Sistema automático para la detección y localización de terremotos en tiempo real con redes neuronales profundas.

The instrumental coverage to detect seismotectonic activity that the seismological observatories
of Costa Rica have installed has increased significantly in recent years. This coverage has been
driven by mobile sensors that can be easily deployed and relocated. The result of the extensive
coverage has allowed the development of seismic models for disaster prevention and a better
understanding of different tectonic phenomena. However, there is still a fundamental challenge in
processing the entire volume of data in real time and with a high degree of accuracy. Fortunately,
recent developments in computer science offer a tremendous opportunity to meet that challenge.
First, computing architectures continue to grow in computing power, particularly those based on
graphics processing cards and multicore processors. Second, the latest deep neural network
models allow us to solve complex processes in nature and make modeling and prediction with
high precision. This project will delve into the inspection of deep neural network models for the
detection and localization of earthquakes in real time. The impact of this system will be immediate
for the work carried out by OVSICORI-UNA in the administration of the country's seismic risk.

Consolidar el procesamiento automático de registros sismológicos en tiempo
real a través de redes neuronales profundas

Sismología Computacional