Fractal analysis of seismoacoustic signalsof near-surface sedimentary rocks in Kamchatka

• Published in: Geofizika
• Main Authors: Imashev, Sanjar, Mishchenko, Mikhail, Cheshev, Mikhail
• Format: Article in Journal/Newspaper
• Language: English
• Published: Andrija Mohorovicic Geophysical Institute 2019
• Subjects: seismoacoustic signal; monofractal; Hurst exponent; multifractal; detrended fluctuation analysis; seizmoakustični signal; monofraktal; Hurstov eksponent; multifraktal; detrendirana fluktuacijska analiza; Karymshina; Koda; Niska; Kamchatka
• Online Access: https://hrcak.srce.hr/232674; https://hrcak.srce.hr/file/338482

We studied, by the mono- and multifractal detrended fluctuation analysis (DFA), time fluctuations in the dynamics of seismoacoustic data, recorded in Karymshina site, which is located in a seismic area of Kamchatka. We took a series of seismoacoustic responses to the regional seismic events with the mag-nitudes M > 4 for the period 2017–2018. The series was divided into three groups (high, medium and low) based on the amplitude of recorded seismoacoustic re-sponse. Background noise segments of the signals demonstrated monofractal behavior similar to white noise by almost constant values of generalized Hurst exponent Hq≈0.5 and very small width of the multifractal spectrum Δa ≈ 0.1. Analysis of the high amplitude seismoacoustic signals with clear P-, S- and coda waves showed that P- and S-waves demonstrate wider multifractal spectrum (ΔaP = 0.37, ΔaS = 0.35) and range of generalized Hurst exponents Hq in com-parison with coda wave, characterized by almost constant Hq and minimal width of multifractal spectrum (ΔaCODA = 0.13). We showed that the properties of the multifractal spectrum could be used in detection of seismic wave arrival, estima-tion its duration and separation of P-, S- and coda waves. Application of the monofractal DFA in a sliding window showed that the acoustic signal transits from monofractal and uncorrelated background noise (Hurst exponent equals to 0.5) into the long-range dependent state during seismic waves arrival, that is helpful in analysis of the signals, particularly in case of low amplitude acous-tic responses, usually demonstrating an unclear waveform. Difference in mul-tifractal spectrum width between the original low amplitude signal and its sur-rogates, obtained by random shuffling showed that the multifractality in the signal is dominantly due to long-range correlations. U ovom smo radu mono- i multifraktalnom detrendiranom fluktuacijskom analizom (DFA) proučavali vremenske fluktuacije u dinamici seizmoakustičnih podataka zabilježe-nih na postaji Karymshina, koja je smještena u seizmički aktivnom području Kamčatke. U analizi smo koristili niz seizmoakustičnih odziva prilagođenih na regionalne potrese s magnitudama M > 4 za razdoblje 2017.–2018. Niz smo podijelili u tri grupe (visoka, sred-nja i niska) na temelju amplitude zabilježenog seizmoakustičnog odziva. Dio mikroseiz-mičkog nemira prisutnog unutar signala iskazuje monofraktalnu strukturu sličnu bijelom šumu s gotovo konstantnim vrijednostima generaliziranog Hurstovog eksponenta Hq ≈ 0,5 i vrlo malom širinom multifraktalnog spektra Δa ≈ 0,1. Analiza seizmoakustičnog signa-la visoke amplitude s jasnim P-, S- i koda valovima pokazala je da P- i S-valovi pokazuju širi multifraktalni spektar (ΔaP = 0,37, ΔaS = 0,35) i raspon generaliziranih Hurstovih eksponenata Hq u usporedbi s koda valovima, koje karakterizira gotovo konstantan Hq i minimalna širina multifraktalnog spektra (ΔaCODA = 0,13). Pokazali smo da se svojstva multifraktalnog spektra mogu upotrijebiti za otkrivanje nailaska seizmičkih valova, pro-cjenu nihovog trajanja i razdvajanje P-, S- i koda valova. Primjena monofraktalne DFA metode na zapise u kliznom prozoru pokazala je da akustički signal prelazi iz monofrak-talnog i nekoreliranog mikroseizmičkog nemira (Hurstov eksponent jednak 0,5) u stanje dugog dometa tijekom dolaska seizmičkih valova, što je korisno u analizi signala, poseb-no u slučaju akustičnih odziva niske amplitude s nejasnim valnim oblikom. Razlika u širini multifraktalnog spektra između izvornog signala niske amplitude i njegovih zam-jenskih oblika, dobivenih nasumičnim odabirom, ukazuje da multifraktalnost u signalu dominantno ovisi o dalekosežnim korelacijama.