Using Geologic and Geomorphic Tools in USACE Hydrologic Loading Evaluation, Carbon Canyon Dam, Orange County, California

USACE dam safety risk assessments consider potential hazards (loading), system responses (fragility) and associated consequences. Where possible, paleoflood data are incorporated to reduce uncertainties in hydrologic loading. Paleoflood analysis of Carbon Canyon Wash provides estimated magnitudes and frequencies of large floods within the past 2,000 years for improving long-term hydrologic loading for a USACE dam.

High-resolution topographic data and field geomorphic mapping identified fluvial terraces that provide geologic/geomorphic evidence of large floods (i.e., paleostage indicators, PSI). Geomorphic and stratigraphic evidence shows that an 8-m-high terrace flooded about 1,725 years ago, and an inset 4.5-m-high terrace formed about 900 years ago (ages from relative soil-profile, radiometric, and OSL dating). Using HEC-RAS software, the detailed computational mesh captures contemporaneous PSI topography; the 2D hydraulic modeling was calibrated using geologic evidence of the 1938 flood of record and later floods, and included a sensitivity analysis using a suite of reasonable channel roughness values and downstream boundary conditions. Comparison of down-valley terrace profiles and HEC-RAS discharge profiles yields estimated paleodischarges of 60,000 cfs (+33%) for the high terrace and 20,000 cfs (+50%) for the inset terrace. Primary sources of uncertainty are ranges in: (1) PSI age; (2) roughness and energy gradient for various discharges; (3) water depth and velocity required for sediment deposition; and (4) inherent variability of water and PSI surfaces. The best-estimates and ranges in PSI discharge and age are included into flow-frequency analysis via perception thresholds and discharge ranges.

The Expected Moments Algorithm was used in a Bulletin 17C flow-frequency analysis of the systematic/historic record with and without paleoflood data. While including paleoflood data does not significantly change estimates for expected mean and standard deviation, the paleoflood frequency model contains a less-negative skew, and a longer equivalent record length (53 vs 99 years), nearly doubling the data record. Overall, inclusion of paleoflood data allows an interpretation that large discharges may be more frequent than would be estimated from only systematic and historic records.