Comparison of Semi-Quantitative and Statistical Regression Models in Assessing Landslide Prone Areas in Wahig-Inabanga Watershed, Bohol, Philippines

抄録

The study attempted to compare and evaluate the two landslide hazard assessment models, semi-quantitative (index-based) and statistical regression (bivariate statistical analysis and logit regression) in predicting landslide prone areas in Wahig-Inabanga Watershed, Bohol, Philippines. This was performed by comparing the predictive power of each model based on the frequency distribution of past landslide events. Findings revealed that the combined bivariate statistical analysis and logit regression model outdone index-based method in predicting landslide occurrences. Results indicated high prediction accuracy on statistical model greater than the 75% threshold level set for evaluation on both pooled moderate to very high hazard zone and the combined high and very high hazard zone with accuracy values of about 83.82% and 76.72%, respectively. Conversely, the semi-quantitative model failed to meet the accuracy threshold. The study showed that statistical regression model, though relatively difficult to implement, can be a better substitute to the most commonly used semi-quantitative method as a decision-support tool for watershed management and land use planning in relation to landslide risk mitigation, reduction, adaptation, and management.