Recording teleseisms on the sea-floor; an example from the Juan de Fuca plate.

by Brian T.R. Lewis and LeRoy M.Dorman

January 1998

1. Abstract.

In 1991, during an experiment to compare low frequency seismic noise on a basaltic and a sediment covered sea-floor (NOBS), we recorded teleseisms on the Juan de Fuca ridge, the Gorda ridge and the adjacent Cascadia Basin with the SNAG ocean bottom seismometers (OBS). These data provide an indication of the type of data that may be obtained from future experiments to record teleseisms and may be helpful in designing these experiments and analyzing the results.

We found that although seafloor noise is dominated by microseisms in the band 0.1 to 0.3 Hz there is a well developed minimum in noise from about 0.03 to 0.1 Hz (the noise notch). It is in this noise notch that teleseisms can be most easily detected. In the Cascadia area the overall noise levels are such that only teleseismic events with magnitude greater than 6.5 were usefully recorded. A magnitude 6.6 event in the New Britain area (delta=89 degrees) produced usable P and surface wave data only in this noise notch. Above 1 Hz Stoneley waves, generated by scattering of acoustic waves, propagate energetically in unconsolidated sediments at the seafloor and contribute a large component of noise to vertical seismometer sensors.

In the band .03 to 0.1 Hz the character of compressional waves is very sensitive to water depth and the type of sensor. We show that pressure sensors are especially sensitive to reverberation in the ocean and that motion sensors (seismometers) are less sensitive to ocean reverberations and will record teleseismic phases with less distortion than pressure sensors.

The Cascadia data indicate enhanced P amplitudes at sites on the ridge axes which could be due to focusing caused by a low velocity lens. These data suggest that amplitude information may be as, or even more, useful than P delay times for determining upper mantle structure.

2. Introduction and objectives.

The structure of the oceanic crust has been the subject of considerable study with seismic methods over the past two decades. Although much has been learned about the physical properties of the crust from these studies, and as a result inferences about processes of generation of the crust at mid-ocean ridges (MOR) have been greatly refined, little is known about physical properties in the upper mantle to depths of hundreds of kilometers under MOR. This is because the velocity of the upper mantle under MOR is likely to be low, which mitigates against standard refraction methods, and the depth is sufficiently great that standard reflection methods are inadequate. The use of Love waves has been most successful to date in terms of getting the general velocity structure under MOR, Zhang and Tanimoto (1992). On land the use of the teleseismic P delay times and tomographic inversion methods have contributed to knowledge of the velocity structure under features beneath the western United States ( Humphreys and Dueker, 1994 ) and to strong constraints on the depth of origin of these features. On the seafloor similar studies have not been done because the instrumentation has not been available and knowledge of the noise levels was sparse. This situation is now changing with the advent of Ocean Bottom Seismometers (OBS) with large bandwidth and data storage .

The upper mantle under the oceans is and will be a focus of much study. Example are the 1994-1995 LABATTS (Lau Basin Transmission Tomography Study) (Zhao, Xu, Wiens, Dorman, Webb, Hildebrand, 1997) and the 1995-1996 MELT (Mantle Electromagnetics and Tomography) ( Forsyth and Chave 1994) experiments, in which the goal is to use teleseismic data to determine if the mantle upwells under the East pacific Rise in a narrow conduit due to dynamic flow, or whether it upwells over a broad area as a passive response to spreading.

The general objective of experiments like LABATTS and MELT is summarized in Figure 1 . In these experiments it is expected that teleseisms will be recorded that will allow the use of P delay times from different distances and azimuths to constrain the upper mantle structure. This paper reports results from an experiment in which similar objectives can be addressed in a limited way. Teleseismic data were recorded in 1991 during the NOBS experiment (Noise On Basalt and Sediment), even though the primary objective of this experiment was to compare noise on basaltic seafloor with noise on seafloor with various sediment thicknesses.

