Each year in Southern California, roughly 10,000 earthquakes are detected. The earthquakes are much more common than people think. They are occurring more frequently in the central United States probably due to the wastewater injection. While a large percentage of these earthquakes is not felt, it is important that we still have some understanding of them. By gaining an understanding and knowledge we can determine what is causing those earthquakes and can create tools for prevention as well.
This can also help in determining the future of earthquakes of all sizes and can improve the way communities will respond to those earthquakes. A research team led by deep learning scientist Thibaut Perol has discovered a way to use the artificial intelligence (AI) to improve the earthquake detection. The team named their AI system ConvNetQuake. It is the first neural network which is designed to detect and locate earthquakes. The specialized algorithm looks at the ground motion measurements called the seismograms. After observing the motions it will be able to tell whether the ground movement is just noise or actual earthquake. In the past, it has been difficult to detect minor earthquakes due to the noise but it will be possible to distinguish them with ConvNetQuake.
The researchers also used the seismic activity in Oklahoma to test and train ConvNetQuake. In their observations, they found that the earthquakes detected were 17 times more than there were recorded in the Oklahoma Geological Survey earthquake catalog. This particular study shows how the system can be improved to detect the earthquakes within the central United States. ConvNetQuake can be applied beyond this region. By effectively detecting and categorizing the seismic activity the information on earthquake’s life can also be calculated. This way the communities in that area can be moved depending on how quick the earthquake is picking up steam and how severe it might be.
While ConvoNetQuake is superior to other earthquake detection methods still it can only detect the earthquakes and cannot predict them in advance. However, early detection is still best for us since it can also help to save lives if the area is evacuated in time.