USING DEBRIS RETENTION BASIN DATA TO UPDATE MODELS THAT PREDICT SEDIMENT VOLUME FROM DEBRIS FLOWS AND FLOODS

Debris flows and sediment-laden floods pose serious hazards to residents living in southern California. Wildfire can amplify these hazards, resulting in extreme sediment accumulations in response to relatively moderate rainfall. Such extreme events may overwhelm structures designed to mitigate the flood and debris-flow hazards. The design of these structures is generally based on multiple-regression models derived from historical measurements of sediment accumulations in debris basins. Information on rainfall and associated sediment collected in debris basins in Ventura County, CA from 1968 to 2005 is used to identify rainfall thresholds above which sediment will be produced, generate new multivariate statistical models that predict sediment yields, and determine the accuracy of the models.

Sediment removed from debris basins may include material deposited by many storms. In addition, many of the storms during periods between sediment removals may not have been large enough to produce any significant sediment from runoff. In order to identify these storms, we examined the rainfall records from periods where no debris was deposited to define the threshold rainfall intensity-duration conditions below which we would not expect sediment delivery to a basin. A weighted average of the remaining threshold-exceeding storms was used to divide the volumes of sediment removed from debris basins into single event data. Multiple regression was used to generate models that predict debris flow volume as a function of storm rainfall, fire history, soil properties and drainage network.

A test of the models, using addtional rainfall and volume of debris flow data from 2003 in San Bernardino, indicated that the models generally under-predict the volume of sediment runoff. This may be due to the lack of large sediment volumes in the data used to generate the models. Sediment volumes were divided among many different storms, when only one or two storms may have produced the majority of the sediment. As new information on the relationship between individual storms and deposited sediment volumes, and the extent and severity of wildfire preceding a depositional event becomes available, better models can be developed.