3. Experiment layout

In the NOBS experiment 30 OBS's belonging to the Office of Naval Research (ONR) (Jacobson et al 1991) and operated by WHOI , SIO and UW, were deployed in four arrays on the Juan de Fuca plate. These OBS's record the output of three orthogonally oriented 1 Hz seismometers and a differential pressure gauge and are capable of recording data from .01 Hz to at least 10 Hz. The location of the arrays is shown in figure 2 . The arrays generally were two dimensional and covered an area of about 1 to 2 km2 with inter-instrument spacing of order of a hundred meters. The JDF array represents a sediment free site on the Juan de Fuca Ridge, Cascadia Basin array represents a thickly sedimented site with 1000 m sediment and was the location of RV FLIP, the BLANCO array was on a flat site at the intersection of the BLANCO fracture zone and the Gorda ridge and had about 10 m sediment, and the FLANK array was to the east of the Gorda ridge at a site with about 100 m sediment.

The instruments were set with a sampling rate of 32 samples/second to record higher frequency noise. Even with 383 megabytes of data recording capacity this sampling rate precluded continuous recording for a month so the OBS's were run on a 50% duty cycle Recording started on Sept. 7 1991 and ended on Oct. 11 1991. The data were searched for teleseismic events that occurred during the recording period, using the IRIS-DMC event listing as a guide. Only one event, of magnitude 6.6 in New Brittain, had a signal to noise ratio that allowed identification of the P wave. Table 1 shows events of magnitude greater than 6 that occurred between July 7, 1991 and Oct. 19, 1991, giving an indication of the probability of having an event with magnitude greater than 6. The remainder of the paper describes the noise levels on the seafloor and the analysis of the data from this event.

The results are best described in terms of the noise levels in the ocean, the effect of the ocean on teleseismic signals, and character of the data recorded from the New Brittain event..

Noise on the seafloor

Noise levels on the seafloor are dominated by pressure variations induced by motions at the seasurface. In the frequency range .01 to about .06 Hz infra-gravity surface waves dominate the noise. The response of the seafloor to this noise largely depends on the compliance of the seafloor (Crawford, Webb and Hilderbrand 1991). Hence noise levels on a vertical seismometer and the relationship between noise on the pressure sensor and noise on the seismometer are dependent on the nature of the seafloor.

From about .1 to .3 Hz the noise is dominated by Rayleigh waves with a speed of about 2 km/sec and a frequency twice that of the causal seasurface waves . The wavelength of the noise in this band is about 10 km and this is the so-called microseism noise band. In this band, seafloor noise is correlated with local swell. At frequencies higher than about 0.8 HZ, the seafloor noise correlates with locak wind (Dorman, Schreiner, Bibee and Hildebrand, 1993)

From about 1.0 to 10 Hz motion of the seafloor is dominated by very low velocity interface waves (Scholte or Stoneley waves) caused by scattering of pressure waves at and within the seafloor (Tuthill and Lewis, 1981; Dougherty and Stephen, 1988; Screiner and Dorman, 1990). The speed of these waves is of order 50 to 100 m/sec, giving a wavelength of order 100 m or less. These waves primarily affect seafloor seismometers and not seafloor pressure sensors. This is supported by our data and by the numerical modeling experiments of Dougherty and Stephen, 1988. The consequence of this is that above 1 Hz the vertical seismometer channel is considerably noisier than the pressure sensor

The best signal to noise ratio for teleseisms is to be had in the range .03 to .08 Hz. This can be seen in figure 3 , which shows a typical NOBS pressure spectrum of noise before teleseismic arrivals together with the noise plus the signal from a magnitude 6.5 event in New Brittain. In fact this was the only teleseismic event of sufficient size to be recorded with a usable signal to noise ratio during the month long recording period.

Teleseismic data

The New Brittain event occurred on Sept. 28 1991 at 20 hr 26 m UT. The depth and magnitude were reported by the IRIS- DMC as 33 km. and 6.6. The epicenter was reported as 5.1 South and 150.6 E, a distance of about 89 degrees from the NOBS experiment.

