Accurate storm surge forecasts are of paramount importance for coastal areas which are sensitive to flooding, especially if they are densely populated and well-developed. As an example, in the Netherlands an important part of the economic activity and population concentration is close to the coast in a part of the country which is near or even below the sea level. Hurricane Katrina and the subsequent flooding of New Orleans have shown the disastrous effects of failing coastal defences.

Operational systems provide deterministic sea level forecasts for the North Sea for up to 48 hours ahead. Current research focuses on adding information on the uncertainty in the forecasts for which there is an increasing demand.

Material & Method:

The standard tool for storm surge forecasts in the Netherlands is the 2D shallow-water model WAQUA/DCSM, which covers all of the NW European Continental Shelf with a resolution of currently 8 km, and receives its atmospheric forcing from the Hirlam limited area model.

There are several ways to add uncertainty information. This paper will present a way to generate probability forecasts with WAQUA/DCSM forced with data from the ECMWF Ensemble Prediction System (EPS). EPS generates a control forecasts and 50 perturbed forecasts of 240 hours. For optimum performance with ECMWF input, the model needs to be recalibrated. Moreover, an additional calibration of the ensemble with the aid of Rank Histograms is performed.

Result & Conclusion:

Brier Skill Scores show that useful probability forecasts can be made available for 2 - 5 days ahead.

The ensemble has been running in experimental mode since February 2007. The results are made available to forecasters at KNMI and the Storm Surge Warning Service.

A storm in March was successfully forecasted as much as a week ahead.