TP Phase Observations at the PIDC

R. W. Cook and J. L. Stevens
Maxwell Technologies
1998 December

Abstract

Oceanic acoustic signals are frequently observed by stations in the seismic network of the prototype International Data Center. The IDC collects seismic, hydroacoustic, infrasonic and radionuclide data from the International Monitoring System which has been developed to monitor the Comprehensive Test Ban Treaty. Maxwell Technologies is currently investigating transfer functions for hydroacoustic to seismic energy conversion. We have observed numerous examples of the hydroacoustic T-phase signals converted to seismic P signals at the ocean/continent interface in many geographic areas. The signals originate from sources having mb(IDC) less than 4.0 to the great Mw 8.2 intraplate earthquake of 1998 Mar 25. Some of the signals have traveled far inland reaching distances greater than 1600km for the seismic portion of the path. Though the amplitudes attenuate proportionally to frequency when the hydroacoustic T converts to the seismic P, TP waveforms do retain envelope characteristics similar to that of the oceanic T portion. Events greater than mb(IDC) = 4.6 have a reasonable chance of being observed as TP at an IDC seismic station somewhere on the Earth at a distance of less than 30 degrees. NORSAR and ARCES, in Norway for events in the far northern oceans, ESDC, Sonseca, Spain for the northern mid-Atlantic, BOSA, Botswana, South Africa for the southern Atlantic and Indian Oceans, PLCA, Paso Flores, Argentina and LPAZ, La Paz, Bolivia for portions of the southeast Pacific and many others may provide a useful auxiliary network for oceanic events. DBIC, Dimbokro, Africa often detects events where mb(IDC) is less than 4.0 out in the nearby central mid-Atlantic ridge. The Mw=8.2 Antarctic earthquake produced TP phases at the ASAR, Alice Springs, and WRA, Warramunga seismic arrays in Australia. Signals have also been seen at BGCA, Bogoin, Central African Republic, well inland and at BDFB, Brasilia, Brazil. While observation of TP phases is not a new phenomena, their prevalence and extent of continental penetration may be. It is conceivable they are responsible for some detections and might even combine with legitimate arrivals to form bogus events. TP signals are often strong and may be as useful as some of the Earth's interior seismic phases (eg. PP, PPP, PKKP) for use in event location and identification. More attention should be given these phases in future analysis and event bulletins.

Examples of Far Inland TP Phases

The figures that follows, though of somewhat reduced quality from time considerations, show waveforms and some of their characteristics, maps of the source to receiver path, and bathymetry and topography along the path. The part of the path on land is shown in red on the maps, the part in water in white. The number "AV" which is marked on several of the cross sections is the apparent velocity of the water calculated by predicting the P-wave travel time on land, subtracting this from the total time, and then dividing the distance in the ocean by the remaining time. If propagation is occurring as P-waves on land, and the wave takes the predicted great circle path, then the apparent velocity should be close to the average water velocity of 1.48 km/second, and this is the case for most of the paths.

For this 1997/12/29 earthquake south of Australia, a montage has been created. The panels in the montage might be the results of appropriate tools available to an analyst. TP phase arrivals have traversed rather simple paths for a relatively small Southeast Indian Rise event. Continental penetration is assumed to occur at the SOFAR depth of 700 meters. Assuming a P travel time for the continental portion of the path and computing the T portion velocity yields 1.45, 1.37, 1.42 and 1.46 km/sec for stations WOOL, ASPA, TOO and STKA respectively. Stations STKA and ASPA are in the Primary Seismic Station network of the International Monitoring System. WOOL was a Primary 3-component station of the GSETT3 experiment while both WOOL and TOO are now not IMS seismic network stations, either primary or auxiliary.




















Figure 2. Path from earthquake of 1998/09/30 recorded at PLCA.

Figure 3. Waveforms from earthquake of 1998/09/30 recorded at PLCA.


Figure 4. Bathymetry and topography of path from 1998/09/30 earthquake to PLCA.


