New prognostic technology for analysis of low-frequency seismic noise variations (on the example of the Russian Far East)

Abstract

A new technology for predicting strong earthquakes with a magnitude range of Mw about 7 and more is considered, based on the use of continuous recordings of seismic noise on a network of 21 broadband stations of the GS RAS in the region of the Kamchatka Peninsula, the Commander Islands and the Paramushir Island. The article is described a forecasting algorithm created by A.A. Lyubushin, IPE RAS, and the state of its implementation in the Kamchatka Division GS RAS for the purpose of an advance (months - first years) assessment of the strong earthquakes preparation sites. The data processing algorithm includes the calculation of four noise statistics time series for each station and the construction of their spatial distribution maps for different time intervals. We used four noise statistics, including the minimal entropy of the orthogonal wavelet coefficients squares and three characteristics of the multifractal spectrum of singularity – the generalized Hurst exponent, the carrier width, and the spectral wavelet exponent. Based on previous research, characteristic features of the four seismic noise statistics behavior at preparation stages of the local earthquakes 2013-2016 with Мw=6.6-8.3 were revealed, corresponding to similar changes before the two earthquakes with Мw=8.3 and 9.0 in Japan. It was found that an increase in the danger of a strong earthquake is accompanied by an increase in minimal entropy and a decrease in the carrier width and other parameters of the singularity spectrum. Since 2020, the processing of current data from the network of broadband stations of the GS RAS in the Far East region has been carried out in accordance with the seismic forecasting algorithm for drawing up quarterly forecast conclusions, which are sent to the Russian Expert Council on Earthquake Forecasting, Seismic Hazard and Risk Assessment (REC) and to Kamchatka Branch of REC

Variability in the Statistical Properties of Continuous Seismic Records on a Network of Stations and Strong Earthquakes: A Case Study from the Kamchatka Peninsula, 2011–2021

“…Since 2021, the authors have been implementing this method with the issuance of quarterly forecast conclusions about the danger of strong earthquakes to the Russian Expert Council for Earthquake Prediction and its Kamchatka branch [39].…”

“…Over the next three years, 2019-2021, significant changes took place in the spatial distribution of seismic noise parameters www.videleaf.com (Figure 4, maps on the right). The danger area moved to southern part of the region in the latitude range of 50-54 N [39]. On 25 March 2020, the North Kuril earthquake with Mw = 7.5 (No 5 in Table 3) occurred to the south of the indicated area.…”

“…To analyze spatiotemporal variations of seismic noise, time series of the daily values of four statistical parameters were used: generalized Hurst exponent α*, singularity spectrum support width ∆α, wavelet-based spectral exponent β, and minimum normalized entropy of squared orthogonal wavelet coefficients En calculated for all stations during the 2011-2021 observation period [35,36,39]. The statistical parameters of seismic noise were estimated according to daily data of 1-min signal registration on BHZ channels for all stations (Figure 1, Table 1).…”

“…. By construction, 0 ≤ En ≤ 1. www.videleaf.com When estimating the wavelet spectral exponent β, the orthogonal wavelet decomposition of the signal fragments in the current daily time window [5] is used, which was previously used in [43] to analyze the noise component of signals from a network of 1203 GPS stations in Japan before the Tohoku mega earthquake on March 11, 2011, and in [36,39] when processing noise data from the Kamchatka Peninsula.…”

“…Such a model of noise parameters' behavior during the preparation for strong earthquakes was used in interpreting the results of processing data from a network of stations in Kamchatka [35,36,39], as well as in the present work. We also note that this simple model of the behavior of seismic noise parameters during the preparation of a strong earthquake is in good agreement with the data on the decrease in the multifractal parameters α* and ∆α in the regions of future earthquake sources with Мw = 8.3 and 9.0 in Japan [8].…”