The field of ambient seismic noise (when no identifiable ac-tive sources emit a signal called background noise) and mul-tiple scattered parts of seismic waveform (seismic coda) have become useful data for seismic interferometry, currently being developed as an ambient noise tomography and coda wave in-terferometry respectively (Derode et al., 2003; Campillo and Paul, 2003, Nakahara et al. (2009), Shapiro and Campillo, 2004, Wapenaar, 2004, Shapiro et al., 2005, Snieder, 2004 and 2007, Snieder at all, 2007, Ruigrok et al., 2009, Yao at all, 2009). It is possible to extract local geological information by fundamental steps of seismic interferometry like cross-correlation and stack-ing. Cross-correlation is the process of measuring the travel time difference of two waves recorded between a pair of stations and it helps us to find similarities between these two waves. Stack-ing is the process of addStack-ing together all records from all virtual sources. Virtual sources are sources with arbitrary locations. Am-bient noise, which is an effect of local or regional background noises (caused by winds, tides, temperature variations, seasonal changes or local vibration from human sources) is treaded as vir-tual seismic source. The Passive Image Interferometry (PII) was successfully applied to monitoring changes in volcanoes and near source regions of large crustal earthquakes (Sens-Schoen-felder and Wegler, 2006, Wegler and Sens-Schoen(Sens-Schoen-felder, 2007, Brenguier, 2008a,b, Nakahara et al., 2009, Wegler et al., 2009).
Following Pei’s discussion about summary of characteris-tics of seismic ambient noise (Pei D., 2007), it can be said that ambient seismic noise is ubiquitous and its amplitude is small. Ambient noise, or background noise, can be divided to natural (microseism) and human (microtremor). Microseism consists of the frequency range between 0.05 Hz to 1 Hz (primary micro-seismic peak 0.05 ÷ 0.1 Hz and secondary micromicro-seismic peak 0.1 ÷ 0.3 Hz), and microtremor between 1 Hz to 10 Hz. The source of microseism is mainly oceans, and of microtremor is traffic, industry and human activity.
Permanent broadband seismic observatories record a large range of frequencies data to register different processes induced by Earth. Broadband instruments as velocity sensors have been continuously and globally recording data of high quality since 1980.
The Polish Seismological Network of the Institute of Geo-physics, The Polish Academy of Sciences consists of 9 stations. The first seismic station in Poland was founded in 1903 by Mau-rycy Pius Rudzki in Kraków. Now, most of the seismological observatories are located far away from urban areas and are de-ployed in such a way to provide observations from the whole area of Poland. The GEOFON network came into existence in 1993 and is now operated jointly with international partners, and in 2014 it consisted of about 80 active stations.
In this study, I used the archive waveform data from across Poland available by the IRIS DMC (Incorporated Research In-stitutions for Seismology Data Management Center) recorded by nine broadband seismometers with the same sampling rate 20 Hz of type STS-2 with a flat velocity proportional to the
re-sponse characteristic in the frequency range from 120 seconds (8.33 mHz) to about 10 Hz. The cross-correlation technique was carried out by the Windows Selection Method (WSM) to mea-sure similarities between two signals of ambient seismic noise recorded between a pair of stations. This method allows to re-cover the empirical wave train, the Green’s function, which can be treated as an impulse response between two receivers. Ev-ery pair of stations is cross-correlated in the frequency domain for a 900 s long time window in a selected period of time. The aim of this paper is to show recovered energetic arrivals in mi-croseism frequency bands observed in correlograms of ambient seismic noise recorded by broadband seismometers across Po-land. The presented results were carried out as part of the project supported by the National Science Centre in Poland no. UMO-2012/05/B/ST10/00512.
METHODS
The data is subjected to several steps of the following calcu-lations (Nakahara et al., 2009; Garus, Wegler, 2011) collected as the Windows Selection Method (WSM):
1) Extract a time window for two stations (length of the time window equals to 900 s).
