Tectonic movements create stress within the Earth’s crust, which gets released in sudden earthquakes, but also in less dramatic slow slip events. Such events are sometimes accompanied by so-called non-volcanic tremors – weak seismic signals of extended duration along major faults. Considerable research effort has gone into classifying the spatio-temporal patterns of such tremors, which may be useful in predicting future slow slip events. In a new paper, LML Fellow Jiancang Zhuang and colleagues analyse data for tremors in the Kii and Shikoku regions of Japan using a two-dimensional hidden Markov model, finding that the activity shows a hierarchical spatial structure, and several distinct types of spatial segments with unique tremor signatures and probabilities of tremor occurrence.
As they show, tremor activity in each region can be grouped into spatial subsystems, with activity tending to migrate more frequently between spatial segments within each subsystem than from one subsystem to another. Moreover, the occurrence patterns also place tremors into three distinct types: episodic segments featuring high tremor activity, weak concentration segments with either low tremor activity or very short average sojourn times, and background activity either spanning across a large spatial region or having very long average sojourn times. By comparing the temporal variation of the three types of tremor with the time periods during which short-term slow slip events were detected, the study shows that tremor from weak concentration segments often precedes or follows short-term slow slip events, but rarely occurs during the time period of slow slip. This observation provides a good basis for forecasting future short-term slow slip events.
The hidden Markov model explored here offers an efficient alternative to laborious manual analysis of tremor data, enabling more effective categorization than was previously possible. The researchers expect that it will be used to produce improved probabilistic forecasts of future tremor activity.
The paper is available at https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2017JB015360 or by contacting the Jiancang Zhang directly (firstname.lastname@example.org).
LML is a charity. If you enjoyed reading this, please consider supporting us.