INTRODUCTION
The seismic sources used in marine geophysics have evolved significantly during the decades since the second World War in response to advances in technology and to the increased appreciation of the effects of loud sounds in the sea upon animal life. One branch of this evolution has led to the use of repetitive sources (airguns and water guns) fired at shallow depths. For seafloor interface wave or detailed refraction experiments, however, sources need to be on or near the seafloor, (Davies, 1965; Whitmarsh and Lilwall, 1982; Sauter and others, 1986; Stoll, 1989; Berge and others, 1991; Caiti, and others, 1991; Chapman, 1991; Christensen, Purdy and Fryer, 1992; Dorman, 1997; Hammer and others, 1994; Wiggins and others, 1996; Christeson, Purdy and Fryer, 1994). For sources on the seafloor, no satisfactory substitute for explosives has yet been found.
Accurately timed and navigated seafloor shots have been used in our lab for a number of years in refraction experiments to image the subbottom structure (Wiggins, 1996; Hammer, 1994) and in Scholte wave studies of and Dorman, 1990; Dorman, Schreiner and Bibee, 1991; Schreiner, Dorman and Bibee, 1991; Bibee and Dorman, 1995; Nolet and Dorman, 1996; Dorman, 1997). Explosives have been used as the energy sources for these studies.
We propose to continue, and hopefully complete, development of an implosive seismic source which can be used on the seafloor. This will have the advantage of concentrating seismic energy at lower frequencies than explosions and will eliminate safety and regulatory concerns as well. It should replace small explosions in usages such as generating Scholte (interface) waves and perhaps be useful in seafloor refraction experiments. Advantages of having the source directly on the bottom include shrinking the footprint area of ray entry and the enhanced excitation of shear waves in comparison with sources some distance above the seafloor.
PAST WORK WITH IMPLOSIONS
Researchers have long recognized the usefulness of collapsing water voids at depth to produce seismo/acoustic signals. In the mid 50's John Isaacs and Art Maxwell (Isaacs and Maxwell, 1952) attached glass floats to piston sediment corers using a mechanism designed to break the float when the the corer hit the bottom. Listeners on board would retrieve the core when they heard the pop. The floats used by Isaacs and Maxwell had a volume of &approx& 0.01 liter. More recently, Urick (1963) recorded acoustic signals from the implosion of small bottles (up to &approx& 4 liters). Orr and Schoenberg (1976) examined the potential of spherical glass high-pressure In their tests, they dropped weakened glass spheres through the water column, recording the implosion, and analyzing the sounds emitted. In all the cases above, the primary interest was in the audible part of the seismo-acoustic spectrum, rather than the subaudible frequencies used for seismic studies of the crust and sedimentary column.
PROBLEMS LIMITING USE OF THESE SOURCES
For a non-explosive source to be useful as a bottom seismic source, it must be able to work on the bottom and it must be initiated at the seafloor by an accurately timed signal.
In the work of Urick and of Orr and Schoenberg, the sources were lowered into the sea until the increasing water pressure caused failure. This lack of control of the depth (making it impossible to initiate an implosion on the seafloor, rather than some uncontrolled distance above it) and the timing of breakage has been difficult to overcome.
There are unpublished reports of efforts to initiate failure of Benthos floats by driving a punch into the sphere from the outside. Evidently the sealing properties and the behavior of glass under high pressure prevented reliable initiation of collapse. There is a photo of a glass sphere with a punch which had been imbedded in the wall at depth but which had not caused collapse. This is truly impressive performance for a float but not what is required for an acoustic source.
SEISMIC PERFORMANCE OF AN IMPLOSION SOURCE
None of the treatments of implosive sources listed above made calculationsof source behavior, other than comparing total radiated energy with the potential energy available. Nolet and Dorman (1996) have calculated the seismic moment for a deep explosion. In that treatment, the significant source parameter at frequencies well below the bubble frequency is shown to be the change in volume between the solid explosive and the equilibrium size of the gas bubble. That treatment is also applicable to an implosion, with the primary difference being that the volume change is negative. A modified version of that argument is presented here.
In an underwater explosion, the explosive charge is rapidly converted into a bubble of gaseous detonation products (Cole, 1948). The pressure within this bubble is much higher than the ambient water pressure and the bubble increases rapidly in size. After the bubble expands to its equilibrium size, the momentum of the outward-moving water mass causes the bubble to expand to a size larger than that necessary to make the gas pressure equal the hydrostatic pressure in the water and the pressure deficit causes a restoring force which makes the bubble smaller. The size of the bubble thus oscillates for a few cycles. During this oscillation, the bubble radiates acoustic energy, especially when the bubble walls undergo large accelerations during the minima of the bubble size. As energy is radiated, the range of oscillation decreases and the bubble volume approaches its equilibrium size at the ambient pressure.