Danuta Garus, PhD
University of Silesia, Faculty of Earth Sciences
Experimental correlograms of the ambient seismic noise across Poland
Fig. 1. Locations of the nine seismic stations used in this study. Seven Polish stations: BEL (Belsk), NIE (Niedzica), GKP (Gorka Klasztorna), KSP (Ksiaz), KWP (Kalwaria Paclawska), OJC (Ojcow), SUW (Suwalki), and two additional stations in Germany: the RUE (Ruedersdorf) station and the MORC (Moravsky Beroun) station in The Czech Republic. The map in Figure 1 was generated
2) Remove the trend, taper and filter the data of the time windows for both stations (Gaussian band-pass filter with eight dominant frequencies: 2 ± 0.5 Hz, 1 ± 0.25 Hz, 0.5 ± 0.125 Hz, 0.25 ± 0.0625 Hz, 0.125 ± 0.0313 Hz, 0.0625 ± 0.0156 Hz, 0.0313 ± 0.0078 Hz, 0.0156 ± 0.0039 Hz).
3) Compute the Root-Mean-Square (RMS) value of the time windows for both stations.
4) Accept or reject the time window according to the RMS value (to exclude both earthquakes and data gaps) for both stations.
5) If both time windows are accepted, compute the cross-correlation.
6) Move to the next time window and restart with step 1).
7) Stack up the cross-correlograms of all accepted time win-dows to obtain cross-correlograms, or simply correlo-grams, corresponding to a stacking time of a few months or for the whole year.
Therefore, this method removes seismic events from con-tinuous data. The RMS time series describes the filtered signal
as an envelope curve with a sampling rate of 900 s and ‘detects’ seismic events in the form of “RMS peaks” relative to “RMS background” which describes seismic noise (Garus, Wegler, 2011). Time windows which do not fulfill the RMS threshold conditions are rejected.
Energetic arrivals in correlograms are analyzed using calcu-lations of the group velocities versus distance between stations. The basis of the present study is a data set of ambient noise recorded by the nine broadband stations: seven Polish stations - BEL (Belsk), NIE (Niedzica), GKP (Gorka Klasztorna), KSP (Ksiaz), KWP (Kalwaria Paclawska), OJC (Ojcow), SUW (Suwalki), and two additional stations in Germany – the RUE (Ruedersdorf) station and the MORC (Moravsky Beroun) sta-tion in The Czech Republic (Fig. 1).
Periods of calculations and the set of stations were selected base on access to the available data. While carrying out this re-search project, access to the available data was widened. For example, at the beginning of the project, data was available only for five Polish stations, currently it is available for seven. Thus, some of the presented results include the period from 2000 to 2004 for stations KWP, MORC, SUW, RUE and from 2011 to 2013 for stations GKP, KSP, KWP, OJC. Other results concern
Fig. 2. RMS time series from the year 2000 for stations KWP (a), MORC (b), SUW (c), RUE (d) Values of dominant filter frequencies in Hz and in second are located on the right side of each graph
the years from 2011 to 2015. To sum up, in this article I present results of stacking up data from winter months, summer months as well as yearly data used for different years.
RESULTS AND DISCUSSION
Now, in figures 2 and 3, we see respectively several exam-ples of the RMS time series for summer months of 2000 for sta-tions KWP, MORC, SUW, RUE and for summer months of 2011 for stations GKP, KSP, KWP, OJC and in figures 4 and 5 present the RMS time series for 8 dominant filter frequencies from years 2012 and 2014 for Polish stations KSP and OJC. These calcula-tion are necessary to estimate in different frequencies the RMS threshold values depend on the local conditions of seismic sta-tions. The RMS threshold values help to delete seismic events and data gaps from the continuous seismic signal.
In the highest frequencies, the daily changes (day, night, weekend) of the RMS data are distinctly visible. Moreover, in the winter, the global noise level is higher in Europe because of seismic noise produced in the North Atlantic and in the summer, noise level is lower because the global seismic noise is produced
in the Indian Ocean (Yang and Ritzwoller, 2008, Stehly et al., 2006). Such annual changes were observed for noise recorded at stations of the German Regional Seismic Network (Garus, Wegler, 2011). So, it seems to be very interesting to look into results from each year divided into three periods of calculation: the summer months (from July to September), the winter months (from January to June and from October to December) and the whole year.
