Spatial Prediction of Slope Failures in Support of Forestry Operations Safety

Abstract

This study produces a slope failure susceptibility map for evaluation of the Caspian Forest for its capacity to support road construction and timber harvesting. Fifteen data layers were used as slope failure conditioning factors, and an inventory map of recent failures was used to detect the most susceptible areas. Five different datasets of conditioning factors were constructed to compare the efficiency of one over the other in susceptibility assessment. Slope failure susceptibility maps were produced using an adaptive neuro-fuzzy interface system (ANFIS) and geographical information system (GIS). The accuracy of the maps was then evaluated by the area under curve (AUC). The validation results suggest that the ANFIS model with input conditioning factors of slope degree, slope aspect, altitude, and lithology performed the best (AUC=83.74%) among the various ANFIS models explored here. The five ANFIS models have performed reasonably well, and the maps allow development of prudent hazard mitigation plans for the safety in forestry operations.