Abstract
AbstractDetecting slow slip events (SSEs) at offshore subduction zones is important to understand the slip behavior on offshore subduction megathrusts, where tsunamis can be generated. The most widely used method to detect SSEs is to measure the vertical seafloor deformation caused by SSEs using seafloor pressure data. However, due to the small signal‐to‐noise ratio and instrumental drift, such detection is very difficult. In this study, we trained a machine learning model using synthetic data to detect SSEs and applied it to real pressure data in New Zealand between 2014 and 2015. Our method detected five events, two of which are confirmed by the onshore GPS records. Besides, our model performs better than the traditional matched filter method. We conclude that machine learning could be used to detect SSEs in real seafloor pressure data. The method can be applied to other regions, especially where near trench GPS is not available.
Highlights
Our ability to forecast catastrophic earthquakes and tsunamis on offshore subduction zone such as the 2004 Sumatra and 2011 Japan earthquakes remains weak (Burgmann & Chadwell, 2014)
The results show that machine learning methods are promising as a means to detect SSEs in seafloor pressure data and could be useful to detect SSEs in other subduction zones, especially where the onshore GPS is too far away from the trench
We combined convolutional neural network (CNN) and recurrent neural network (RNN) to train the model and we showed RNN is very useful in this study (Figure S2)
Summary
Our ability to forecast catastrophic earthquakes and tsunamis on offshore subduction zone such as the 2004 Sumatra and 2011 Japan earthquakes remains weak (Burgmann & Chadwell, 2014). One main limitation is our lack of ability to measure, detect, and quantify small and slow tectonic deformation including SSEs at the seafloor (Burgmann & Chadwell, 2014). This situation is changing as several large projects using seafloor pressure sensors have been conducted in the last decade in Japan (DONET, S-NET), Cascadia (Cascadia Initiative, OOI), New Zealand (HOBITSS), and Alaska (AACSE). Kong et al, (2019) and Bergen et al, (2019) reviewed recent applications of machine learning in seismology and solid earth geoscience, respectively Among these studies, convolutional neural network (CNN) plays an important role. The results show that machine learning methods are promising as a means to detect SSEs in seafloor pressure data and could be useful to detect SSEs in other subduction zones, especially where the onshore GPS is too far away from the trench
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