In figures 6 and 7, I presented correlograms of vertical com-ponent in primary microseismic band from the summer months of years 2012-2013 and 2014-2015. Figures 8 and 9 show cor-relograms in secondary microseismic band from the summer months of years 2012-2013 and 2014-2015. Then, figure 10 shows correlograms in primary microseismic band from win-ter months of years 2012 and 2014 and figure 11 correlograms in secondary microseismic band from winter months of years 2012 and 2014. The next figures 12 and 13 show correlograms in primary microseismic band for the whole year 2012 and 2014 and in secondary microseismic band for the whole year 2012 and 2014.
The Rayleigh surface wave as a Green function reconstruct-ed from random noise correlations is stronger at the frequency
Fig. 3. RMS time series from year 2011 for stations GKP (a), KSP (b), KWP (c), OJC (d) Values of dominant filter frequencies in Hz and in second are located on the right side of each graph
band 0.05 – 0.1 Hz and is often a time-symmetrical signal (Pei D., 2007). We can observe it in almost all of the graphs in figures no. 6, 7, 8, 9, 10, 11, 12 and 13. Looking at the correlograms for the summer months, the winter months and the whole year al-lows us to observe the differences in the received signal depend-ing on the period of calculation.
In figure 6 (primary microseismic band), we can observe the received time-symmetrical signal at the dominant frequency 0.0625 Hz and 0.0313 Hz for the summer months of 2012 and 2013. But at the dominant filter frequency 0.0156 Hz, we see a signal near the zero lag time. In figure 8 (secondary microseis-mic band), we can also observe a weak coherent time-symmet-rical signal at filter frequencies 0.25 Hz and 0.125 Hz for the summer months of both 2012 and 2013. At the frequency 0.5 Hz, it looks like there is no coherent signal.
In figure 7 (primary microseismic band), a time-symmetrical signal can be observed at all frequencies for the summer months,
whereas for the summer months of 2015, it only appears at the frequency 0.0625 Hz. At the frequency 0.0156 Hz, we can see an amplified signal near the zero lag time and at the frequency 0.0313 Hz we cannot observe any coherent signal. In figure 9 (secondary microseism) we can observe very similar results for the summer months of 2014 and 215 as in figure 7. If we look only at results calculated for the winter months of the years 2012 and 2014, shown in the figures 10 and 11, we can observe the time-symmetrical signal at a frequency of 0.0625 Hz and 0.0313 Hz in primary microseismic band (Fig. 10), and at 0.25 Hz, and 0.125 Hz in the secondary microseismic band (Fig. 11). At fre-quency 0.0156 Hz in the primary microseismic band (Fig. 10) there is a weak coherent signal near the zero lag time and at the frequency 0.5 Hz in the secondary microseismic band (Fig. 11) there is no coherent signal. When we sum up all correlograms for the whole year, then the signal to noise ratio is not always bet-ter (figures 12 and 13), however, characbet-teristics of the received signal at each frequency remain the same as described above.
Fig. 4. RMS time series at all filters frequencies from year 2012 for Polish stations KSP (a) and OJC (b) Values of dominant filter frequencies in Hz are located on the right side of each graph
Fig. 5. RMS time series at all filters frequencies from year 2014 for Polish stations KSP (a) and OJC (b) Values of dominant filter frequencies in Hz are located on the right side of each graph
Fig. 6. The Cross-correlations of vertical component records between stations plotted with respect to the distance between these stations stacked for the months between July and September (summer) 2012 (a, c, e) and 2013 (b, d, f) in the primary microseismic band.
Fig. 7. The Cross-correlations of vertical component records between stations plotted with respect to the distance between these stations stacked for the months between July and September (summer) 2014 (a, c, e) and 2015 (b, d, f) in the primary microseismic band
Fig. 8. The Cross-correlations of vertical component records between stations plotted with respect to the distance between these stations stacked for the months between July and September (summer) 2012 (a, c, e) and 2013 (b, d, f) in the secondary microseismic band
Fig. 9. The Cross-correlations of vertical component records between stations plotted with respect to the distance between these stations stacked for the months between July and September (summer) 2014 (a, c, e) and 2015 (b, d, f) in the secondary microseismic band
Fig. 10. The Cross-correlations of vertical component records between stations plotted with respect to the distance between these stations stacked for the months between January and June and also between October and December (winter) 2012 (a, c, e) and 2014 (b, d, f) in the primary microseismic band
Fig. 11. The Cross-correlations of vertical component records between stations plotted with respect to the distance between these stations stacked for the months between January and June and also between October and December (winter) 2012 (a, c, e) and 2014 (b, d, f) in the secondary microseismic band
Fig. 12. The Cross-correlations of vertical component records between stations plotted with respect to the distance between these stations stacked for the whole year 2012 (a, c, e) and 2014 (b, d, f) in the primary microseismic band
Fig. 13. The Cross-correlations of vertical component records between stations plotted with respect to the distance between these stations stacked for the whole year 2012 (a, c, e) and 2014 (b, d, f) in the secondary microseismic band.
Strong noise sources generate surface waves, however in the case of very long distances the surface wave in the secondary microseismic band is attenuated stronger than in the primary mi-corseismic band, thus observation are better in correlograms in the secondary microseismic band (Landes et al., 2010). Those waves can be generated by the interaction of storm waves with the seabed, as they show seasonality correlated with the seasonal migration of the strong oceanic storms. For Poland, an efficient transfer of energy from oceanic to seismic waves occurs during summer months and the noise sources are located in the south-ern Pacific and Indian oceans. During winter months, they are located in the northern Pacific and the Atlantic Ocean. These summer observations suggest that the surface waves for the secondary microseism observed in the correlograms of ambient noise across Poland are induced by the oceanic waves in the deep southern oceans.
In the next step of analysis, the values of the group veloci-ties of the recovered signal at each frequency band versus dis-tance between the stations were estimated. It is assumed that the envelope of the received coherent signal moves at the group velocity. For detection of the envelope of the signal, the Hilbert transform was used. The calculations were carried out for casual (right) and anticasual (left) parts of correlograms. For clarity, the next figures 14, 15 and 16 show the graphs and values of group velocities only at dominant frequency 0.125 Hz, 0.0625 Hz and 0,0313 Hz for the whole 2012 year. The values of the group velocities at each frequency bands were estimated with the cor-relation coefficient (the normalized covariance function). The correlation coefficient based on observations and for the causal part of the correlograms (the right side of correlograms) this pa-rameter equals approximately 0.4 and for the anticausal part of the correlograms equals approximately 0.5.
Fig. 14. The group velocity at dominant frequency 0,125 Hz for the whole year 2012
Table 1. The median values of the group velocities at different dominant filter frequencies (the first column), where vgr_all are the median values of
the yearly data, vgr_left and vgr_right contain the separate median values of
the left and the right side of the correlograms.
[Hz] vgr_all [km/s] vgr_left [km/s] vgr_right [km/s]
2.000 3.921 4.688 3.921 1.000 1.981 1.886 2.334 0.500 2.487 2.297 2.508 0.250 2.276 2.105 2.429 0.125 3.504 3.553 3.504 0.063 3.423 3.378 3.532 0.031 3.892 3.153 4.702 0.016 1.796 1.796 0.363
The table 1 includes three sets of median values of the group velocities at each frequency band. Median values vgr_all (ta-ble 1) are calculated for the yearly data from both the causal and the anticausal parts of the correlograms, whereas vgr_right and vgr_left are calculated only for the causal and the anticausal respectively.
These sets of median values of the group velocities indicate that the recovered signals at each frequency band in the correlo-grams are mainly surface waves.
The phenomenon of a dispersion curve concerns the depen-dency of a velocity of a surface wave on a frequency. This phe-nomenon is caused by changes in properties of heterogeneous
Fig. 15. The group velocity at dominant frequency 0,0625 Hz for the whole year 2012
Fig. 16. The group velocity at dominant frequency 0,0313 Hz for the whole year 2012
medium with depth. In figure 17, we can see these median val-ues from the table 1 as a dispersion curve of group velocity of the retrieved signal estimated for the yearly data (upper graph), for the winter data (middle graph) and for the summer data (low-er graph).
Wave propagation is influenced by properties of a geological profile. Most of the energy of surface waves, characterized by the group velocity, propagates depending on the frequency on various depths. The depth of penetration of surface waves ap-proximately equals one wavelength.
In figure 18, there are four group velocity versus period. The first, the second, and the third profile are plotted based on the median values presented in the table 1. In the fourth profile, all the previous results are presented together. In this graph, we can distinguish four main changes of values of the group velocities. Based on results of the seismic experiments in Central Eu-rope: POLONAISE’97 – the result of the inversion for profile
Fig. 17. The dispersion curves of group velocity of the retrieved signal estimated for yearly data (the upper graph), for the winter data (the middle graph) and for the summer data (the lower graph).