Inherent to the analysis, the arrival of TP as a relatively high frequency signal (~1-10Hz) often following Rayleigh motion invokes the idea that it may be an unusual component of the Rayleigh wavetrain. These two events observed by PLCA to the west and south should dispel this thought and support the TP phase identification. Note how the TP from the western event arrives considerably after the initial low frequency Rayleigh arrival. This is a result of the low continental to oceanic travel path ratio. Not so for the southern Drake Passage event as the TP has arrived nearly simultaneously with the early Rayleigh. Calculating the total path in water at 1.48 km/sec plus the predicted travel time for a 'P' wave travel time on land seems to reconcile the observed total travel time for the TP arrival. Does this mean the path went around the Tierra Del Fuego peninsula? Consulting bathymetry, the depth of the water is ~100 meters and the ratio of continental to oceanic is greater than one-half. How is T, at apparently high efficiency (note ratio of T to P), transported across these shallow depths?



Figure 5. Path from earthquake of 1998/08/27 recorded at PLCA.



Figure 6. Waveforms from earthquake of 1998/08/27 recorded at PLCA.

For the Figure 7 map, original calculations of the TP junction location based on a visually picked waveform arrival yielded an entry that was too deep ie. too far below the SOFAR depth intersect with the continent. Near the continental shelf off the coast of Spain for that event observed 1998/07/09 along the mid-Atlantic ridge, the calculated entry was at a depth of four kilometers. When the profile was examined, the signal path was observed to have been interrupted by the presence of two Azores islands. The implication is the signal path was slightly greater on land than the model we were using assumed. As a result, the apparent entry was pulled further offshore. Similar model problems were observed for an event occurring in Drake Passage that crossed or went around Tierra del Fuego to be detected as TP at station PLCA in Argentina and for station MEEK in Australia to an event south of MacQuarie Island.



Figure 7. Path from earthquake of 1998/07/09 recorded at ESDC.

We would like to fully investigate the nature of these signal paths to determine what the signal path actually is. This would require finding additional events in these areas of sufficient size to produce TP and hopefully with sufficiently varying location of the source to provide for propagation raypaths around and through the islands of the above mentioned sites. In some cases, events with well defined P arrivals, that is quasi-explosive, have little or no crustal, surface wave arrivals. Yet, there are significant TP signals. In other cases, both types of signal are present and yet others the reverse is true. And of course, there are many cases where neither is present. Due to the natural filtering capability of the seismic regimen, few TP signals are expected for the very high frequency content of manmade explosive phenomena. Nevertheless, low frequency explosive content may be observed. Comparison of hydroacoustic signals to seismic counterparts may provide considerable screening potential. It may not be possible to prove that an event is an explosion, but it may be possible to prove it is an earthquake.

FK analysis on the early portion of TP waveforms seem to produce nice P velocities as shown; however, as the FK window is moved further into the wavetrain window the velocities take on a slower character and I believe this to occur from the multiplicity of refractive angles at the TP junction. Further into the waveform, chaotic arriving wavefronts render FK analysis invalid rather than an indicator of transverse motion. FK analysis prior to the TP arrival is that of the lower velocity surface waves. In fact, a continuously sampled FK window preceeding the TP may indicate an accurate starting time for the arrival of the TP phase. I do believe we should look for other forms such as TPP or maybe TS if we ever have the luxury of time; but I think continued processing of previous events and their analysis to see what is happening is VERY important in principle. Determining and graphically analyzing TP/P, TP/MS ratios could reveal discriminatory attributes, also. There may be times when either LR or TP are exclusive.


Figure 8. Waveforms from earthquake of 1998/07/09 recorded at ESDC.

Figure 9. FK result using a 40 second window at the beginning of TP phase.












A magnitude 5.19 REB earthquake in the south Atlantic Ocean, shows how many stations of the International Monitoring System might see an event and in this case covering both sides of that ocean. Though quite sizeable, penetration was well inland and received quite well with nice signal to noise ratios. BDFB in Brazil appears to have a double apparition of the event or is possibly mixed. The first arrival would have entered the solid crust at a point nearer the source.


