Researchers Used AI to Develop Early Warning Systems for Tsunamis

By Pranav Mahapatra

April 26, 2023
landscape with green trees and a white car drowning in a tsunami

Researchers have developed a new method for rapidly estimating the properties of tectonic events and predicting tsunami wave heights in near-real-time.

A new study published in the Journal Physics of the Earth and Planetary Interiors, Numerical validation of an effective slender fault source solution for past tsunami scenarios1, presents a method that can provide early estimations of a tectonic event’s properties and the resulting tsunami’s characteristics.

The approach relies on analyzing acoustic waves generated by the tectonic event to infer its characteristics and inputting those into a numerical model to predict the tsunami’s propagation.

To capture the sound waves generated by the earthquakes, researchers used recordings from underwater microphones, also known as hydrophones.

Tsunamis pose a significant threat to coastal communities worldwide, early warning systems that can accurately predict the height and arrival time of a tsunami wave can help mitigate the damage caused by these events.

Using Acoustic Waves to Detect Tsunamis

Acoustic waves are sound waves that propagate through the Earth’s crust and the ocean, and they can provide valuable information about the source of tectonic events.

When a tectonic event occurs, it generates acoustic waves that propagate through the Earth’s crust and the ocean.

The method developed by the researchers relies on analyzing the acoustic waves generated by a tectonic event to infer its properties, including the location, orientation, and size of the fault responsible for the event.

Using a technique called Inverse Problem Analysis, the researchers were able to estimate the effective rectangular ground area displacement caused by the tectonic event, which provided information on the characteristics of the fault.

Simulating Tsunami Waves with Computer Models

Researchers used the estimated fault characteristics as input for a numerical model called COMCOT or Cornell Multi-grid Coupled Tsunami model, which simulates the propagation of water waves generated by tectonic events.

It can simulate both linear and non-linear effects of water waves and predicts the tsunami wave’s arrival time and height at specific locations.

The rectangular fault was used as a source descriptor in the model because it is a simple geometry and allows researchers to relate it to the recorded acoustic signals.

This makes it easier to keep the computational costs low while simulating the propagation of water waves generated by tectonic events.

The model has been developed by researchers to take into account various factors that affect the propagation of tsunami waves, including the depth of water, the shape of the coastline, and the presence of other sources of energy.

Using the input parameters calculated by the Inverse Problem Analysis technique, the model is able to accurately predict the arrival time and height of the tsunami wave at specific locations.

Testing the Model’s Accuracy with Real-World Data

Researchers validated the accuracy of the model by comparing its predicted wave heights and arrival times with real-world data recorded by DART buoys.

Advertisements

The comparison evinced that the predicted wave heights and arrival times were in the same order of magnitude as the recorded values.

The partial agreement between the predicted and actual dynamics suggests that the model could be used as a complementary tool for tsunami warning systems to provide early estimations on submarine tectonic event properties in almost real time.

However, the researchers observed some discrepancies in the maximum wave amplitudes, which were overestimated in some cases, resulting in larger tsunamis.

They speculate that the discrepancies could be due to the inaccuracy of the ML algorithm in estimating the source parameters or the neglecting of the elasticity of the ground and the effects of gravity, which can play a significant role in tsunami propagation.

The researchers suggested that further improvements in the direct problem’s realistic description, including the inclusion of elasticity in the seabed, gravity effects, and the consideration of real bathymetry, are necessary to achieve a more accurate description of the tectonic events’ acoustic signals.

Future Applications and Limitations of the Model

The researchers acknowledge that there are discrepancies between the predicted and actual values, which could be due to neglecting the elasticity of the ground and the effects of gravity in the numerical model.

They suggest that including these effects and improving the accuracy of the source parameter estimation through machine learning algorithms could lead to more accurate results.

Nevertheless, the method provides a complementary tool for tsunami warning systems that can provide early estimations of tectonic event properties in near-real-time.

While there are still areas for improvement, the method provides a valuable tool for early warning systems that can help mitigate the damage caused by tsunamis.

References

  1. Bernabe Gomez, and Usama Kadri, ‘Numerical validation of an effective slender fault source solution for past tsunami scenarios‘, American Institute of Physics (AIP), 25 April 2023, https://pubs.aip.org/aip/pof/article/35/4/046113/2885298/Numerical-validation-of-an-effective-slender-fault[]
Share: