S13C-1070 1340h
Characterization of the Late 2003 Microseismic Signal Offshore Southern California
The Scripps OBS (ocean Bottom seismometer) group that provides instruments for the national instrument pool has recently test--deployed their new passive seismic instruments. These instruments have a LC2000 data logger system, a Nanometrics Trillium 40 3--component seismometer and a Cox--Webb differential pressure gauge. The test deployment took place between November 6th, 2003 and January 8th, 2004 and included five units operating at a water depth of about 1000~m, 40~km offshore from La Jolla, CA. Data were recorded continuously, at a sampling rate of 31.25~Hz. During the deployment, we recorded numerous local, regional and teleseismic earthquakes but this study focuses on the observation of signal in the frequency band 0.01--0.5~Hz that is not related to earthquakes. We determine the dominant bands of the infragravity and microseismic noise recorded on the ocean floor and compare its amplitudes with those recorded on land on broad--band seismometers of the Trinet, Anza and GSN networks. We also compare our seismic data with oceanographic and weather data. We identify four time periods of high microseismic activity that occurred in December 2003. Two of these are associated with major storm events that exhibit a marked increase in wave height and wind speed. These events correlate well with the microseism recorded on the ocean floor but are recorded only poorly on land. The sharp onset of one of the events allows us to track its propagation. The microseism signal increases first at station CIA (Catalina Island), then one hour later at a buoy at the La Jolla Scripps Pier, and another hour later at the OBSs, which are actually closer to CIA than the pier is. The two other events are somewhat puzzling: the dominant frequency is about 1.5 times lower than for the two storms and the signals show no obvious dispersion. They are extremely well recorded on the OBSs as well as on land but do not correlate with any regional weather or wave height data. The fact that these events are also clearly seen at other stations in the Pacific suggests a powerful source and/or favorable, nondispersive propagation conditions.
S13C-1071 1340h
Regional and Global Variations of Microseisms
We developed a methodology to analyze microseisms using seismic data from Southern California. Our basic approach consists of (1) selection of Rayleigh wave signals using three-component seismograms, (2) determination of the direction of propagation (excitation source), and (3) measurement of seasonal variations in the horizontal to vertical amplitudes that we refer to as the HZ ratio. We have now collected data from seismic stations in North America extensively. The total number of stations is over 200 and covers various climatic regions in North America. Based on this extensive data set, we report on comparisons of various characteristics of microseisms such as source locations and their seasonal variations (changes in source azimuth), and seasonal variations in the HZ ratios. There are some interesting differences between microseisms on the Pacific coast side and those on the Atlantic coast side. The first obvious difference is in the peak frequencies. At stations close to the Pacific ocean, the predominant frequencies of microseisms are about 0.15 Hz. At stations close to the Atlantic oceans, on the other hand, is about 0.2 Hz. While substantial energy is found over an interval of 0.1 Hz and thus there exist overlapping frequency ranges, the systematic difference of the peak frequency is unmistakable. We speculate that this feature is related to the ocean depth differences in the source regions of microseisms, meaning the microseisms are excited at shallower source region (ocean depth region) on the Atlantic side. However, there still remains a speculation. The second difference is found in the source locations. For stations in Southern California, the source locations were (surprisingly) concentrated within narrow angles; the microseism energy propagated through a station almost in the same direction throughout the year. On the Atlantic side, there seems to be large variations in azimuth, including some indications of seasonal changes in azimuth. The HZ ratios on the Atlantic ocean side display similar seasonal variations with stations on the Pacific side, meaning that they show the minimum in July and the maximum in January. We postulate that these variations are caused by changes of water-content in the vadose zone and thus expect different behaviors for different climatic regions; however, we have not found significant differences in the HZ ratios among different climatic regions yet. We will report on our comparisons between relatively wet region and dry region and discuss the effects on the HZ ratios.
S13C-1072 1340h
Correlation of oceanic microseisms at Californian seismic stations
We attempt to reconstruct Green functions between pairs of stations by cross-correlating records of the ambient seismic noise at those stations. We compute cross-correlations between vertical component records for several days of ambient seismic noise observed at various station-pairs located in California and separated by distances of a few hundreds kilometers. Emerging coherent waveforms are dominated by fundamental mode Rayleigh waves with travel times similar to those measured for the same paths from earthquake excited ballistic surface waves. This reconstruction is expected to work perfectly when the correlated wavefield is completely random and isotropic. Therefore, results of the cross-correlations provide also information about the degree of randomness and isotropy of the target ambient seismic noise.At periods corresponding to the oceanic microseisms (around 7-8 s), the amplitude of the emerging Rayleigh wave is larger for paths nearly perpendicular to the coastal line than for those that are nearly parallel to the coastal line. For paths perpendicular to the coastal line, where the ocean-solid Earth coupling occurs, the resulting cross-correlation are strongly asymmetric demonstrating that more energy is propagating from the coast than in the opposite direction. However, coherent Rayleigh waves also emerge from cross-correlations for inter-station paths that are almost parallel to the coastal line suggesting that, in addition to direct waves excited at the coast, oceanic microseisms contain a non-negligible amount of surface waves that were scattered on inhomogeneities within the Earth. We observe that the azimuthal distribution of energy in the noise is changing with the period band considered.Finally we show that meaningful geological information can be obtained by extracting surface waves from cross-correlations of the seismic noise, providing a way for a passive imaging of the Earths structure.
