← Back to Sessions

Exploring Paired Neural Networks to Rapidly Characterize Aftershock Events

Description: 

Mainshocks and their aftershock sequences occur unexpectedly and increase the burden of seismic analysts. Cross-correlation methods show promise for detecting aftershocks but require careful curation of template libraries, which is challenging for in-progress sequences. Also, cross-correlation methods are more computationally intensive, in comparison to use of pre-trained, deep-learning models. We explore the effectiveness of using a paired neural network (PNN) to identify similar earthquakes, such as aftershocks. The PNNs were trained to recognize similar earthquake pairs from a global dataset of real earthquakes augmented with different levels of background noise. We test the PNNs with a suite of closely-controlled test datasets, including a subset of high-quality aftershock templates with their cross-correlation matches (Emry et al., 2023), a dataset of highly similar event pairs from China (Schaff and Richards, 2021), and a dataset of synthetic waveforms created with a 1-d velocity model. This approach allows us to discern where the existing PNN models succeed and where they struggle, with the goal to understand whether PNN models may be trained differently to better generalize to in-progress aftershock sequences. This Ground-based Nuclear Detonation Detection (GNDD) research was funded by the U.S. Department of Energy. SNL is managed and operated by NTESS under DOE NNSA contract DE-NA0003525.

Session: Advancements in Forensic Seismology and Explosion Monitoring [Poster Session]

Type: Poster

Date: 5/2/2024

Presentation Time: 08:00 AM (local time)

Presenting Author: Erica

Student Presenter: No

Invited Presentation: 

Authors