NOAA/USGS DEMONSTRATION FLASH FLOOD AND DEBRIS FLOW EARLY-WARNING SYSTEM FOR RECENTLY BURNED AREAS IN SOUTHERN CALIFORNIA, AND LESSONS LEARNED FROM FOUR YEARS OF OPERATION

Flash floods and debris flows are common following wildfires in southern California. On December 25, 2003, sixteen people were swept to their deaths by debris flows generated from basins that burned the previous fall. In an effort to reduce loss of life and property by floods and debris flows, the National Oceanic and Atmospheric Administration (NOAA) and the United States Geological Survey (USGS) established a warning system for recently burned areas in eight counties of southern California in the fall of 2005. The prototype system was initially based on comparisons of precipitation (including radar estimates, in-situ measurements, and short-term forecasts) with rainfall intensity-duration thresholds defined for the occurrence of post-fire flash floods and debris flows. Separate sets of thresholds were defined for three geologically- and morphologically-similar areas of southern California for the first winter after a fire and following a year of vegetative recovery. The system was advanced in 2007 to provide forecasters with web-based information on each burned area, and to display hazard maps that show those basins most likely to produce the largest debris flows. During the four winters of operation, 64 warnings were issued with a success rate of 54%.

Three primary lessons have been learned during the four years of operation, and future work is focused on improving the program based on these lessons. First, a warning system based only on rainfall intensity-duration thresholds applied across entire burn areas or geologic/morphologic zones can only provide a rough approximation of life-and property-threatening conditions. The rainfall intensity-duration thresholds were not sufficiently precise to distinguish debris-flow triggering conditions from those that resulted in flooding, and the incorporation of more storm and response data did not increase the accuracy of the thresholds. Thus, warning systems that rely on rainfall intensity-duration information alone will result in high false alarm rates. Second, hazard mapping is necessary to provide the spatially explicit information that emergency-response personnel need. Finally, the lead times of hours to days provided by watches are more useful for emergency response than are warnings with their minute to hour lead times.