Estimating the Response of Physical Processes in the South San Francisco Bay for Flood Stage Frequency Analysis

Coastal flooding in the far south San Francisco Bay (SSFB) can be a function of astronomical tide, residual tide, in-bay wind speed and direction and fluvial discharge. These physical processes and coastal levee failure were considered as input parameters into a Monte Carlo Simulation (MCS) to estimate coastal flood stage frequency in the SSFB. Limited data is available in the SSFB to estimate the contribution of these physical processes to coastal flood statistics. Over 100 years of measured water surface elevation (WSE) is available at the San Francisco tide station which can be used as input to hydrodynamic model simulations to estimate the WSE response in the SSFB. Data sampling criteria have been developed to select significant events at the San Francisco tide station for data transfer to the project site and statistical analysis. The coincidently sampled astronomical and residual tides at the San Francisco tide station were analyzed to cover the full range of the combinations of astronomical and residual tides that contribute to coastal flood statistics at the project site. A look-up table of astronomical and residual tide in the form of WSE responses at the project site from the hydrodynamic simulations was established for the interpolation in the MCS. The hydrodynamic model simulations indicated that the higher-high astronomical tides between 5.15 and 7.25 feet MLLW amplifies with a factor of 1.40 to 1.90 as a function of tidal frequency and water depth, including tidal range. The residual tide varies minimally as it propagates into the SSFB. In-Bay wind set-up from a significant event was found to contribute on the order of one foot to the total WSE in the SSFB; however, wind events with strong magnitudes along the primary axis of the bay occur infrequently making an insignificant contribution to the overall flood statistics. The fluvial discharges of Guadalupe River and Coyote Creek were considered in the hydrodynamic simulations as they are located within the study area. The San Francisco tide station and fluvial flow stream gauges were used to correlate the peak fluvial discharge with coincident residual tide. A mean phase shift of 6 and 11 hours for Guadalupe and Coyote Creek, respectively were calculated between peak fluvial flow and peak residual tide. The PDFs of astronomical tide, residual tide, local wind speed and direction, and levee failure were developed as input into the MCS. It was concluded that using hydrodynamic model simulations to estimate the response of the physical processes and levee failure at the project site in conjunction with the application of MCS is a reasonable way to estimate coastal flood stage statistics within acceptable uncertainty limits.