LANDSLIDE SUSCEPTIBILITY MAPPING IN GUERRERO, MEXICO

The tectonically active mountain landscape in southern Mexico has been recognized as a region highly susceptible to landslides because of frequent hurricanes, tropical cyclones and earthquakes. Thus identifying areas susceptible to landslides is fundamental to planning for land management and safe human occupation in the region. The goal of this study was to map susceptibility to landslides in Guerrero based on the analysis of hurricane Manuel induced landslides and high resolution LiDAR topographic data.

Landslides triggered by hurricane Manuel on September 2013 were mapped based on two approaches, including manual mapping using satellite images, and automatic identification of landslide morphology employing the CCMFlows algorithm (Contour Connection Method). Landslides were dominantly debris flows that occur mostly along zones of topographic convergence. A map of susceptibility to landslides was developed by computing probability of landslide occurrence from statistical relationships of existing landslides with LiDAR elevation derived landslide-causing factors using a logistic regression method. The accuracy of the model was computed based on receiver operating characteristics (ROC) curve approach.

Results showed high susceptibility zones, defined as probability of landslide occurrence > 0.6, encompass ~30% of the study area and occur mostly along topographic convergence. Area under the ROC curve indicated that the model has ~70% overall prediction accuracy. The approach identified most of the landslides within the high susceptibility zone and suggested that the approach is valid and applicable for mapping areas susceptible to landslides in southern Mexico.