Magnitude Estimation for Earthquake Early Warning Using a Deep Convolutional Neural Network

Abstract

Magnitude estimation is a vital task within earthquake early warning (EEW) systems (EEWSs). To improve the magnitude determination accuracy after P-wave arrival, we introduce an advanced magnitude prediction model that uses a deep convolutional neural network for earthquake magnitude estimation (DCNN-M). In this paper, we use the inland strong-motion data obtained from the Japan Kyoshin Network (K-NET) to calculate the input parameters of the DCNN-M model. The DCNN-M model uses 12 parameters extracted from 3 s of seismic data recorded after P-wave arrival as the input, four convolutional layers, four pooling layers, four batch normalization layers, three fully connected layers, the Adam optimizer, and an output. Our results show that the standard deviation of the magnitude estimation error of the DCNN-M model is 0.31, which is significantly less than the values of 1.56 and 0.42 for the τc method and Pd method, respectively. In addition, the magnitude prediction error of the DCNN-M model is not affected by variations in the epicentral distance. The DCNN-M model has considerable potential application in EEWSs in Japan.

Rapid earthquake magnitude estimation combining a neural network and transfer learning in China: Application to the 2022 Lushan M6.1 earthquake

“…Meanwhile, some researchers have used neural networks to establish magnitude estimation models [17][18][19]. [20] used a deep convolutional neural network (CNN) to establish a magnitude estimation model (DCNN-M model) based on the strong-motion data recorded by the Japan Kyoshin Network (K-NET) stations.…”

CREIME—A Convolutional Recurrent Model for Earthquake Identification and Magnitude Estimation

“…Unlike Zhu et al. (2021) where a set of 12 features extracted from 3 s of data is used to perform magnitude estimation and Ochoa et al. (2018) where features extracted from 5 s of P wave data are fed to a Support Vector Machine Regression algorithm to estimate magnitude, CREIME is end‐to‐end using a combination of Convolutional and Recurrent neural network to extract features directly from the raw waveform.…”

“…In this paper, we present a novel approach to achieve multi-tasking Convolutional Recurrent model for Earthquake Identification and Magnitude Estimation (CREIME), which can simultaneously perform earthquake identification, local magnitude estimation, and first P-wave arrival time regression solely based on 1-2 s P-wave recording. Unlike Zhu et al (2021) where a set of 12 features extracted from 3 s of data is used to perform magnitude estimation and Ochoa et al (2018) where features extracted from 5 s of P wave data are fed to a Support Vector Machine Regression algorithm to estimate magnitude, CREIME is end-to-end using a combination of Convolutional and Recurrent neural network to extract features directly from the raw waveform. Unlike traditional approaches such as Nakamura (1988), Allen and Kanamori (2003), and Wu and Zhao (2006) no a-priori information on hypocentral distance is required for magnitude estimation.…”

CREIME: A Convolutional Recurrent model for Earthquake Identification and Magnitude Estimation

“…In this paper we present a novel approach to achieve multi-tasking Convolutional Recurrent model for Earthquake Identification and Magnitude Estimation (CREIME), which can simultaneously perform earthquake identification, local magnitude estimation and first P-arrival time regression solely based on 1-2 seconds P-wave recording. Unlike [47] which uses a set of twelve features extracted from 3 seconds of data to perform magnitude estimation, CREIME is end-to-end using a combination of Convolutional and Recurrent neural network to extract features directly from the raw waveform. The motivation for using such a small duration of P-wave data lies in its potentially easier utility in applications such as rapid earthquake characterisation for EEW systems ( [16] [48] and references therein).…”