2:45 PM - 3:00 PM
[SSS03-05] Spectral characteristics, time variations and estimation of source directions of marine microseisms recorded in Cuba
Keywords:marine microseisms, seasonal variation, particle motion, sources directions
The ambient seismic wavefield consists of seismic signals generated from different sources, excluding earthquakes and explosions, coming from different directions and recorded simultaneously at the receiver stations. Understanding noise characteristics and source distribution is crucial to image the subsurface structure of the Earth reliably, using these signals in conjunction with earthquake data. Our study aims analyzing the spectral characteristics of marine microseisms recorded by 13 broadband seismic stations in Cuba and estimate the location of the sources that contribute to the generation of the seismic noise in the frequency band of marine microseisms. We focus on three period ranges (1s-2s, 2s-4s and 4s-8s) and analyze the data for the year 2020.
First, we determined the power spectra by the Welch periodogram method (Welch, 1967) applied to data recorded on the vertical component, in windows of 3,600s with a 50% overlap. The method reduces the variance without additional smoothing or binning the spectral estimates for an appropriate spectral resolution (e.g., Anthony et al., 2020). Since the occurrence of teleseismic events may bias the spectral estimations, we eliminated the teleseismic signal of earthquakes with magnitudes larger than or equal to 5.5 by removing the two-hour data after such events. We next analyzed the mean power spectral density behavior in space, time, and frequency for the primary and secondary microseisms peaks. Our results reveal a broad secondary microseisms peak (1s-10s) with a dominant period at almost all stations between 2.9s to 4.1s, slightly changing from station to station. A more stable primary microseism peak, compared with the secondary one, is found at 15.6s for the majority of stations. The power spectral density ranges from -118 to -128 dB and -150 to -156 dB for the secondary and primary microseisms, respectively. We also analyzed the energy variation of microseisms in the two seasons (i.e., dry and wet) in Cuba. The strongest signals at the main period of the secondary and primary microseism are found during the dry season. This seasonal variation is weak, which is consistent with findings by Stutzmann et al. (2009) that report stable temporal variations of the microseisms’ energy at stations located near the equator.
Second, we applied polarization analysis (Takagi et al., 2018) for the data recorded at each station to estimate the distribution of the direction of sources, as well as the intensity of the seismic energy in that direction. We focus our analysis on the Rayleigh and P-waves polarization, assuming both are uncorrelated. We found highly directional incident Rayleigh and P waves at periods between 1s-2s. Source directions of Rayleigh waves at these periods suggest that microseisms are developed in a nearby sea area, relative to the stations’ location. At 2s-4s, source contributions of Rayleigh waves are pointing to the sea in the north of Cuba, for stations located in the northwest and central regions, and to the Caribbean Sea for stations located in the southeast region. At 4s-8s, the scatter sources’ directions of Rayleigh waves indicate that the sources’ contributions are coming from different directions and there is no dominant source region. There is no evidence of seasonal variation of the source locations for the Rayleigh waves at periods 1s-2s and 2s-4s. On the contrary, the seasonality distribution of the sources of the P waves at 4s-8s is strong at most of the stations: sources are coming from north and south, corresponding to the local winter of the northern and southern hemispheres, respectively. This seasonality of the P-waves source locations is consistent with P-waves global distributions (e.g. Nishida and Takagi, 2022). Our findings provide useful information to be considered in further studies to investigate the crust structure in Cuba using seismic noise.
First, we determined the power spectra by the Welch periodogram method (Welch, 1967) applied to data recorded on the vertical component, in windows of 3,600s with a 50% overlap. The method reduces the variance without additional smoothing or binning the spectral estimates for an appropriate spectral resolution (e.g., Anthony et al., 2020). Since the occurrence of teleseismic events may bias the spectral estimations, we eliminated the teleseismic signal of earthquakes with magnitudes larger than or equal to 5.5 by removing the two-hour data after such events. We next analyzed the mean power spectral density behavior in space, time, and frequency for the primary and secondary microseisms peaks. Our results reveal a broad secondary microseisms peak (1s-10s) with a dominant period at almost all stations between 2.9s to 4.1s, slightly changing from station to station. A more stable primary microseism peak, compared with the secondary one, is found at 15.6s for the majority of stations. The power spectral density ranges from -118 to -128 dB and -150 to -156 dB for the secondary and primary microseisms, respectively. We also analyzed the energy variation of microseisms in the two seasons (i.e., dry and wet) in Cuba. The strongest signals at the main period of the secondary and primary microseism are found during the dry season. This seasonal variation is weak, which is consistent with findings by Stutzmann et al. (2009) that report stable temporal variations of the microseisms’ energy at stations located near the equator.
Second, we applied polarization analysis (Takagi et al., 2018) for the data recorded at each station to estimate the distribution of the direction of sources, as well as the intensity of the seismic energy in that direction. We focus our analysis on the Rayleigh and P-waves polarization, assuming both are uncorrelated. We found highly directional incident Rayleigh and P waves at periods between 1s-2s. Source directions of Rayleigh waves at these periods suggest that microseisms are developed in a nearby sea area, relative to the stations’ location. At 2s-4s, source contributions of Rayleigh waves are pointing to the sea in the north of Cuba, for stations located in the northwest and central regions, and to the Caribbean Sea for stations located in the southeast region. At 4s-8s, the scatter sources’ directions of Rayleigh waves indicate that the sources’ contributions are coming from different directions and there is no dominant source region. There is no evidence of seasonal variation of the source locations for the Rayleigh waves at periods 1s-2s and 2s-4s. On the contrary, the seasonality distribution of the sources of the P waves at 4s-8s is strong at most of the stations: sources are coming from north and south, corresponding to the local winter of the northern and southern hemispheres, respectively. This seasonality of the P-waves source locations is consistent with P-waves global distributions (e.g. Nishida and Takagi, 2022). Our findings provide useful information to be considered in further studies to investigate the crust structure in Cuba using seismic noise.