P4 (Guterch et al., 2004), the lithospheric structure across Po-land can be compared with the main changes of values of group velocities. The profile P4 runs from the Palaeozoic Platform in the southwest, across the Trans-European Suture Zone, onto the East European Craton to the northeast. In figure 19, the sketch of the lithospheric structure across Poland with the group veloc-ity value are presented. The first layer reaches 2 km deep (group
Fig. 19. A simplified sketch of the lithospheric structure across Poland and the group velocity values extracted from correlograms developed by Windows Selection Method (the ambient seismic noise crosscorrelation technique)
velocity equals approximately 3 km/s), the second layer: 2 – 10 km (velocity ~ 2.4 km/s), the third layer: 10 – 30 km (velocity ~ 3.5 km/s), the fourth layer: 30 – 55 km (velocity ~ 3.4 km/s) and the fifth layer: from 55 km (velocity ~ 4 km/s). The first layer corresponds to sedimentary strata, the second corresponds to the Polish basin (filled with sedimentary strata from the Pal-aeozoic and Mesozoic times), the third layer corresponds to the upper crust, the fourth layer corresponds to the lower crust. At the depth of roughly 55 km The Mohorovičić discontinuity (The Moho) is located – the boundary between the Earth’s crust and the mantle. The last layer seems to be the upper mantle.
CONCLUSIONS
This study uses surface waves contained in ambient seis-mic noise recorded mainly by broadband seismometers of The Polish Seismological Network and by additional stations of the GEOFON network. The recovered energetic seismic arrivals in correlograms of ambient seismic noise are achieved by the Windows Selection Method. The RMS time series describes the filtered signal as an envelope curve with a sampling rate of 900 s and ‘detects’ seismic events (Garus, Wegler, 2011). Energetic arrivals in correlograms are analyzed by calculating the group velocities versus distance between stations. The sketch of the lithospheric structure across Poland and the group velocity val-ues extracted from correlograms developed by the Windows Se-lection Method are presented in figure 19.
This lithospheric structure is based on the result of the inver-sion for POLNAISE’97 profile P4 (Guterch et al., 2004). The profile P4 runs from the Palaeozoic Platform in the southwest, across the Trans-European Suture Zone, onto the East European Craton to the northeast.
I predict, based on this research, that in the future I can use the Windows Selection Method to estimate the main lithospheric structure using ambient seismic noise data in aseismic regions successfully.
DATA AND RESOURCES
The facilities of the IRIS Data Management System, and specifically the IRIS Data Management Center, were used for access to waveform and metadata required in this study. The IRIS DMS is funded through the National Science Foundation and specifically the GEO Directorate through the Instrumenta-tion and Facilities Program of the NaInstrumenta-tional Science FoundaInstrumenta-tion under Cooperative Agreement EAR-1063471. The map in Fig-ure 1 and other graphs in all figFig-ures were generated using the MATLAB software, License Number 836302.
REFERENCES
1. Brenguier F., Shapiro N. M., Campillo M., Ferrazzini V., Duputel Z., Coutant O., Nercessian A. (2008a): Towards forecasting volcanic eruptions us-ing seismic noise, Nature Geoscience 1, 126 – 130, doi.: 10.1038/ngeo104.
2. Brenguier F., Campillo M., Hadziioannu C., Shapiro N. M., Nadeau R.M., Larose E. (2008b): Postseismic Relaxation Along the San Andreas Fault at Parkfield from Continuous Seismological Observations, Science, vol. 321, 1478 – 1481, doi.: 10.1126/science.1160943.
3. Campillo M. (2006): Phase and Correlation in ‘Random’ Seismic Fields and the Reconstruction of the Green Function, Pure appl. Geophys. 163 (2006) 475-502, doi.: 10.1007/s00024-005-0032-8.
4. Campillo M., Paul A. (2003): Long-range correlations in the diffuse seismic coda, Science, vol. 299, 547-549, doi.: 10.1126/science.1078551.
