TEMPO.CO, Jakarta - Tropical storms are the most destructive natural disasters in the United States. Since 1980, storms have caused more than US$1.5 trillion in damages and claimed the lives of more than 7,000 people. The main factor behind this devastation is storm surge, the rise in sea level caused by a combination of strong winds, low air pressure within the storm, and breaking waves near the coast.
With technological advancements, artificial intelligence (AI) has the potential to improve the accuracy of storm surge forecasts. "Accurate storm surge predictions are critical for giving coastal residents time to evacuate and giving emergency responders time to prepare," as stated by Space on Monday, October 27, 2025.
Storm surges have historically been predicted using hydrodynamic models based on the physics of water flow. These models estimate wave heights and identify areas of risk by considering factors such as wind speed, storm direction, tide times, and the shape of the seabed and coastline.
Although modern computer performance is significantly faster, high-resolution simulations that determine details down to the level of residential environmental impact still require several hours. This time gap is critical for coastal residents needing to evacuate and for responders preparing for storm impacts.
To expedite the process, forecasters typically lower the simulation resolution. However, this leads to increased uncertainty regarding flood risks in specific areas. Some researchers have identified two main sources of uncertainty in storm surge predictions: the input data used in the model, such as storm wind patterns, and the computational grid resolution employed to calculate wave dynamics and water levels.
Here, the role of artificial intelligence becomes vital. AI models can produce detailed predictions faster. "Engineers and scientists have developed AI models based on deep neural networks that can predict water levels along the coastline quickly and accurately by using data about the wind field," as cited by Space.
In some cases, AI-based forecasts have proven to be more accurate than traditional hydrodynamic models. This technology is also capable of generating predictions in areas with limited historical data, or for unprecedented extreme conditions.
Artificial intelligence can be trained with synthetic data from physics models to address scenarios that may occur in the future. Once trained, AI models can rapidly produce storm surge forecasts based on air pressure and wind speed data.
Training AI with data from hydrodynamic models is also said to enhance its ability to create flood risk maps rapidly. This intelligent system can indicate which roads or houses are likely to be inundated during extreme events.
AI is now being used to a limited extent in storm surge forecasts to complement physics-based models. Several research teams are also developing AI applications to assess post-storm damage and process images from cameras to estimate flood intensities.
"As artificial intelligence models rapidly spread through every aspect of our lives and more data becomes available for training them, the technology offers potential to improve hurricane and storm surge forecasting in the future," according to the researchers.