Scientists are increasingly using the combination of seismic information with machine learning techniques to deepen earthquake prediction.
In recent years, the prediction of an earthquake has become a much more viable concept. These phenomena are traditionally characterized by their unpredictability. However, researchers do not stop working for be able to obtain information that allows some preventioneven if it’s last-minute.
Knowing with some margin when an earthquake is going to take place would be extremely useful to be able to mitigate its effects. And this is where the artificial intelligence, more specifically, machine learning. The scientific journal Seismological Research Letters has published a section showing the advances being made for the prediction of earthquakes using these techniques.
The idea is to better understand the patterns that are triggered in an earthquake, both before and during. In this way it seeks to be able to make predictions about the seismic activity of an area, or identify the epicenter of the phenomenon. It is also intended to catalog different types of seismic waves and distinguish seismic activity from other types of unrelated movements.
Of course, artificial intelligence has long been explored to calculate when an earthquake will come. The creation of algorithms based on this technology and the growing volume of datafrom the seismological records, they have given rise to this.
Food is the data
There is a key issue in this work of achieving earthquake prediction. This is analog information. Although it is now when more records are kept and when more complete they are. But, over decades, information has been collected in analog formats. The challenge is to digitize this large collection to be able to introduce it into algorithms and, above all, to prevent it from being lost.
Machine learning can also help in this work, as it is able to identify and categorize images. In this way, it would not be necessary to digitize the entire archive, but only the parts that actually provide useful information.
Efforts in earthquake prediction have crystallized in several applications, whose creators claim to allow some anticipation. Although MyShake, which aims to create a seismic detection network, or other apps, which try to get 30 or 100 seconds ahead of the earthquake, are only anticipations of future more refined tools. Everything moves in the same direction: know when an earthquake will occur.
Images: IFRC, Angelo Giordano
