SPATIALLY-EXPLICIT PREDICTION OF RAINFALL INTENSITY-DURATION THRESHOLDS FOR THE GENERATION OF POST-FIRE DEBRIS FLOWS IN THE WESTERN UNITED STATES

Population expansion into fire-prone steeplands has resulted in an increase in post-fire debris-flow risk in the western United States. To reduce risk in southern California, the United States Geological Survey (USGS) and the National Weather Service (NWS) have developed a post-fire debris-flow early-warning system in Los Angeles and San Diego. Rainfall intensity-duration (ID) thresholds for the generation of runoff-induced post-fire debris-flows comprise a major component of the warning system. This system compares forecasted and real-time estimates of rainfall intensity to the empirically-derived ID thresholds. How closely the forecasted or observed rainfall rates compare to the ID thresholds is a major factor in the decision-making process for issuing debris-flow outlooks, watches, and warning. The success of this program in southern California has increased interest in the expansion of the warning system into other regions. A major limitation of the current empirical approach is the need for an extensive library of rainfall and basin response information from which the ID thresholds can be calculated, which is time-consuming to develop.

Traditional ID threshold methodology can be combined with logistic regression models that predict post-fire debris-flow probability to create a framework for the calculation of spatially-explicit ID thresholds. The approach presented here synthesizes these methods by incorporating rainfall terms into each model variable within the logistic regression equation. Model performance is evaluated not only by statistical results, but also by how well the model equation defines the rainfall intensity-duration threshold for individual basins for which a threshold has been identified. The model can be applied in a spatially-explicit manner, which allows for site-specific characterization of both the likelihood that a debris-flow will occur at a given rainfall intensity and direct calculation of the rainfall rates that will result in a given likelihood. With additional validation, this model may permit the prediction of spatially-explicit ID thresholds for debris-flow generation and allow for more rapid expansion of the USGS/NWS early-warning system into other susceptible regions.