Several time scales are important in underwater explosions. The first is that of the detonation itself. The velocity of detonation of TNT (trinitrotoluene) is about 6.9 km/s and the radius of a 2.27 kg (5 lb) sphere of TNT is 0.069 meters so the time scale of the detonation process is about 0.01 microseconds. The second significant time scale is the period of bubble oscillations. At a depth of 3800 meters this is 2.9 milliseconds from the theory of the adiabatic expansion of an ideal gas (Arons 1948).
The third significant time scale is the round-trip travel time of an acoustic wave between the shotpoint and the sea surface. This separates the purely dynamic regime from that where the concept of hydrostatic pressure has significance. The fourth is the time of the cooling of the detonation products and their dissolution into water. The time scale of the dissolution is a few minutes (See Figure 3 of Hammer et al., 1994).
To model the low-frequency spectrum of an explosion, we need to calculate the moment, which is proportional to the volume change caused by the source. This volume is the difference between the volume initially occupied by the unexploded charge and the volume occupied by the combustion products at the end of the process. We are interested in the spectrum in the 0.3-3 Hz frequency range, since this is the range in which we observe Scholte waves. The time scale of interest is thus the 0.3-3 second range, much larger than the detonation and bubble oscillation time scales but smaller than the surface reflection and dissolution time scales, so we can treat the moment time dependence as a step function in time.
The theory of volume seismic sources is treated by Aki and Richards (1980), at the end of chapter 3. They consider a spherical volume source of radius a as it carries out a transformational expansion. For this case, the moment M is given by their equation 3.34, (equations omitted until the web can better handle them)For a shot of 5 pounds (2.27 kg), the scale length from the adiabatic theory is 0.309 meters (Arons, 1948). The equilibrium bubble radius is 0.62 times the scale length so the equilibrium volume is 0.0296 m **3.
The shots for this experiment were placed in pressure cases which were about half full. Since the explosive volume is 0.00138 m **3 we will take v0 to be 0.00276 m **3. We take v1 to be the equilibrium gas volume from the ideal gas theory, 0.0296 m **3.
We are using a source depth of 1 meter within the sediments, where k is 3.15 times 10 **9 Nt /m **-2. Thus the moment tensor is the diagonal identity tensor times 7.62 x 10 **7
The adiabatic bubble theory is widely used in practice for computing the relationship between bubble oscillation period and source size, but the measurements used for calibration were largely taken at much shallower depths than the 3800 meter depth of this experiment. For this reason, another independent, though also empirical, estimate of the bubble size was made.
In an explosion, the detonation wave compresses the solid explosive, moving along the solid Hugoniot in PV space. The exact trajectory depends on the geometry of the problem. When detonation occurs, the locus in PV space was taken to move to the Chapman-Jouguet point, which allows the minimum detonation velocity satisfying conditions for steady-state propagation of the detonation. From the C-J point, the isentrope of the detonation products of TNT was followed down to a pressure of 380 bars. These calculations were made using the Becker-Kistiakowski-Wilson (BKW) equation of state developed at Los Alamos Scientific Laboratory (Mader, 1979). At 380 bars, the specific volume is 0.00945 m **3/kg. This yields a gas volume of 0.02145 m **3, a slightly smaller value than produced by the adiabatic theory.
To match the data presented in that paper, (Scholte waves at distances of 400 meters and 1070 meters) we had to adjust calculated moment upward by a factor of 1.5. Since seismic source theory (as well as cosmology) is best treated by logarithmic laws, we view this level of agreement as "exact".
Since the equilibrium volume of a gas bubble produced by an explosion is pressure dependent, the performance (as measured in seismic moment) of explosive sources decreases with depth. The moment of an implosive source, however, is not depth-dependent since it is determined by the volume of the container. We have calculated comparisons between the seismic moments of small explosions at a range of depths and present these in the following figures.
Figure 1 Compares the bubble volume from a 0.5 lb of TNT with the volume of our prototype 20 liter (0.02 m **3 ) source. Figure 2 shows the same data in units of scalar seismic source moment. Note that the effectiveness of the two sources is the same at about 500 meter depth, and that the implosion source is superior at greater depths.
PROGRESS WE HAVE MADE SO FAR
We have developed and are in the process of testing a pressure-induced seismic source (PISS) gun which consists of an evacuated cylinder with a ceramic plate blocking an opening on the endcap. We break the plate by electrically firing a projectile from the inside of the pressure case into the plate. This initiates an implosion as seawater at ambient pressure floods the cylinder; similar in effect to a watergun. We lowered th PISS gun on the .322 CTD cable on both the deployment and pickup cruises.
