Artificial Intelligence Algorithm Helps Predict Dynamics of Breaking Waves

According to a recent scientific study published in Nature Communications, researchers have developed a new method to simulate the dynamic changing of breaking waves using a machine learning algorithm known as a neural network with test data from wave tanks. Accurate predictions of wave interactions can help optimize the design of offshore structures, as well as improve understanding of how the ocean interacts with the atmosphere.

Once the waves swell to a critical height, they break and fall into droplets and bubbles at the crest. For decades, the dynamics of how and when waves break have been too complex for scientists to predict. The mathematical equations of traditional calculation methods cannot capture the complexity of breaking waves.

In this new study, the researchers started with a set of equations believed to be a standard description of wave behavior and improved the model by “training” it on wave-breaking data obtained in practical experiments. Overall, the team used data obtained from about 250 experiments to train a neural network machine-learning algorithm comparing the actual waves in the experiments with those predicted in a simple model. Depending on the difference between the two, the algorithm adjusted the model to simulate reality. The researchers found that the model, modified by the machine learning algorithm, could more accurately predict how and when the waves would break. Compared to traditional wave equations, the model provides a more accurate estimate of a wave’s steepness before its break, as well as its energy and frequency after breaking.

Given the entire ocean area, the researchers believe breaking waves play a fundamental role in climate predictions. For example, with better estimates of how waves break up, it will be possible to predict how much carbon dioxide and other atmospheric gasses the oceans can absorb. The team’s updated wave model comes in the form of open-source code that can also be used for simulated testing of offshore platforms and coastal structures.