Article
The Seismo-Performer: A Novel Machine Learning Approach
for General and Efficient Seismic Phase Recognition from
Local Earthquakes in Real Time
Andrey Stepnov
1,
* , Vladimir Chernykh
2
and Alexey Konovalov
1
Citation: Stepnov, A.; Chernykh, V.;
Konovalov, A. The Seismo-Performer:
A Novel Machine Learning Approach
for General and Efficient Seismic
Phase Recognition from Local
Earthquakes in Real Time. Sensors
2021, 21, 6290. https://doi.org/
10.3390/s21186290
Academic Editor: Nunzio Cennamo
Received: 21 August 2021
Accepted: 16 September 2021
Published: 19 September 2021
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2021 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
1
Far East Geological Institute, Far Eastern Branch, Russian Academy of Sciences, 690022 Vladivostok, Russia;
director@fegi.ru
2
Khabarovsk Federal Research Center, Far Eastern Branch, Russian Academy of Sciences,
680000 Khabarovsk, Russia; admvc@ccfebras.ru
* Correspondence: a.stepnov@geophystech.ru; Tel.: +7-914-757-66-27
Abstract:
When recording seismic ground motion in multiple sites using independent recording
stations one needs to recognize the presence of the same parts of seismic waves arriving at these
stations. This problem is known in seismology as seismic phase picking. It is challenging to automate
the accurate picking of seismic phases to the level of human capabilities. By solving this problem, it
would be possible to automate routine processing in real time on any local network. A new machine
learning approach was developed to classify seismic phases from local earthquakes. The resulting
model is based on spectrograms and utilizes the transformer architecture with a self-attention
mechanism and without any convolution blocks. The model is general for various local networks
and has only 57 k learning parameters. To assess the generalization property, two new datasets
were developed, containing local earthquake data collected from two different regions using a wide
variety of seismic instruments. The data were not involved in the training process for any model to
estimate the generalization property. The new model exhibits the best classification and computation
performance results on its pre-trained weights compared with baseline models from related work.
The model code is available online and is ready for day-to-day real-time processing on conventional
seismic equipment without graphics processing units.
Keywords:
seismogram; spectrogram; transformer; attention; CNN; deep learning; seismic phase;
real-time automation; classification; computational efficiency; local seismic network
1. Introduction
Phase picking is a routine task in the processing of local seismological monitoring
data. The complete automation of this task has become increasingly important, especially
in connection with the growth of seismic networks with inexpensive instruments and the
increase in the number of Internet of Things (IoT) devices [1,2].
Phase picking automation is a challenging problem. The number of lower magnitude
earthquakes has grown exponentially [
3
]; however, the amplitudes of many earthquake
signals are weakened to the level of seismic noise or less with decreasing earthquake magni-
tudes. Improving the completeness of the earthquake magnitude catalog is a central goal of
local seismological monitoring since a comprehensive catalog provides more information
about the seismic regime.
Another issue is the configuration of the seismic network. A wide variety of sensor
types, site soil conditions, and levels of seismic noise can exist inside a single network. Of
course, this can differ from one network to the next. Consequently, the registered shape of
a seismic signal can vary significantly, and a general algorithm is needed to address the
phase picking task in a manner meeting or exceeding human effort for data coming from
any local seismic network.
Sensors 2021, 21, 6290. https://doi.org/10.3390/s21186290 https://www.mdpi.com/journal/sensors
