NORMA eResearch @NCI Library

Seismic Phase Detection & Picking using EfficientNet

Ramakrishnan Arularasan, Arunprasath (2022) Seismic Phase Detection & Picking using EfficientNet. Masters thesis, Dublin, National College of Ireland.

[thumbnail of Master of Science]
PDF (Master of Science)
Download (1MB) | Preview
[thumbnail of Configuration manual]
PDF (Configuration manual)
Download (989kB) | Preview


The monitoring of seismic waves for the detection of earthquakes and the picking of the arrival of P and S waves has been a challenging task in the field of observational seismology. While the use of deep learning techniques has led to improved performance, the models tend to suffer from poor generalizability and poor picking performance with the S waves. This research proposes the use of EfficientNet architecture on the spectrogram and waveform plots of the signals for phase detection and phase picking. The performance of the model was evaluated using the Italian Seismic dataset. The model was also trained using the Stanford Earthquake dataset for comparing it with the existing models. The EfficientNet baseline model was outperformed by a simple CNN architecture on the Italian Seismic dataset. When trained on the Stanford dataset, the EfficientNet model had an F1 score of 0.95 which is slightly lesser than the existing models. In phase regression, the EfficientNet architecture had poor picking performance compared to the existing models. While the models trained on the Stanford dataset had a higher accuracy compared to the Italian dataset, both the models suffered with poor generalizability when tested with waveforms from different regions.

Item Type: Thesis (Masters)
Subjects: Q Science > QA Mathematics > Electronic computers. Computer science
T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science
Q Science > QC Physics
Q Science > QE Geology
Divisions: School of Computing > Master of Science in Data Analytics
Depositing User: Tamara Malone
Date Deposited: 01 Mar 2023 15:25
Last Modified: 01 Mar 2023 15:25

Actions (login required)

View Item View Item