Mejora del sistema de alerta temprana de tsunamis en España mediante técnicas de Machine Learning

IP: Jorge Macías Sánchez

Acronym: UMA-CEIATECH-05

Abstract: The objective of this project is to improve the Spanish Tsunami Early Warning System (TEWS) managed by the IGN and of which the EDANYA group is a scientific advisor, and doing so using Machine Learning techniques. Currently this system uses the Tsunami-HySEA (TH) code for the direct simulation of hypothetical events that may affect Spain. For this, once a potentially tsunamigenic earthquake has been detected and the main characteristics such as magnitude, location, type of fault that generated it have been identified, we proceed to simulate the propagation of the tsunami, as well as the identification of the possible affected areas using the model TH. This system, although it has been a considerable advance with respect to the technology used up to now in TEWS, since it uses real-time simulations, does not consider the initial uncertainty associated with the identification of the earthquake that generated the event. In order to include this uncertainty, it would be necessary to simultaneously perform a large number of simulations that include the initial uncertainty. In order to perform this task, it is intended to design a trained neural network with the possible events that may affect the Spanish coasts simulated with TH and to use the predictions obtained with the network to be able to measure the initial uncertainty of the possible tsunamigenic events.