Extended Abstract (204K)7.4
Probabilistic guidance for hurricane storm surge
Arthur A. Taylor, NOAA/NWS, Silver Spring, MD; and B. Glahn
Currently, the National Hurricane Center (NHC) begins operational storm surge model runs 24 hours before the forecast landfall of a hurricane. The resulting storm surge forecast products, which include the maximum the forecast storm surge attained in a given area and animations of the forecast storm surge, provide emergency managers an estimate of the likely storm surge based on the current NHC hurricane forecast. However, these operational storm surge products are based on a single run of the storm surge model. This is a problem since numerical models of storm surge inundation are dependent upon an accurate forecast of the hurricane's track and surface wind structure, and today's best forecasts still have considerable uncertainty. This makes it difficult for even the most accurate storm surge model to produce reasonable guidance from a single run.
To provide emergency managers with better guidance, National Weather Service storm surge experts have historically used ensembles of hypothetical storms to create potential flooding products. While these products are good for creating evacuation plans, they are not directly associated with the current hurricane forecast, and so can result in overestimates. An alternative probabilistic ensemble method has been developed for operational use. It creates an ensemble of hypothetical storms by combining the parameters, from the current hurricane forecast, used by the storm surge model to describe the hurricane's track and surface wind structure, with the historic error distributions associated with those parameters. The results created by running the ensemble of hypothetical storms through the storm surge model can then be combined, based on the likelihood of each storm's particular variation of the input parameters, to create probabilistic guidance for hurricane storm surge.
This paper will describe the probabilistic ensemble method and demonstrate the reliability of the probabilistic forecast. It will also describe the two experimental products used to portray the storm surge hazard.
Session 7, Probability Forecasting
Tuesday, 22 January 2008, 3:30 PM-5:15 PM, 219
Previous paper Next paper
Recorded presentation