Scientists have developed a system that can estimate the magnitude and location of a powerful earthquake almost immediately after it occurs, based on extremely weak changes in the Earth's gravitational field. Unlike traditional methods, the technology uses a signal that appears before the seismic waves, and artificial intelligence helps to extract from it the necessary information to assess the scale of the event. . The development has already gone beyond the limits of the laboratory experiment. the researchers created an application that allows the algorithm to be directly integrated into the SeisComP seismic monitoring system. The technology is being tested and implemented in various countries, including tsunami early warning systems in Peru and Alaska. Quentin Blettery, a researcher at the "Geoazur" laboratory of the French Development Research Institute (IRD), presented the principle of the system's work and the results of its tests. The system should solve the problem of the first minutes after the earthquake According to him, in the event of a strong underwater earthquake, specialists have a relatively long time before the tsunami arrives. depending on the distance from the shore, the wave can reach the shore in one or two times An hour later. However, the decision about the need for a warning should be made much earlier. The main parameter here is the magnitude of the earthquake. According to the expert, the possible size of the tsunami largely depends on the scale of the event itself. The problem is that traditional early warning systems estimate an earthquake by the first seismic waves. For normal events this approach works well, but for extremely large earthquakes a saturation effect occurs; the first data may give a significantly reduced estimate of the magnitude. Blatter cited the 2011 magnitude 9 earthquake in Japan as an example. Such an earthquake does not develop instantly. its magnitude is formed in about two minutes. At the same time, Japan's early warning system initially underestimated the magnitude of the event and actually stopped at about 8 magnitude. This is a fundamental difference for tsunami forecasting. According to the researcher, in this case, there can be a 30-fold difference in the amplitude of the tsunami between earthquakes of magnitude 8 and 9. That is why it is especially important for early warning systems to get information about the true scale of the largest earthquakes as soon as possible. Gravity provides information earlier than seismic waves After the 2011 Tohoku earthquake, researchers turned their attention to the so-called instantaneous elastic-gravity signal, PEGS. When a huge earthquake occurs, huge masses of rock are moved considerable distances. This causes a very small change in the Earth's gravitational field. The signal is so weak that its magnitude is about 1 nanometer per second squared. However, it can be recorded by seismological devices, because seismographs measure acceleration, and the gravitational effect also manifests itself as acceleration. The main advantage of PEGS is the speed of propagation. A gravitational disturbance travels at the speed of light, while seismic waves travel through the Earth much more slowly. This creates a unique time window. information about changes in the gravitational field can be obtained before the arrival of seismic waves. At the same time, the signal contains information not only about the fact of the earthquake, but also about its scale. Bleteri showed that the calculated signals for earthquakes of magnitude 8.5 and 9 are noticeably different. In theory, this allows you to determine how big an event is before traditional seismic information gives the full picture. Created thousands of virtual earthquakes for AB The main technical problem was that the gravitational signal itself is too weak. To teach the algorithm to recognize it, the researchers created a large number of virtual earthquakes at different points in the subduction zones of Japan. For each scenario, synthetic signals were calculated that the seismic network would record if such an earthquake were to occur. Then real seismic noise collected during a year of observations was added to that data. Thus, the scientists received an artificial data set, which is as close as possible to real monitoring conditions. A neural network was trained on this data, which determined the magnitude and location of the earthquake based on the characteristic distribution of the signal in the seismic network. Later, researchers switched from convolutional neural network to graph neural network (GNN). In this case, the algorithm takes into account not only the signals themselves, but also the geometrical position of the seismic stations relative to each other. This has been particularly important for smaller earthquakes, where the gravitational signal is weaker. The technology is already being tested in various countries The method was tested in autonomous mode on data from Japan, Chile and Alaska. Now the algorithm is also being piloted in Peru's early warning system, is being used in work on the Alaska system, and has begun to be used in New Caledonia. In Alaska, the system was additionally taught to determine the moment tensor of an earthquake, a parameter that allows to determine the geometry of the fault and the nature of the movement of the earth's crust. This is of direct importance for tsunami assessment. Two earthquakes of the same magnitude do not necessarily create a tsunami of the same size. a lot depends on how the earth's crust shifted. In experiments, the algorithm worked well for earthquakes of magnitude 8 and above. For weaker events, the gravitational signal became too small. However, the use of a graph neural network allowed to increase the stability of the results and expand the working range of the system. The algorithm can be integrated into an existing monitoring system The researchers' next step is to make the technology part of a real seismological monitoring infrastructure. According to Bleteri, an application was created for this, which allows connecting the algorithm directly to the SeisComP platform used by professionals. So it is no longer just about a neural network working in a research environment. The developers are creating a tool that can be integrated into existing monitoring centers and used together with traditional methods for determining earthquake parameters. The task of the system is to obtain an additional estimate of the magnitude and location of a strong earthquake almost immediately after its occurrence, and later to use this information to determine the need for a tsunami warning. When deciding. At the same time, the new technology is not designed to replace traditional seismic systems. Its task is to give specialists an additional source of information at the very moment when conventional methods still cannot show the true scale of a super-large earthquake. If further tests confirm the results, the gravity signal in combination with AB could become an additional level of tsunami early warning.