Evaluation of Sub-Seasonal Coastal Flooding Forecasts for Extreme Events; Hurricane Isabel and Katrina Case Studies

Extreme weather events frequently impact coastal areas of the United States (US) resulting in severe flooding. The current operational lead-time of coastal flood forecasts in the US is limited to short-range temporal scales (7 days), which limits the response time for emergency preparations. The sub-seasonal prediction project (SubX), which produces weather predictions up to four weeks lead time, presents an opportunity to evaluate the potential for developing longer-lead time probabilistic flood forecasts. Using the coastal storm surge model ADCIRC, two major hurricanes, Isabel (2003) and Katrina (2005) were used as case studies for evaluating coastal flooding forecasts forced by wind and pressure fields derived from five global weather models within SubX. These coastal flooding forecasts were compared to storm surges generated by the same modeling framework forced by the NCEP-Reanalysis, and measured storm surges at NOAA gauges. The results were evaluated in terms of lead-time and forecast skill metrics. Ensemble-based storm surge forecasts driven by several members of SubX models demonstrate skill beyond a 4-day lead-time and up to ~10 days. Some ensemble members forecast hurricanes and storm surges as early as 3-4 weeks lead-time, however offsets develop in the timing and magnitude of the peak at these lead-times and the significance of these longer lead time results is hard to establish based on a sample size only two events. While flood reforecasts over the entire SubX reforecast period (1999-2015) are needed to build robust statistics of the forecast skill, these case studies demonstrate the potential feasibility of probabilistic flood forecasting at subseasonal timescales using the SubX models.