Speaker
Fedor Vinberg
((IEPT RAS / MSUIECS) International Institute of Earthquake Prediction Theory and Mathematical Geophysics)
Description
The algorithm put forward in this paper is designed for devices that provide ultrashort warning of earthquakes. It is based on the time difference between the arrivals of P and S waves. The difference is small, but rapid identification of a P wave arrival from a large earthquake is useful for mitigating the impact of such an earthquake, in particular, human losses. The method we propose is based on neural network recognition and can deal with the problem as stated above rather accurately and rapidly. Indeed, the training of a local network by local data (in contrast to the training using a "universal" sample) can significantly reduce the warning error. This study demonstrates the point by a Japanese example.
Primary author
Fedor Vinberg
((IEPT RAS / MSUIECS) International Institute of Earthquake Prediction Theory and Mathematical Geophysics)