October 31, 1996, after deploying all the instruments and navigating their positions, we steamed to 650 meters of water and lowered the Piss gun to 630 meters. It had a 1/2" thick piece of ceramic plate attached, with a calculated breaking depth of 560 meters. An ohm meter, monitoring resistance between the 2 leads into the pressure case, indicated that the plate did not break. We brought the PISS gun up, steamed to a deeper site, and relowered the gun. The plate did not break again even when lowered to 750 meters. We ceased operations and headed home. We broke the ceramic plate used on this trip in the MPL pressure tank. It broke at 1700 psi, equivalent to 1156 meters depth. Four other 1/2" plates were broken in the tank, all broke at lower pressures from 450 to 1120 psi.
On the evening of November 22, 1996, after recovering all the OBS's, we steamed out to 840 meters of water and lowered the PISS gun with a 1/2" ceramic plate down to 600 meters. It did not break, so we brought it up and rigged the glass breaker inside and relowered it. At 327jd 02:18:00Z we fired the breaker, which propelled a diamond-tipped nail into the plate. No audible signal could be heard, but a recording on the TEAC DAT recorder showed a signal coincident to firing time. When the gun was raised back on deck, it was filled with water and the plate was cracked. We rigged the PISS gun with another 1/2" ceramic plate, lowered it to 600 meters, and fired the glass breaker at 327 03:19:00Z. Back on deck, the gun was full of water and the plate was cracked in half.
These results were presented at the 1996 fall AGU meeting in a poster: Development of a seismo/acoustic implosive source .
PROPOSED WORK
The problems remaining to be addressed are the following:
(1) We do not know how close to the ultimate breaking pressure the plates must be subjected to in order to allow reliable breaking by the Glass Popper. This will require a run of tests in a pressure tank.
(2) We do not know the conditions required to produce shattering (Figure 3 ) rather than cracking. The upper end of the source spectrum is determined by the time required for the chamber to fill with water. When shattering occurs, filling is quite rapid. We do not know the filling time when cracking occurs.
(3) We do not know how large a chamber can be "rapidly" filled by the 4 inch by 6 inch plates which are available "off the shelf". Special production runs of larger plates will be more expensive. We propose to make a larger chamber than our present 20-liter device.
(4) On the test which produced the spectacular shattering, there was some damage to the sealing surface of the chamber endcap. We do not know if this is due to the small size of the test pressure tank (plate fragments bouncing back from the walls) or if it is a direct result of the plate failure.
We propose to attack these problems by the following sequence of tests: (1) Question (1) requires a series of pressure tank tests using a low-volume implosion chamber. (2) Questions (2) and (3) can be addressed by a combination of at-sea tests and perhaps some finite-element shock calculations. The combination of small-volume pressure tank and small-volume imploder chamber does not allow laboratory testing of these questions. One thing we will also try is using an oblong hole in the end cap to expose more of the area of the plate to hydrostatic pressure and increase the area of the filling orifice exposed when the plate breaks. (3) Question (4) will require deep-water tests. If seal damage is a direct result of plate failure, we will test the use of a protective plate placed between the endcap and the ceramic plate (sealed on both sides with O-rings).
In all the sea tests, we will monitor the acoustic signature with a deep hydrophone, so that we can get valid spectra down to a Hertz or so. In the later tests, we will monitor the results with OBSs, to see that the source is functioning as required.
EXTENSIONS
It is clear that this source has directional properties. Thus far we have fired it only in midwater and have chosen a vertical orientation for the 7"x4' pipe chamber. If the chamber were laid on its side and coupled to the seafloor, perhaps by fins or short legs, significant horizontally polarized shear radiation should occur. This could provide a useful SH seafloor source. We propose to attack this question when the other questions are under control.
There seems nothing to inhibit the construction of a 6-shooter or a 9-shooter, allowing multiple detonations on a single wire-lowering. The controlelectronics used in the Hess Deep field work would probably be suitable for controlling multiple detonations.
REFERENCES:
Aki, K. and Richards, P. G.; Quantitative Seismology: Theory and Methods; W. H. Freeman and Co. San Francisco, Calif.,1980.
Arons, A. B.; Secondary Pressure Pulses Due to Gas Globe Oscillation in Underwater Explosions. II. Selection of Adiabatic Parameters in the Theory of Oscillation; J. Acoustical Soc. Am.; V 20; P 277-282; 1948.
Berge, Patricia A., Mallick, Subhashis, Fryer, Gerard J. A , Barstow, Noel, Carter, Jerry A., Sutton, George H., and Ewing, John I.; Refraction measurement of shear wave anisotropy in shallow marine sediments and implications for reflection processing Shear Waves in Marine Sediments; edited by J. M. Hovem, M. D. Richardson and R. D. Stoll; pp. 203-212; 1991; Kluwer Academic Publishers; Dordrecht, The Netherlands; 1991.
