IMPROVING OUR ABILITY TO PREDICT THE PROBABILITY OF DEBRIS-FLOW OCCURRENCE IN RECENTLY BURNED AREAS OF SOUTHERN CALIFORNIA

Two recent examples highlight the destructive nature of post-fire debris flows. On December 25, 2003, debris flows within the Grand Prix and Old burn areas killed 16 people near San Bernardino, CA. On February 6, 2010, debris flows produced in the Station burn area overtopped sediment retention basins and damaged or destroyed 46 homes in La Crescenta, CA. These events provide sobering examples of the threat that post-fire debris flows pose to lives, properties, infrastructure and important resources within and downstream of recently burned steeplands. Given these hazards, an important goal of the USGS Landslide Hazards Program is to provide guidance to government agencies, emergency managers and concerned citizens regarding the likelihood of debris-flow occurrence in or below recently burned areas.

We improve our estimates of post-fire debris-flow probability by analyzing new information related to debris-flow response and rainfall, basin morphometry (from higher resolution data), burn severity and soil characteristics from 14 burn areas. These new data increased the database from 306 records (fire years 2003 – 2006) to 1748 records (fire years 2007 – 2010). Instrumental records of debris-flow timing and associated rainfall intensities also provide new insights into the rainfall conditions that initiate post-fire debris flows. Formerly, probabilities were calculated using data from storms with durations in excess of three hours. Monitoring data highlight the importance of short-duration (≤ 30 minutes), high-intensity bursts of rainfall in generating debris flows, with little correlation between debris-flow initiation and rainfall over longer durations.

These advances warrant revisiting the current model of calculating post-fire debris-flow probability. The updated probability model incorporates readily available precipitation frequency estimates and variables derived from digital elevation, soils and burn severity data. The new empirical model of post-fire debris-flow probability outperforms earlier modeling efforts and relies upon more physically meaningful variables than earlier models. The updated model will be implemented in future post-fire debris-flow hazard assessments in southern California.