Skip Links

USGS - science for a changing world

Open-File Report 2009-1200

Multivariate Statistical Models for Predicting Sediment Yields from Southern California Watersheds

By Joseph E. Gartner, Susan H. Cannon, Dennis R. Helsel, Mark Bandurraga


Thumbnail of cover and link to report PDF (3.3 MB)

Debris-retention basins in Southern California are frequently used to protect communities and infrastructure from the hazards of flooding and debris flow. Empirical models that predict sediment yields are used to determine the size of the basins. Such models have been developed using analyses of records of the amount of material removed from debris retention basins, associated rainfall amounts, measures of watershed characteristics, and wildfire extent and history. In this study we used multiple linear regression methods to develop two updated empirical models to predict sediment yields for watersheds located in Southern California. The models are based on both new and existing measures of volume of sediment removed from debris retention basins, measures of watershed morphology, and characterization of burn severity distributions for watersheds located in Ventura, Los Angeles, and San Bernardino Counties. The first model presented reflects conditions in watersheds located throughout the Transverse Ranges of Southern California and is based on volumes of sediment measured following single storm events with known rainfall conditions. The second model presented is specific to conditions in Ventura County watersheds and was developed using volumes of sediment measured following multiple storm events. To relate sediment volumes to triggering storm rainfall, a rainfall threshold was developed to identify storms likely to have caused sediment deposition. A measured volume of sediment deposited by numerous storms was parsed among the threshold-exceeding storms based on relative storm rainfall totals.

The predictive strength of the two models developed here, and of previously-published models, was evaluated using a test dataset consisting of 65 volumes of sediment yields measured in Southern California. The evaluation indicated that the model developed using information from single storm events in the Transverse Ranges best predicted sediment yields for watersheds in San Bernardino, Los Angeles, and Ventura Counties. This model predicts sediment yield as a function of the peak 1-hour rainfall, the watershed area burned by the most recent fire (at all severities), the time since the most recent fire, watershed area, average gradient, and relief ratio. The model that reflects conditions specific to Ventura County watersheds consistently under-predicted sediment yields and is not recommended for application. Some previously-published models performed reasonably well, while others either under-predicted sediment yields or had a larger range of errors in the predicted sediment yields.

First posted October 8, 2009

Part or all of this report is presented in Portable Document Format (PDF); the latest version of Adobe Reader or similar software is required to view it. Download the latest version of Adobe Reader, free of charge.

Suggested citation:

Gartner, J.E., Cannon, S.H., Helsel, D.R., 2008, Multivariate statistical models for predicting potential sediment yields from Southern California watersheds: U.S. Geological Survey Open-File Report 2009–1200







Summary and Conclusions


Appendix 1. Data Used To Develop Multiple Regression Models

Appendix 2. Statistics for Multiple Regression Models

Appendix 3. Application of Model 1 To Generate Maps

Accessibility FOIA Privacy Policies and Notices

Take Pride in America logo logo U.S. Department of the Interior | U.S. Geological Survey
Page Contact Information: GS Pubs Web Contact
Page Last Modified: Wednesday, 07-Dec-2016 22:01:07 EST