The closest long period seismic land station to the NOBS site is Corvallis, and it is situated almost directly east of the NOBS site. Figure 4 shows a comparison of a pressure sensor at the Cascadia site with the 3 components of motion from Corvallis long period sensors in the band 0.02 to 0.07 Hz. One can clearly identify the Rayleigh Waves and the P phase on the OBS but other phases, such as S, are less clear. In this frequency band the horizontals from the OBS's did not prove to be useful and the vertical was similar to the pressure. Because waves and swell are ubiquitous on the surface of the ocean the noise levels on the sea-floor are generally much higher than at a quiet land site.

Rayleigh waves from all four OBS sites and Corvallis are shown in figure 5 and indicate that the waveforms are sufficiently stable that dispersion properties could be calculated, but only in the noise notch band.

Effect of water depth on teleseismic P waves

Because of the large velocity difference between the ocean and the underlying rocks reverberation of acoustic waves in the water column can have a significant effect on the wave forms of teleseismic P waves (Blackman, Orcutt and Forsyth 1995) We demonstrate this by computing synthetic seismograms for a very simple case, an oceanic water layer overlying a half space with mantle like properties. In these computations we also used the impedance relationship between pressure and displacement to allow comparable computations of pressure and displacement. The impedance relationship in water is P = -i*(rho)*(alpha)*(2pi)*f*Uz;

where P=pressure, Uz =vertical displacement; f=frequency; rho=density and alpha=speed of sound in water.

Reflectivity synthetic seismograms (Kennett, 1983) were computed for P waves traveling almost vertically in the water and for several water depths. A sampling interval of 0.5 sec. and an impulsive displacement source function with time history +1,-1 was used for calculating seismograms of displacement. This approximates a step in moment as seen by a velocity sensor. For calculating seismograms of pressure the derivative of the displacement source function (the second derivative of a step) was used. The synthetic seismograms were then bandpass filtered from .01 to .08 Hz to simulate the teleseismic data.

Figure 6 shows the pressure and displacement seismograms for a 5 km thick water layer and the impulsive source. One can clearly see the reverberations associated with the reflections at the sea surface and the seafloor in the pressure seismograms. The reflections are labeled t1, t2, t3 and they are larger than the incident pulse t0 because of reinforcement at the seafloor. Note also the inversion due to the reflection at the seasurface. On the displacement seismograms the reverberations are present but much smaller because of phase cancellation at the seafloor. (A downgoing wave with positive displacement is reflected at the seafloor with a negative displacement. The incident and reflected waves tend to almost cancel with the result that a displacement sensor is far less sensitive to the reverberations than a pressure sensor.).

Figure 7 shows the displacement seismograms bandpass filtered to simulate the teleseisms and, as one expects, the P wave is little affected by the water depth. However in figure 8 we show the same situation for the pressure seismogram and here one sees a dramatic affect of water depth on the record. The water depth not only change the amplitudes but also changes the relative arrival times of the P wave.

Confirmation of these calculations can be seen in data from a small earthquake recorded by a NOBS instrument (figure 9 ). This figure shows the direct P arrival and it’s sea-surface reflection as recorded by a pressure sensor and a vertical seismometer. On the pressure sensor the surface reflection is larger than, and inverted with respect to, the direct P arrival. On the seismometer the surface reflection is essentially nonexistent, because of the cancellation effect at the seafloor.

The resonance frequency of the ocean can be approximately expressed by

fr = (water depth * 4)/(water velocity).

For a13 second period P wave (0.078 Hz) there is a maximum in the resonance curve at 5 km water depth whereas for 2.5 km water depth there is a null in the resonance curve. Clearly water depth will be a major factor in the distortion of teleseismic P waves recorded by pressure sensors on the seafloor. This distortion is much less of a problem for P waves recorded by vertical component seismometers.

