Seismic Noise Correlations Research Articles

Seasonal variations of cross correlations of seismic noise in Israel

The long-period microseismic noise was recently found to carry deterministic information about the crust and upper mantle structure in the cross-section between the two station sites in terms of the surface wave Green function, which is theoretically proportional to the noise cross-correlation function (NCF) for the long-term observations at the pair of the BB stations. We performed daily-long cross correlations for the period of 2–3 years between the 7 BB stations, distributed in the Eastern Mediterranean, and stacked them for every month of a year. A preprocessing of the broadband waveforms included whitening of the direct and inverse DFT of the waveforms to avoid influence of earthquakes recordings and to equalize energy of different types of microseisms. As the result, we have found that the NCFs obtained exhibit clear seasonal variations within period band 2–20 s persistent from year to year. For the best description of these variations, we applied seasonal diagrams, which presented the distribution of the NCF maximal amplitudes in three narrow frequency bands, with respect to the month of a year. The diagrams helped in determining that these variations can be split into four types. The different types of the seasonal NCF variations are assumed to be attributed to the four certain remote deep ocean regions, which are responsible for the increased microseismic activity due to the specific interaction between the ocean waves and the bottom during ocean storms.

Convergence of correlations in multiply scattering media

Correlations of ambient seismic or acoustic noise are now widely used to reconstruct the impulse response between two passive receivers as if a source was placed at on of them. Applications include terrestrial and solar seismology, underwater acoustics and structural health monitoring. Nevertheless, for a given set of data, correlations do not only yield automatically the Green function between the sensors, but also contains residual fluctuations that might "blur" the images. [Gizon et al, Astrophys. J. 614 (2004); Weaver and Lobkis, J. Acoust. Soc. Am. 117 (2005); Sabra et al., J. Acoust. Soc. Am. 118 (2005)]. We propose a model to describe the “signal‐to‐fluctuations” ratio in the correlations in the case of nonstationary wavefields, and more particularly in the case of scattering media. The work includes theoretical derivations and numerical simulations. The role of multiple scattering in the rate of convergence of the correlations toward the Green function is quantitatively evaluated [Larose et al, (submitted 2008)].

Surface wave tomography of China from ambient seismic noise correlation

We perform ambient noise tomography of China using data from the China National Seismic Network and surrounding global and regional stations. For most of the station pairs, we retrieve good Rayleigh waveforms from ambient noise correlations using 18 months of continuous data at all distance ranges across the entire region (over 5000 km) and for periods from 70 s down to about 8 s. We obtain Rayleigh wave group velocity dispersion measurements using a frequency‐time analysis method and invert for Rayleigh wave group velocity maps for periods from 8 s to 60 s. The tomographic maps display significant features that correlate with surface geology. Major basins, including Tarim, Junggar, Qadaim, Sichuan, Bohai‐Wan, and Songliao, are all well delineated by slow group velocities at shorter periods (10 to 20 s). The overall trend of crustal thickening from east to west is well represented by group velocity decreases from east to west at periods around 30 s.

Small-scale seismic inversion using surface waves extracted from noise cross correlation

Green's functions can be retrieved between receivers from the correlation of ambient seismic noise or with an appropriate set of randomly distributed sources. This principle is demonstrated in small-scale geophysics using noise sources generated by human steps during a 10-min walk in the alignment of a 14-m-long accelerometer line array. The time-domain correlation of the records yields two surface wave modes extracted from the Green's function between each pair of accelerometers. A frequency-wave-number Fourier analysis yields each mode contribution and their dispersion curve. These dispersion curves are then inverted to provide the one-dimensional shear velocity of the near surface.

