TY - GEN

T1 - ConvLSTM Neural Networks for seismic event prediction in Chile

AU - Fuentes, Alex Gonzalez

AU - Nicolis, Orietta

AU - Peralta, Billy

AU - Chiodi, Marcello

N1 - Publisher Copyright:
© 2021 IEEE.

PY - 2021/8/5

Y1 - 2021/8/5

N2 - Predicting seismic risk is a challenging task in order to avoid catastrophic effects. In this work, two models based on Convolutional Network (CNN) and Long Short Term Memory (LSTM) networks are proposed to predict the seismic risk in Chile. In particular, a ConvLSTM and a Multi-column ConvLSTM network are used for the prediction of the average number of seismic events greater than 2,8 magnitude on the Richter scale, in the Chilean regions of Coquimbo and Araucania between the years 2010 and 2017. For this model, the values of the intensity function estimated through an ETAS model and the accumulated displacement prior to a the seismic events are used as inputs. In particular, given the spatial and temporal characteristics of the seismic data, matrices of size 20x20 of the last 20 days are considered to predict the average number of seismic events of the next day in a given area. From the results obtained, the Multi-column ConvLSTM network achieved a coefficient of determination of 0,804 and a lower MSE than other networks.

AB - Predicting seismic risk is a challenging task in order to avoid catastrophic effects. In this work, two models based on Convolutional Network (CNN) and Long Short Term Memory (LSTM) networks are proposed to predict the seismic risk in Chile. In particular, a ConvLSTM and a Multi-column ConvLSTM network are used for the prediction of the average number of seismic events greater than 2,8 magnitude on the Richter scale, in the Chilean regions of Coquimbo and Araucania between the years 2010 and 2017. For this model, the values of the intensity function estimated through an ETAS model and the accumulated displacement prior to a the seismic events are used as inputs. In particular, given the spatial and temporal characteristics of the seismic data, matrices of size 20x20 of the last 20 days are considered to predict the average number of seismic events of the next day in a given area. From the results obtained, the Multi-column ConvLSTM network achieved a coefficient of determination of 0,804 and a lower MSE than other networks.

KW - Deep learning

KW - ETAS

KW - prediction

KW - seismic risk

UR - http://www.scopus.com/inward/record.url?scp=85116274852&partnerID=8YFLogxK

U2 - 10.1109/INTERCON52678.2021.9532946

DO - 10.1109/INTERCON52678.2021.9532946

M3 - Conference contribution

AN - SCOPUS:85116274852

T3 - Proceedings of the 2021 IEEE 28th International Conference on Electronics, Electrical Engineering and Computing, INTERCON 2021

BT - Proceedings of the 2021 IEEE 28th International Conference on Electronics, Electrical Engineering and Computing, INTERCON 2021

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 28th IEEE International Conference on Electronics, Electrical Engineering and Computing, INTERCON 2021

Y2 - 5 August 2021 through 7 August 2021

ER -