P waves from the New Brittain event

P-waves from this event are clear on the pressure sensors and highly coherent across pressure sensors in the individual arrays, but only in the noise-notch, 0.03 to 0.08 Hz, . Examples from the BASIN and BLANCO arrays are shown in figures 10a and 11a . As can be seen from these figures the amplitudes on the pressure sensors are very stable across the arrays.

In this same frequency band the vertical seismometer data are more variable. At the BLANCO array, with about 10 meters of sediment, the data are fairly consistent across sensors (figure 10b ) whereas at the BASIN array, with about 1000 meters of sediment) the data are much more variable (figure 11b ). We attribute this to the thick sediment section under the FLIP sensors. This sediment section will have a relatively large compliance compared to basalt and would result in larger noise levels on the vertical seismometers.

P wave delay times

In figure 12a and 12b are plotted P waves from each of the sites after making a correction for the travel time through a standard earth model. In these figures the amplitudes scales of the seafloor data are uniform. Because of the relatively poor signal to noise ratio and the low frequency of the data it is very difficult to determine delay times to better than a second. The data do indicate that the relative delay times are not more than 1 to 2 sec. To improve delay time resolution in these experiments will require either much larger events or much quieter sites.

P wave amplitudes

Comparing amplitudes between arrays we note that ridge sites (JDF and BLANCO) have larger amplitude P waves than the off axis sites (FLANK and BASIN) both for the vertical seismometer data and the pressure data. Although these amplitude variations are more pronounced on the pressure data, they cannot be explained only on the basis of water depth differences. The water depths at each site were; JDF 2350m, BASIN 3015m, BLANCO 3010M and FLANK 2750m. From the synthetic seismograms in figure 7 and 8 we would not expect large amplitude differences between these sites. and we would expect JDF and FLANK to have the smallest amplitudes. What we see in the data in figures 12a and 12b is that the two sites closest to the ridge axes, JDF and BLANCO, have the largest amplitudes. A possible cause is focusing caused by a low velocity lens in the upper mantle velocity under zero age crust.

5. Discussion and Conclusions

The main purpose of this paper was to evaluate the capability of existing OBS technology to record teleseismic earthquake data. Such data are clearly important for understanding processes in the upper mantle under the oceans.

We have found that the seafloor is in general a noisy environment when compared with quiet sites on land. In particular microseisms generated by seasurface waves contribute much noise from 0.1 to 0.3 Hz. Above 1 Hz Stoneley waves, generated by scattering of acoustic waves, propagate energetically in unconsolidated sediments at the seafloor and contribute a large component of noise to vertical seismometer sensors.

The best signal to noise ratio for teleseisms is to be found in the band 0.03 to 0.1 Hz. In this band surface waves and body waves can be usefully recorded, provided the source event is at least of size magnitude 6 or greater. Below 0.03 Hz infra gravity waves overwhelm teleseismic surface waves and above 0.1 Hz microseismic noise overwhelms teleseismic body phases.

Seafloor pressure sensors are sufficient to record teleseismic P-waves, Rayleigh waves, and some S phases that are converted to P before they reach the seafloor, but there can be considerable distortion of amplitude and phase due to reverberation in the ocean. Seafloor seismometers can record teleseismic phases with less distortion from reverberation than pressure sensors, but coupling and leveling can be a problem here.

The data from the NOBS experiment suggest that events with magnitudes greater than about 6.5 and sites where microseismic noise is low (low ocean swell) will be necessary to acquire teleseismic data that can provide useful information about the upper mantle. Amplitude data could be most diagnostic about low velocity lenses in the upper mantle.

6. Acknowledgments

This experiment was supported by the Office of Naval Research under contracts N00014- 90-1881 and N00014-91-j-1336 under the Marine Geology and Acoustics Programs

7. References

Blackman, D.C., J.A. Orcutt and D.W. Forsyth, Recording teleseismic earthquakes using ocean bottom seismometers at mid-ocean ridges. Bull. Seismol.Soc Am,. Accepted.