Characteristics of ambient seismic noise as a source for surface wave tomography

Interstation cross correlations of ambient seismic noise from 1 year of continuous data at periods between 6 and 50 s are used to study the origin of the ambient noise using stations located in Europe, southern Africa, Asia, and three regions within North America. The signal‐to‐noise ratios (SNR) of Rayleigh waves for positive and negative correlation time lags at periods of 8, 14, 25 and 50 s are used to determine the azimuthal distribution of strong ambient noise sources. Ambient noise in both the primary (10–20 s) and secondary microseism bands (5–10 s) comes dominantly from the directions of relatively nearby coastlines with stronger noise occurring in the Northern Hemisphere in northern winter and in the Southern Hemisphere in southern winter, consistent with the hypothesis that oceanic microseisms are generating this noise. The observed differences in the directivity of noise in the primary and secondary microseism bands are the consequence of propagation and attenuation, rather than the location of generation. At intermediate and long periods (>20 s), there is much less seasonal variation in both signal strength and directivity. We argue that our results are explained most simply by near‐coastal sources rather than deep ocean sources at all periods. Although the dominant ambient noise sources are distributed inhomogeneously in azimuth, strong ambient noise emerges from most directions when using recordings that are 1 year in duration. Simulations illustrate that this is what ensures the accuracy of the empirical Green's functions and ambient noise tomography.

Combining Seismic Data from Passive and Active Sources for understanding the terrane structure of the Eastern Goldfields, Western Australia

A passive seismic survey to investigate variations in crustal structure across the Yilgarn craton has shown significant contrasts in seismic velocity models between neighbouring terranes/superterranes. The Eastern Goldfields showed a unique variability in crustal structure in agreement with a recent reinterpretation of terrane boundaries within the Yilgarn craton. We further investigated the Eastern Goldfields region using a 3-way approach which combined conventional passive seismic analysis with innovative seismic noise-correlation methods and constraints from active source data. Conventional passive seismic analysis enables the Receiver Function S-velocity structure, and hence composition, of the lower crust to be constrained. Noise -correlation analysis allows seismic velocity models in the 5-15 km depth range to be determined and provides medium resolution coverage across regions not previously explored using active seismic methods. Where active source data have been acquired, shallow structure and deeper seismic velocity determinations are added, providing an unprecedented combination of seismic constraints on the structure of this complex and economically important region. We find that, although some individual terrane boundaries within the new Eastern Goldfields reinterpretation are open to question, the concept of the multi-terrane amalgamation is substantially justified by the exceptional variability of the lower crustal structure. Upper crustal structure is often characterised by seismic discontinuities which may represent detachment surfaces or layered structure that varies between terranes over a sub-100 km length scale. The accretionary history of the superterrane and associated regional tectonic setting of numerous formations of economic significance would now appear to be beyond question.

Rayleigh‐wave group‐velocity of the Icelandic crust from correlation of ambient seismic noise

By cross‐correlating 2 years of ambient seismic noise from 31 broadband stations distributed across Iceland, we have been able to measure Rayleigh wave group velocity in three discrete frequency bands centered on 4, 7 and 17 s, sensitive to approximately the top 6, 10 and 25 km of the crust, for 289, 420 and 139 station‐to‐station paths, respectively. These are inverted to yield tomographic group velocity maps of the Icelandic crust. The results correlate with the major tectonic features on Iceland, and agree well with earlier seismic studies based on traditional methods (active source and earthquake). Specifically, we obtain low velocities in the central parts of Iceland, coinciding with the rift zones and young crust. In contrast, the parts of Iceland covered with older Tertiary volcanics reveal relatively higher velocities. At the shorter periods (4 and 7 s) this correlation reflects porosity, degree of fracturing and alteration. At the longest period (17 s), i.e. at greater depth, this correlation may reflect the thermal state of the crust, with warm and seismically slow material near the rifts and cooler and seismically faster material beneath the Tertiary formations to the east and west.

Influence of the seismic noise characteristics on noise correlations in the Baltic shield

It has recently been shown that correlations of seismic noise can contain significant information about the Green's function along the station profile. Using an array of 38 temporary broad-band stations located in Finland between 1998 September and 1999 March, we study the resulting 703 noise correlations to understand how they are influenced by the directivity of the noise field. The latter information is obtained through f-k analysis of data from two permanent seismic arrays in Germany and Norway and from a subset of stations of the array in Finland. Both types of analysis confirm that the characteristics of the seismic noise is strongly frequency-dependent. At low frequencies (0.02–0.04 Hz), we observe diffuse noise and/or randomly distributed sources. In contrast, the noise is strongly direction-dependent and not fully diffuse in the intermediate period ranges (0.04–0.25 Hz) which correspond to the first and second microseismic peak, created at the Irish and Scottish coast and the western coast of Norway. In this frequency interval the noise is sufficiently close to a plane wave to introduce systematic errors in group velocities for station pairs which are not parallel to the direction of the dominant incident noise. Phase velocities calculated by slant stack over many traces are however correct, independently of profile direction. In the high-frequency band (0.25–1.0 Hz), the situation is a mix between the low-frequency and the intermediate frequency cases. Average phase velocities and individual group velocities from well-oriented profiles are in excellent agreement with results from Rayleigh wave studies of the area.

