Several seismic precursors and earthquake-causing variables have been proposed in the last decades based on physical considerations and case observations, however none has been confirmed on long datasets using linear analysis. This work adopts an information-theoretical approach to investigate the occurrence of causal flow between these precursors and causing variables and seismicity. It starts by introducing the key concepts in seismology and presenting the current main precursor candidates. Four variables will be considered as possible precursors or anomalies leading to earthquakes: large tidal amplitudes, temporal fluctuations in the Gutenberg-Richter's b-value, surface gravity changes, and preceding anomalous seismicity patterns. To perform the causality test between these variables and their effects, it is developed a method which allows the fast calculation of Transfer Entropy for any two time-series, detecting the direction of the flow of information between the variables of interest. The method is tested to coupled logistic maps and networks with different topologies before application to geophysical events. The analysis shows mutual information relating to coupling strength and also allows inference of the causal direction from data using the Transfer Entropy, both in bivariate systems and in networks. The method was then applied to the earthquake analysis for an interval of 4018 days on an area comprising the Japan trench. Within a conservative margin of confidence, the results could not at this point confirm any of the four precursor options considered, but future work can clarify initial suggestions regarding tidal amplitudes link to seismicity, and pre-seismic gravity changes and cumulative daily magnitude anomalies. The Matlab/Octavecodes for our method are open-source and available at https://github.com/artvalencio/causality-toolbox We hope the method is able to support the quest for other precursor candidates, and to assist other fields of knowledge.
Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:752655 |
Date | January 2018 |
Creators | Lopes da Silva Valencio, Arthur |
Publisher | University of Aberdeen |
Source Sets | Ethos UK |
Detected Language | English |
Type | Electronic Thesis or Dissertation |
Source | http://digitool.abdn.ac.uk:80/webclient/DeliveryManager?pid=237105 |
Page generated in 0.002 seconds