Downscaling wind and wavefields for 21st century coastal flood hazard projections in a region of complex terrain

While Global Climate Models (GCMs) provide useful projections of near-surface wind vectors into the 21st century, resolution is not sufficient enough for use in regional wave modeling. Statistically-downscaled GCM projections from Multivariate Adaptive Constructed Analogues (MACA) provide daily averaged near-surface winds at an appropriate spatial resolution for wave modeling within the orographically-complex region of San Francisco Bay, but greater resolution in time is needed to capture the peak of storm events. Short-duration high wind speeds, on the order of hours are usually excluded in statistically-downscaled climate models, and are of key importance in wave and subsequent coastal flood modeling. Here, we present a temporal downscaling approach, similar to constructed analogues, for near-surface winds suitable for use in local wave models, and evaluate changes in wind and wave conditions for the 21st century. Reconstructed hindcast winds (1975-2004) recreate important extreme wind values within San Francisco Bay. A computationally-efficient method for simulating wave heights over long time periods was used to screen for extreme events. Wave hindcasts show resultant maximum wave heights of 2.2 m possible within the Bay. Changes in extreme over-water wind speeds suggest contrasting trends within the different regions of San Francisco Bay, but twenty-first century projections show little change in the overall magnitude of extreme winds and locally-generated waves.