5. Derode A., Larose E., Tanter M., de Rosny J., Tourin A., Campillo M., Fink M. (2003): Recovering the Green’s function from field-field cor-relations in an open scattering medium (L), J. Acoust. Soc. Am. 113(6), doi: 10.1121/1.1570436.
6. Derode A., Tourin A., Fink M. (1999): Ultrasonic pulse compression with one-bit time reversal through multiple scattering, J. Appl. Phys. 85, 6343-6352, doi.:10.1063/1.370136 (10 pages).
7. Dziewonski, A., Bloch S., Landisman M. (1969): A technique for the analysis of transient seismic signals, Bull. Seis. Soc. Am., 59, 427-444.
8. Garus D., Wegler U. (2011): The Green’s Functions Constructed from 17 Years of Ambient Seismic Noise Recorded at Ten Stations of the German Regional Seismic Network, Bulletin of the Seismological Society of America, December 2011, v. 101, p. 2833-2842, doi:10.1785/0120110115
9. Gerstoft P., Sabra K. G., Roux P., Kuperman W. A., Fehler M. C. (2006): Green’s functions extraction and surface wave tomography from microseisms in Southern California, Geophysics, 71, SI23-SI31.
10. Guterch A., Grad M., Keller G.: POLONAISE’97, CELEBRATION 2002, ALP 2002, SUDETES 2003 Working Groups, 2004. Huge contrasts of the lithospheric structure revealed by new generation seismic experiments in Central Europe, Przegl. Geol., 52 (8/2), 753-760.
11. Grad M., Guterch A. (2010): Struktura litosfery i geodynamika Europy Centralnej w świetle eksperymentów sejsmicznych POLONAISE’97, CEL-EBRATION 2000, ALP 2002 i SUDETES 2003.
12. POLSKIE I ŚWIATOWE OSIĄGNIĘCIA NAUKI, NAUKI O ZIEMI. KAPITAŁ LUDZKI, Unia Europejska, Europejski Fundusz Społeczny, Gliwice, 109-157.
13. Landès M., Hubans F., Shapiro N. M., Paul A., Campillo M. (2010): Origin of deep ocean microseisms by using teleseismic body waves, J. Geophys. Res. 115, B05302, doi.: 10.1029/2009JB006918.
14. Larose E., Derode A., Campillo M., Fink M. (2004): Imaging from one-bit correlations of wide-band diffuse wavefields, J. Appl. Phys. 95 (12), pp 8393-8399.
15. Larose E., Derode A., Corennec D., Margerin L., Campillo M. (2005): Passive retrieval of Rayleigh waves in disoredered elastic media, Phys. Rev. E. 72, 046607, doi.: 10.113/PhysRevE.72.046607.
16. Larose E., Margerin L., Derode A., van Tiggelen B., Campillo M., Sha-piro N., Paul A., Stehly L., TanterM. (2006): Correlation of random wavefields: An interdisciplinary review, Geophysics, vol. 71, No. 4, P.SI11-SI21, 8FIGS., 10.1190/1.2213356.
17. Larose E., Roux P., Campillo M., Derode A. (2008): Fluctuations of correlations and Green’s function reconstruction: role of scattering, J. Appl. Phys. 103, 114907 (2008).
18. Nakahara H., Wegler U., Shiomi K. (2009): Monitoring Seismic velocity changes using passive image interferometry: An application to the 2005 West off Fu-kuoka prefecture, Japan, earthquake (Mw 6.6), Workshop on “Seismic Wave Scat-tering and Noise Correlation”, Tohoku University, Sendai, Japan (Feb. 16-17, 2009). 19. Pei D. (2007): Modeling and Inversion od Dispersion Curves of Sur-face Waves in Shallow Site Investigations. University of Nevada, Reno.
20. Paul A., Campillo M., Margerin L., Larose E., Derode A. (2005): Em-pirical synthesis of time-asymmetrical Green functions from the correlation of coda waves, J. Geophys. Res. 110, B08302, doi.:10.1029/2004JB003521.
21. Roux P., Sabra K. G., Gerstoft P., Kuperman W. A., Fehler M. C. (2005): P-waves from cross-correlation of seismic noise, Geophys. Res. Lett. 32, L19303, doi.: 10.1029/2005GL023803.
