​

We applied machine learning to detect a special tectonic signal recorded by pressure sensors sitting on the seafloor. This signal represents the release of tectonic stress between earthquakes and thus their existence indicates a lower likelihood of future large earthquake and tsunamis. It is difficult to detect this signal because of the high noise. Here we show that machine learning successfully detected two such signals and three possible cases in data collected near New Zealand between 2014-2015. The method has the potential to transform our way of detecting such signals in seafloor pressure data offshore New Zealand and elsewhere, especially where the signal source is far away from the shoreline. 

​

Publication: He*, B., M. Wei, R. Watts, and Y. Shen (2020), Detecting Slow Slip Events from Seafloor Pressure Data Using Machine Learning, Geophysical Research Letters, accepted.