Improved Prediction of Postfire Debris Flows Through Rainfall Anomaly Maps

Abstract

Predicting where runoff-generated debris flows might occur during rainfall on steep, recently burned terrain is challenging. Studies of mass-movement processes in unburned areas indicate that event locations are well-predicted by rainfall anomaly, R*, in which peak observed rainfall is normalized by local rainfall climatology. Here, we use remote and field methods to map debris flows triggered within the 2020 Dolan Fire burn area in coastal California, demonstrate that a short-duration R* metric predicts debris-flow occurrence more effectively than absolute peak intensity or longer-duration rainfall metrics, and show that incorporating an R* criterion into an existing debris-flow likelihood model can reduce false positive predictions and improve accuracy. We test R* at three other climatically distinct fires in California, demonstrating its utility for mapping likely debris-flow locations in different climates. We also consider how R* can benefit postfire debris-flow prediction given recent increases in climatological variability within individual burn perimeters.