spatial analysis of the relationship between landslide distribution and geology revealed that gravitational processes are more abundant in the mid-upper part of the watershed, where deposits and volcanic rocNs are the pre- dominant. In this worN the landslide susceptibility was modeled by using the statisti- cal model of Multiple Logistic Regression (RLM) through the use of LOGISNET system and SPSS statistical program. The RLM was used to examine the relationship between landslides and cartographic variables (altimetry, slope, terrain curvature, contributing area, flow direction, geology, land use, geology, and edaphology) in or- der to create the susceptibility map. With four of eight independent cartographic var- iables, a resolution of 10m per pixel, the susceptibility model tends to over-predict landslides. However, the model is statistically valid and capable of predicting 72.63%  of the existing landslides. The implementation of the landslide inventory and the sus- ceptibility modeling show the feasibility of the method to be used in other areas of Mexico.