A stress test to evaluate the usefulness of Akaike information criterion in short-term earthquake prediction

Roberta Tozzi,Michael Pezzopane,Fabrizio Masci

doi:10.1038/s41598-020-77834-0

Roberta Tozzi, Michael Pezzopane + Show 1 more

Open Access

https://doi.org/10.1038/s41598-020-77834-0

Copy DOI

Abstract

Akaike information criterion (AIC) has been recently adopted to identify possible earthquake precursors in ionospheric total electron content (TEC). According to the authors of this methodology, their technique allows finding abrupt increases (positive breaks) in vertical TEC rate of change 25–80 min before the occurrence of large earthquakes, highlighting a promising implication of AIC method in Mw > 8 earthquakes alert strategies. Due to the relevance of this matter, a lively scientific debate ensued from these results. In this study, we carefully evaluate AIC method potentiality in searching earthquake TEC precursory signatures. We first investigate the dependence of the detected breaks number on the adjustable AIC method parameters. Then, we show that breaks occurrence clusters around specific local times and around moderate and high solar and geomagnetic activity. The outcome of this study is that AIC method is not concretely usable for issuing large earthquakes alerts.

Highlights

Akaike information criterion (AIC) has been recently adopted to identify possible earthquake precursors in ionospheric total electron content (TEC)
Considering the extreme relevance and complexity of this matter, and the potential societal relevance of the results found by Heki and Enomoto[27], i.e. the possibility to be able to detect signs of an incoming strong earthquake in real-time monitored vertical TEC (vTEC) data, is of uttermost importance to rigorously check the reliability of their results
Open questions on AIC method, to which we attempt to answer, concern: 1) the compatibility of breaks frequency with alert issuing for incoming strong earthquakes, 2) the possibility to univocally set AIC method parameters and 3) the possibility to relate detected breaks to sources other than earthquakes