Predicting coastal waves with buoy observations and global model output, a Southern California case study
- Author(s): Crosby, Sean Christopher
- Advisor(s): Guza, Robert T
- et al.
Accurate coastal wave predictions are needed to support nearshore process modeling (transport, erosion) and local risk assessment (flooding, inundation). Small errors in direction or frequency have relatively large impacts to transport, run-up estimates. Nearshore prediction error is often dominated by offshore uncertainties in regional models. Detailed, frequency-direction, spectra are needed offshore to accurately estimate blocking and refraction in sheltered regions, e.g.\ the Southern California Bight (SBC).
Here, nearshore prediction skill using both global wave model predictions and offshore buoy observations in the offshore boundary condition is compared for swell-band wave energy (0.04-0.09Hz). Despite inherent directional ambiguity, offshore buoy observations yield a more accurate boundary condition. Analytical and ah hoc combinations global model predictions and offshore buoy observations are tested and yield worse and marginally improved predictions respectively (as compared to buoy-driven predictions).
Coastal regions often contain sheltered nearshore wave observations. Assimilating nearshore observations in regional wave models theoretically will improve regional skill. However, nonlinear wave propagation models (e.g. SWAN, WW3) are difficult to invert. Recent developments employ variational methods, but success in complex real-world environments has yet to be observed. At swell-bands wave energy propagation across narrow shelves (e.g. U.S. West Coast) is dominated by linear processes (refraction, shoaling) and well modeled by self-adjoint ray tracing. Here, a general assimilation framework is developed to estimate physically smooth (time, direction), accurate, offshore boundary conditions from offshore and sheltered buoy observations with global model predictions included as a model prior. Case studies show error reduction at validation (non-assimilated) buoy sites suggesting that assimilation of nearshore observations improves regional skill. Initial results suggest that few (1 offshore, 1-2 nearshore) buoys are needed to sufficiently resolve offshore conditions, which has implications for cost-effective buoy array design. Additionally, buoy sites with significant misfit to assimilated observations identify regions of model error suggesting missing model physics (e.g. diffraction, reflection).