Qualitative and semi-quantitative debris flow susceptibility mapping of mining communities in Peru using logistic regression models

Abstract

Debris flows are geological phenomena in which saturated masses of soil and fragmented rock surge down slopes, entrain objects in their paths, and form thick, muddy deposits on valley floors. The purpose of this study was to develop protocols for debris flow hazard characterization for the Ocoña River Valley in the Arequipa region of Peru by qualitatively and semi-quantitatively analyzing the relationship between debris flow causative factors and the occurrence of debris flows. Logistic regression was used to identify areas susceptible to debris flows based on a combination of input parameters. The models produced logit equations which were then input into ArcGIS to create susceptibility maps of debris flow deposits and sources. The results of the study revealed that the semi-quantitative approach models performed best for the deposits, whereas the qualitative approach models performed better for the sources. This was the case for both the visualization of the susceptibility maps and the comparison of them to the inventories. According to the ROC curve results, the qualitative approach produced models that outperformed the semi-quantitative approach for both deposits and sources. This work should result in improved safety for the nearby communities by identifying potential locations of future debris flows.