Using weather data, a traditional climate calculation model, and artificial intelligence, Dr. Daniel Krieger of the University of Hamburg’s Center for Earth System Research and Sustainability (CEN) is able to provisionally forecast the frequency and height of storm surges for the next 10 years. The results were published in the journal Geophysical Research Letters.
The examples of Cuxhaven, Esbjerg (Denmark), and Delfzijl (Netherlands) show that the study’s forecasts are reliable. For instance, an average of 11.6 storm surges per year were recorded in Cuxhaven for the last 10 years. The model predicted 12.8 storm surges per year for the same period, with a tolerance range of +/-1.6 storm surges.
"Until 2029, the value remains similar at 12 storm surges per year," says climate modeler Daniel Krieger. The height of the floods, however, is different. "While the highest annual storm surge in the last 10 years averaged 2.5 meters, our model indicates an average of 3 meters for the next 5 years," says Krieger.
To date, climate models have been able to calculate only whether more storms will occur in the North Sea in the future but not how they will affect specific coastal locations. However, such information is vital, as places can be affected very differently depending on their location, nature, and orientation to the wind. The information is useful, for example, for coastal protection, planned levee construction, and secure port infrastructure.
Krieger’s team based their analysis on the hourly water levels of locations, which have been measured for decades. For instance, around 700,000 measured values have been collected for Cuxhaven since 1940. The researchers used these values, weather maps, and air pressure data to feed a statistical model with an algorithm that can learn by itself. The model learned only 80 percent of the data, with the rest remaining secret so that the model could be tested later. The team then linked 10-year forecasts to the AI model to obtain targeted forecasts for a specific location.
It took less than a second to calculate such a forecast. This is several hundred times faster than traditional climate models, which require a lot of computing time. The forecasts are expected to provide interesting results, particularly for the 2030s. This is because an internal climate fluctuation is still dampening the effects of the rise in sea level. It has a cycle of around 35 years. Krieger expects this to switch completely in a few years’ time, which could lead to higher storm surges locally. He wants to use the new method to determine where.
Original publication:
Krieger D, Weisse R, Baehr J, Borchert L (2025): Machine learning-driven skillful decadal predictions of the German Bight storm surge climate; Geophysical Research Letters. https://doi.org/10.1029/2024GL111558
Innovating and Cooperating for a Sustainable Future in a Digital AgeUniversity of Hamburg
