Recurrent Convolutional Neural Networks help to predict location of Earthquakes
This work addresses earthquake prediction for disaster preparedness, but it is incremental as it builds on existing neural network architectures with modest improvements over baseline models.
The paper tackles the problem of midterm earthquake prediction by developing a recurrent convolutional neural network model to forecast earthquakes above magnitude 5 in specific areas within 10-60 days, achieving a ROC AUC of 0.975 and PR AUC of 0.0890, which outperforms a baseline in reducing false alarms from 1,004,000 to 192,000 while maintaining similar correct predictions.
We examine the applicability of modern neural network architectures to the midterm prediction of earthquakes. Our data-based classification model aims to predict if an earthquake with the magnitude above a threshold takes place at a given area of size $10 \times 10$ kilometers in $10$-$60$ days from a given moment. Our deep neural network model has a recurrent part (LSTM) that accounts for time dependencies between earthquakes and a convolutional part that accounts for spatial dependencies. Obtained results show that neural networks-based models beat baseline feature-based models that also account for spatio-temporal dependencies between different earthquakes. For historical data on Japan earthquakes our model predicts occurrence of an earthquake in $10$ to $60$ days from a given moment with magnitude $M_c > 5$ with quality metrics ROC AUC $0.975$ and PR AUC $0.0890$, making $1.18 \cdot 10^3$ correct predictions, while missing $2.09 \cdot 10^3$ earthquakes and making $192 \cdot 10^3$ false alarms. The baseline approach has similar ROC AUC $0.992$, number of correct predictions $1.19 \cdot 10^3$, and missing $2.07 \cdot 10^3$ earthquakes, but significantly worse PR AUC $0.00911$, and number of false alarms $1004 \cdot 10^3$.