22. Ruigrok E., Campman X., Wapenaar K. (2009). Lithospheric-scale seismic interferometry: a comparison of approaches to deal with an irregular source distribution and source-side reverberations, Deutsche Geophysikalische
Gesellschaft e.V., Noise and Diffuse Wavefields, Extended Abstracts of the Neustadt Workshop, Neustadt an der Weinstrasse, Germany, 5-8 July 2009.
23. Sawazaki K., Sato H., Nakahara H., Nishimura T. (2009): Time-lapse changes of seismic velocity in the shallow ground caused by strong ground mo-tion shock of the 2000 Western-Tottori earthquake, Japan, as revealed from coda deconvolution analysis, Bull. Seism. Soc. Am. 99, 352-366.
24. Sens-Schönfelder C., Wegler U. (2006): Passive image interferometry and seasonal variations of seismic velocities at Merapi Volcano, Indonesia, Geo-phys. Res. Lett. 33, L21302, doi:10.1029/2006GL027797.
25. Shapiro N. M., Campillo M.: (2004): Emergence of broadband Ray-leigh waves from correlations of the ambient seismic noise, Geophys. Res. Lett. 31, L07614, doi.:10.1029/2004GL019491.
26. Snieder R. (2009): A Guided Tour of mathematical Methods for the Physical Sciences Second Edition, Cambridge University Press.
27. Snieder R. S., Wapenaar K., Wegler U. (2007): Unified Green’s func-tion retrieval by cross-correlafunc-tion with energy principles, Physical Review E 75, 036103.
28. Snieder R. (2007): Extracting the Green’s function of attenuating het-erogeneous acoustic media from uncorrelated waves, J. Acoust. Soc. Am. 121(5). 29. Snieder R. (2004): Extracting the Green’s function from the correlation of coda waves: A derivation based on stationary phase, Physical Review E 69, 046610.
30. Stehly L., Campillo M., Shapiro N. M. (2006): A study of the seismic noise from its long-range correlation properties, J. Geophys. Res. 111, B10306, doi.: 10.1029/2005JB004237.
31. Styles P. (2012): Environmental Geophysics, Everything you ever wanted (needed!) to know but were afraid to ask! European Association of Geo-scientists & Engineers (EAGE), The Netherlands, ISBN 978-90-73834-33-0
32. Wapenaar K. (2004): Retrieving the elastodynamic Green’s function of an arbitrary inhomogeneous medium by crosscorrelation, Physical Review Letters 93, 254301.
33. Wapenaar K., Fokkema J., Snieder R. (2005): Retrieving the Green’s function in an open system by cross correlation: A comparison of all approaches (L), J. Acoust. Soc. Am. 188(5).
34. Wegler U., Sens-Schönfelder C. (2007): Fault zone monitoring with passive image interferometry, Geophys. J. Int. (2007) 168, 1029-1033, doi.: 10.1111/j.1365-246X.2006.03284.x.
35. Wegler U., Nakahara H., Sens-Schönfelder C., Korn M., Shiomi K. (2009): Sudden drop of seismic velocity after the 2004 Mw 6.6 mid-Niigata earthquake, Japan, observed with Passive Image Interferometry, J. Geophys. Res. 114, B06305, doi:10.1029/2008JB005869.
36. Wiszniowski J. (2002): Broadband Seismic System: Effect of Transfer Band on Detection and Recording of Seismic Waves, Publication of the Institute of Geophysics Polish Academy of Sciences, Monographic volume B-27(339), Warsaw, ISBN-83-88765-11-6, ISSN-0138-0109
37. Yao H., Campman X., de Hoop M. V., van der Hilst R. D. (2009): Es-timation of surface wave Green’s functions from correlation of direct waves, coda waves, and ambient noise in SE Tibet, Physics of the Earth and Planetary Interiors 177 (2009), 1-11.
38. Yang Y., Ritzwoller M. H. (2008): Characteristic of ambient seismic noise as a source for surface wave tomography, Geochemistry, Geophysics, Geo-systems, vol. 9, No. 2, Q02008, doi.: 10.1029/2007GC001814, ISSN 1525-2027.
ACKNOWLEDGMENTS: A big thank you to the people who gave the sci-entific world the IRIS Data Management System.
FUNDING: This work was supported by the National Science Centre [grant numbers: UMO-2012/05/B/ST10/00512].