On the correlation of seismic microtremors

We present results of the analysis of microtremor measurements using a small array. Our analysis is based on the computation of cross‐correlation functions between stations in both frequency and time domains. We obtain similar results in both domains and link those results to the application of Aki's spatial autocorrelation method of using a single station pair and to recent studies that have shown that the Green's function between two stations can be retrieved from the temporal cross correlation of seismic noise. We show that the same simple subsoil structure allows interpretation of our correlation results in time and frequency. We observe both Love and Rayleigh waves; however, Love waves dominate the records in our lower‐frequency range (between 3.6 and 6 Hz), while Rayleigh waves are prevalent in the records in our higher‐frequency band (from 6 to 20 Hz). Frequency domain cross correlation yields better results in the lower‐frequency range, while time domain cross correlation shows very clear results in the higher‐frequency band. Thus both analyses are complementary. Our results show that ambient vibration recorded at the free surface includes different types of waves but that the correlation between any two stations is governed by the more stable propagation mode between them, surface waves in the case of a layered medium. Our results shed some light on the nature of microtremors and on the reasons why the spatial autocorrelation method or the horizontal‐to‐vertical spectral ratios are useful in geophysical and site response studies.

Lunar subsurface investigated from correlation of seismic noise

By correlating seismic noise recorded by four sensors placed on the Moon during the Apollo 17 mission, we have retrieved a well-defined dispersed Rayleigh wave pulse. Inversion of its group velocity provides new constraints on the lunar subsurface structure. The estimated ”signal-to-noise” ratio (SNR) of the retrieved Rayleigh wavetrain is strongly dependent on solar illumination, effectively making solar heating a source of seismic noise on the Moon. This result suggests that in future planetary missions it is feasible to extract information on the internal structure of extraterrestrial objects by correlating seismic noise even when natural quakes are absent.

Emergence of broadband Rayleigh waves from correlations of the ambient seismic noise

We demonstrate that the coherent information about the Earth structure can be extracted from the ambient seismic noise. We compute cross‐correlations of vertical component records of several days of seismic noise at different pairs of stations separated by distances from about one hundred to more than two thousand kilometers. Coherent broadband dispersive wavetrains clearly emerge with group velocities similar to those predicted from the global Rayleigh‐wave tomographic maps that have been constrained using ballistic surface waves. Those results show that coherent Rayleigh waves can be extracted from the ambient seismic noise and that their dispersion characteristics can be measured in a broad range of periods. This provides a source for new types of surface‐wave measurements that can be obtained for numerous paths that could not be sampled with the ballistic waves and, therefore, can significantly improve the resolution of seismic images.

Volcanic tremors at Deception Island (South Shetland Islands, Antarctica)

Deception Island (62°43′S, 60°57′W) is an active volcano located at the spreading centre of the Bransfield Strait backarc on the southwestern side of the Scotia Sea region (Antarctica). Since 1986, the Spanish Antarctic National Program has supported geophysical and geological surveys addressed to the study of seismic and volcanic activities in the area. The present volcanotectonic activity is restricted to significant seismic activity and several fumarole fields. Numerous earthquakes, with magnitudes ranging from 0.1 to 2.2, have been recorded, most of them distributed along the main fractures of the island, which coincide with the regional structural trend of the Bransfield Strait back-arc. Several types of volcanic tremors have also been recorded, all of them with high spectral stability contents with few and well-defined spectral peaks. The location of the stations that record tremors, the correlation of seismic noise with the tremors and the geological characteristics of Deception Island, together suggest that the tremors are generally associated with geothermal noise originated in the uppermost ducts of the fumarolian system. If we associate the surficial ducts to harmonic oscillators, it is observed that the motion is basically unidirectional, as is expected in the degasification of an aquifer. Taking the main observed frequencies and applying the classical mechanical expressions of organ pipes, we obtain a duct length which agrees with the length of a water column necessary to maintain the equilibrium as indicated by the application of the Clausius-Clapeyron equation.