Cox,C.S., Deaton,T., and Webb,S.C., A deep-sea differential pressure gauge: J.Atmos. and Ocean Tech., 2, 237-246, 1984

Crawford, W.C., S.C. Webb and J.A. Hilderbrand, Seafloor compliance observed by long- period pressure and displacement measurements: J.Geophys.Res., 96,16151-16160, 1991

Dorman,L.M, Schreiner,A.E, Bibee,L.D., and Hiderbrand,J.A., 1993, Deep-water seafloor array observations of seismo-acoustic noise in the eastern Pacific and comparisons with wind and swell, in B. Kerman, Ed., Natural Physical Sources of Underwater Sound, Cambridge, England, Kluwer,.

Dougherty M.E. and R.A Stephen, Seismic energy partitioning and scattering in laterally heterogeneous ocean crust. PAGEOPH, 128, 195-229, 1988.

Forsyth, D.W. and A.D. Chave, Experiment investigates magma in the mantle beneath mid-ocean ridges, EOS, 75, 537-540, 1994

Humphreys, E.D., and K.G. Dueker, Western U.S. upper mantle structure, J. Geophys. Res., 99, 9635-9650, 1994.

Jacobson,R.S., Dorman,L.M., Purdy,G.M.,Schultz,A., and Solomon,S.C., Ocean Bottom Seismometer facilities Available, EOS, Trans. of AGU, 72,506-515. 1991.

Kennett,B.L.N., Seismic wave Propagation in stratified media: Cambridge University Press, 342pp, 1983.

Liu, J., Schmidt,H. and Kuperman, W.A., Effect of a rough seabed on the spectral composition of deep ocean infrasonic ambient noise: J. Acoust.Soc.Am., 93, 753-769, 1993.

Sauter, A.W., Hallinan, J., Currier, R., Barash,T., Wooding, B., Schultz, A., and Dorman, L.M., Anew ocean bottom seismometer: Proceedings of conference: Marine Instrumentation ‘90, Marine Technology Society, 99-104, 1990.

Schreiner,A.E. and Dorman,L.M., Coherence lengths of seafloor noise: Effects of seafloor structure: J.Acoust.Soc.Am, 88, 1503-1514. 1990.

Tuthill,J.D., Lewis, B.T.R., and Garmany,J.D., Stoneley waves, Lopez Island noise and deep sea noise from 1 to 5 Hz: Mar. Geophys.Res., 5, 95-108, 1981

Zhang, Y., and T. Tanimoto, Ridges, hotspots and their interaction as observed in seismic velocity maps. Nature, 235, 45-49, 1992

Zhao, D., Xu, Y., Wiens, D.A., Dorman, L.M., Hidebrand, J., and Webb, S., in press for issue of October 1997, Depth extent of the the Lau back-arc spreading center and its relationship to subduction processes: Science, 1997

8. Figure captions

Figure 1. Hypothesis

Figure 2. Location of seismometers

Figure 3. Spectrum of Noise and Earthquake signal.

Figure 4 comparison of COR with OBS12 pressure, p,s, and rayleigh

Figure 5 comparison of Rayleigh waves

Figure 6. Synthetic seismograms for impulse source

Figure 7. Synthetic displacement for band pass signal at 3 depths

Figure 8 Synthetic pressure for band pass signal at 3 depths

Figure 9. Comparison of vertical and pressure sensors using the P-wave from a local earthquake

Figure 10a P-wave across BLANCO array (pressure)

Figure 10b P-wave across BLANCO array (vertical)

Figure 11a. P-waves across BASIN array (pressure)

Figure 11b. P-waves across BASIN array (vertical)

Figure 12a. P-wave amplitudes and delay time (pressure)

Figure 12b. P-wave amplitudes and delay time (vertical)

Table 1. List of events

Brian T.R.Lewis

University of Washington, WB-10

School of Oceanography

Seattle, WA, 98195

LeRoy M. Dorman

University of California

Scripps Inst. of Oceanography

La Jolla, Ca. 92093