Validated coastal flood modeling at Imperial Beach, California: Comparing total water level, empirical and numerical overtopping methodologies

Abstract

Flood extent field observations collected during a winter storm (Hs∼1.8m, Tp∼14s) coinciding with a spring high tide are used to evaluate the accuracy of static (‘bathtub’) and hydrodynamic coastal flood modeling methodologies. Static models rely on empirically calculated wave setup or runup and simply compare total water level (TWL) to land elevation. The dynamic model resolves temporally variable overtopping rates, overland flow, urban features and storm system drainage. SWAN, a numerical wave model, transformed deep water buoy spectra to the nearshore. Static TWLs were calculated using SWAN output and an empirical runup model. Numerical (XBeach) and Empirical (EurOtop) overtopping models parameterized with survey data and SWAN bulk wave statistics estimated temporally variable overflow rates along representative transects for overland flow model input. XBeach model mode (hydrostatic, nonhydrostatic), boundary depth and random realizations significantly affected overtopping rates. Nonhydrostatic mode estimated order of magnitude larger overflow volumes suggesting the importance of incident waves, particularly in near threshold conditions. Boundary depth and random realizations varied overflow rates approximately fourfold. Field observations showed static TWL models performed poorly, maximum runup substantially overestimated flood extent and setup predicted no flooding. All dynamic models reasonably predicted flood extent despite significant overflow volume differences. Backshore topography and flow dynamics are important flood extent controls. Accurate near threshold coastal flood predictions require dynamic overland flow modeling parameterized with temporally variable overtopping estimates and site specific beach and backshore topography.