S13C-1073 1340h
Locating Sources of the Earth's "hum" using an array-based Method
Until now, the continuous background free oscillations of the earth have been studied mainly using normal mode approaches, which do not have the temporal and spatial resolution needed to investigate the geographical distribution of the sources of excitation of this weak signal. For this, a propagating wave approach is needed, to capture the directional dependence of the long period Rayleigh wave energy that contributes to the "hum". We have developed an array-based method which utilizes two (or three) large aperture regional arrays of very long period seismometers, and combines a traditional beam-forming approach with an anti-dispersion filtering technique. This method is designed to enhance the consistent surface wave energies over the array and measure their direction of arrival. The two main arrays, equipped with STS-1 seismometers are in California (Berkeley Digital Seismic Network) and in Japan (F-net). A third array can be formed using STS-1 equipped stations in Europe, but provides weaker directional constraints. Data for intervals of time free of earthquakes of $M_w$ 5.5 or larger are either band-pass filtered between 100-400 sec, or filtered with a gaussian filtered centered at 240 sec. Dispersion corrections are applied, following the reference spherically symmetric PREM model. We analyze maximum stack amplitude as a function of back-azimuth, taking into account the distortion due to non-uniform array responses. Our results show that the Rayleigh wave background energy is generated primarily over the oceans and that the location of the sources shift seasonally, correlating with maxima in significant wave height. We perform forward modelling experiments which confirm that a uniform distribution of sources or a distribution of sources over continental areas does not fit the observed patterns. In order to further investigate the atmosphere/ocean/sea floor coupling mechanism inferred for the source of excitation of the "hum". we need to expand the available seismic time series, which is limited by rigorous selection criteria to avoid contamination by earthquake signals. For this purpose we have developed a filtering method which allows us to remove significant contributions from moderate size earthquakes ($M_w$ $<=$ 6) and further assess the space/time correlations with ocean wave data.
S13C-1074 1340h
Frequency Content of Ambient Seismic Noise in North-Central Illinois
High and low-frequency ground motion vibration measurements were made in DeKalb and Kane Counties, Illinois, to characterize different sources of seismic noise and determine directions and magnitude of motion produced by each source. The Seismic Analysis Code 2000 (SAC2000) was used to process seismic noise data recorded with a low-frequency digital system,, earthquake records downloaded from the WILBER Web site, and noise data recorded by an engineering seismograph with high-frequency geophones. Power-spectral density estimates were computed from an autocorrelation series in most cases. Using the power density spectra routine of SAC2000, selecting autocorrelation windows of 20 s for the low-frequency data and 2 s for the high-frequency data, we have found distinctive seismic noise peaks among the different sites, regardless of the background noise level. The earthquake low-frequency data showed a microseism peak at a frequency of 0.2 Hz for stations located in the Midwestern U.S. Microseism peak frequency did not decrease with increasing distance from Lake Michigan, suggesting the lake is not the primary source of the microseisms, which may be generated in ocean basins. Ambient ground motion recorded by the Northern Illinois University seismic station with a 2 Hz natural period seismometer exhibited peaks around 0.8 and 2.2 Hz. Ground motion from trains, traffic, air-conditioning units and water pumping equipment was recorded with an engineering seismograph. Vertical geophones of natural frequency 2, 8, and 50 Hz were used, as well as a 4.5 Hz horizontal geophone. Train noise exhibits strong peaks in the 5-10 Hz range, both for the vertical and horizontal geophones. Noise peaks at 25, 60, 90 and 115 Hz are probably related to traffic, electrical transformers, and air-conditioning units. Measurements made near a sewage treatment plant in DeKalb showed peaks at 10, 30, 93, and 109 Hz, probably related to pumping and mechanical equipment. Seismic noise collected about 100 m from municipal swimming pool with vertical geophones showed spectral peaks at about 10-30 Hz and 180 Hz. Horizontal geophones at the same location exhibited peaks at 5-15 Hz and 140 Hz. Pool activities, pumping equipment and traffic may contribute to this noise. Delineating particle motion at the recording sites will determine whether dominant motion is due to Rayleigh, Love or body waves. Geophone arrays may also be deployed to determine the origin direction and velocity of each seismic noise component.
S13C-1075 1340h
High Quality Seismological Recordings From the SN-1 Deep Seafloor Observatory at the Base of Mt. Etna
We show results from the SN-1 seafloor multidisciplinary observatory 4 month long experiment, off-shore Eastern Sicily (Italy) at 2105 b.s.l.. SN-1 3 component broadband seismometer shows background noise levels above 0.1 Hz, similar to the ones found at a nearby ground station. We find that tilt of the sensor induced by sea currents is a noise source below 0.01 Hz. Seismic noise levels above 0.01 Hz are strongly characterised by Etna volcanic activity. In spite of such intense noise sources, SN-1 could record a large amount of events on local, regional and global scale. The quality of seismic recordings confirms a valid installation procedure and ground coupling of the 3 component broadband sensor, suggesting that SN-1 can be proposed as permanent observatory in this high seismic and volcanic hazard area.