Berge, Patricia A., Mallick, Subhashis, Fryer, Gerard J., Barstow, Noel, Carter, Jerry A. Sutton, George H., and Ewing, John I.; In situ measurement of transverse isotropy in shallow-water marine sediments; Geophys J. Int.; V 104; pp. 241-254; 1991.
Bibee, L. D. and Dorman, L. M.; Full Waveform Inversion of Seismic Interface Wave Data; pp.377-382; in: Full Field Inversion Methods in Ocean and Seismo-Acoustics; edited by O. Diachok, A. Caiti, P. Gerstoft, and H. Schmidt; Kluwer Academic Publishers; Dordrecht, The Netherlands; 1995.
Chapman, N.R.; Estimation of geoacoustic properties by inversion of acoustic field data Shear Waves in Marine Sediments; in Shear Waves in Marine Sediments; edited by J. M. Hovem, M. D. Richardson and R. D. Stoll; pp. 511-520; Kluwer, Dordrecht; 1991.
Christeson, G. L., Purdy, G. M., and Fryer, G. J.; Structure of young upper crust at the East Pacific Rise near 9 deg 30'N; Geophys. Res. Lett.; V 19; pp. 1045-1048; 1992.
Christeson, G. L., Purdy, G. M., and Fryer, G. J.; Seismic Constraints on shallow crustal emplacement processes at the fast spreading East Pacific Rise; J. Geophys. Res.; V 99; pp. 17957-17973; 1994.
Davies, D.; Dispersed Stoneley waves on the ocean bottom; Bull Seismol. Soc. Am.; V 55; pp. 903-918; 1965.
Dorman, LeRoy M.; Encyclopedia of Acoustics; Edited by Malcolm J. Crocker; Chapter 37: Propagation in Marine Sediments; 409-416; John Wiley; New York; 1997.
Hammer, Philip T. C., Dorman, LeRoy M., Hildebrand, John A., and Cornuelle, Bruce; Jasper Seamount Structure: Seafloor Seismic Refraction Tomography; J. Geophys. Res.; V 99; pp. 6731-6752; 1994.
Isaacs, J. D. and Maxwell, A. E.; The ball-breaker, a deep water bottom signalling device; J. Marine Res.; V 11; pp. 63-68; 1952.
Koelsch, D. E., Purdy, G. M., and Broda, J. E.; A multiple charge deep deployed explosive source for high resolution seismic refraction profiles in the deep ocean; Proceedings Fall MTS Conference; Marine Technology Society; New Orleans; 1991.
Mader, Charles L.; Numerical Modeling of Detonations; University of California Press, Berkeley; 1979.
Nolet, G. and Dorman, L. M.; Waveform Analysis of Scholte modes in ocean sediment layers Geophys. J. Intl.; pp. 385 - 396; 1996.
Orr, M. and Schoenberg, M.; Acoustic signatures from deep water implosions of spherical cavities; J. Acoust Soc. Am.; V 59; pp. 1155-1159; 1976.
Sauter, A. W., Dorman, L. M., and Schreiner, A. E.; A study of sea floor structure using ocean bottom shots; in: Ocean seismo-acoustics, low frequency underwater acoustics; Edited by T. Akal and J.M. Berkson; Plenum; New York, 1986.
Schreiner, A. E. and Dorman, L. M.; Correlation Lengths of Seafloor Noise: Effects of Seafloor Structure; J. Acoust. Soc. Am.; V 88; pp. 1503-1514; 1990.
Schreiner, A. E., Dorman, LeRoy M., and Bibee, L. D.; Shear Wave Velocity Structure from Interface Waves at Two Deep Water Sites in the Pacific Ocean; in Shear Waves in Marine Sediments; Edited by J. Hovem, R. Stoll, and M. Richardson; pp. 231-238; Kluwer, La Spezia, Italy; 1991.
Stoll, R. D.; Sediment Acoustics; V 26 of Lecture Notes in Earth Sciences; 153 pp.; Springer-Verlag; Berlin; 198 9.
Whitmarsh, R. B. and Lilwall, R. C.; A new method for the determination of in-situ shear-wave velocity in deep-sea sediments; Proceedings of Conference "Oceanology International 1982"; 1982.
Wiggins, Sean, Dorman, LeRoy M., Cornuelle, Bruce D., and Hildebrand, John A. Hess Deep Rift Valley Structure from Seismic Tomography; J. Geophys. Res; pp. 22335-22353; 1996.
