Correcting model bias in multiscale coastal storm surge models with climate change

Assessment of flood risk in urban areas is critical to maintaining sustainable cities, especially with expected sea level rise and changes to the climate. The development of hydrodynamic models to predict coastal storm surge is well established, but results inherently contain errors when compared with historical validation data. Typically, storm surge is modeled at a large regional scale, with nested high-resolution models computing local O(m) scale fields. The correction to model water level results from observations involves not only time series analysis at point locations, but also spatial mapping of errors to the high-resolution nested model grid. We present a novel method to adjust model bias in storm surge predictions using a set of historical storm events, that maps the errors to the nested model boundary using a lagged covariance function. The method is illustrated for several field test cases with positive results.