Figure 10. Path from earthquake of 1998/01/03 recorded at several stations in Africa and South America.


Figure 11. Waveforms from earthquake of 1998/01/03 recorded at several stations both in Africa and South America.

Two depictions of the waveforms are shown, the first being true time aligned and the second aligned to the theoretical arrival time of a T phase traveling almost entirely an oceanic path. It is easy to see by this portrayal that the earlier the error of TP relative to the T, the more continental is the path.


Figure 12. Waveforms from earthquake of 1998/01/03 aligned on the T phase.

The following event in the south west Indian Ocean ridge area was very nicely detected by the South African station BOSA, the Antarctic MAW and western Australian station MEEK, the latter two maintained by Australia. MEEK appears to be an outstanding station for detection of TP phases. Iron mining occurs nearby. This event has generated comparatively and considerable high frequency energy, as evidenced by the 4-8 Hz band-pass filter although MEEK has an apparent proclivity for higher frequencies.


Figure 13. Path from earthquake of 1998/05/21 recorded at stations in Australia, Antarctica, and Africa.


Figure 14. Waveforms from earthquake of 1998/05/21 recorded at stations in Australia, Antarctica, and Africa.

The spectrogram of the waveforms received at BOSA, South Africa nicely shows the two distinctive longitudinal P phases while revealing little significant energy for the crustal waves LQ, LR. However, on the unfiltered and natural broadband channel there is little evidence of TP.


Figure 15. Spectrogram of 1998/05/21 earthquake recorded at BOSA showing a strong TP phase at 1500 seconds.


Spectrum of a noise sample (black) just prior to the P arrival, the P arrival (green) and the TP arrival (blue) all from a 30 second window at the time of the respective signals. Note the clearly defined energy passband revealed in the 1-8Hz passband for this event at BOSA.


Figure 16. Spectrum of P (green) and TP (blue) phases from 1998/05/21 earthquake recorded at BOSA.



Figure 17. 1998/05/21 Cross-section (Source to BOSA)


Figure 18. 1998/05/21 Cross-section (Source to MAW)


Figure 19. 1998/05/21 Cross-section (Source to MEEK)

Cross-sections of the signal path are provided for the three TP stations observing this event. Strangely enough, there may be some depth screening potential for TP. One might expect several aspects to be of importance. The expectation is that a deeper event will couple oceanically further down the path to the station and thus arrive earlier. Signal characterstics may be different, indeed, some feature revealing depth may even be extractable from the envelop of TP. For all cross-sections, the depths of the events are not known, that is their hypocenter location is restrained to the surface of Earth. The only event here displayed having depth is the one off the west coast of Chile seen by PLCA. For this type of discrimination we will not know unless we look.


Figure 20. Path from earthquake of 1998/01/07 recorded at MEEK, STKA, ARMA, YOU, TOO, MCQ.


Figure 21. Waveforms from earthquake of 1998/01/07 recorded MEEK, STKA, ARMA, TOO, MCQ (south of New Zealand).


Table 1. The following table lists all occurrences of continental "T" phases currently in the pIDC database.

DATE

TIME

LAT

LON

DEPTH

MB

ML

MS

STA

PHASE

DELTA

EVID

1996/11/01

14:38:05

-0.13

-18.02

0

4.46

5.39

4.03

DBIC

T

14.769

845700

1997/09/05

03:23:18

-56.21

-27.94

52

4.51

-

5.19

DBIC

T

65.489

1122787

1997/10/02

09:57:31

-24.96

-13.43

24

4.35

-

-

DBIC

T

32.534

1149445

1997/10/07

17:53:37

-52.17

15.40

25

4.78

-

-

DBIC

T

61.113

1155146

1997/10/22

22:20:29

-54.32

5.84

0

4.41

-

4.57

DBIC

T

61.421

1168106

1997/11/28

17:54:58

.27

-15.81

0

4.03

4.55

-

DBIC

T

12.647

1206850

1997/12/03

02:59:01

.02

-16.78

0

3.93

4.41

-

DBIC

T

13.605

1209657

1998/01/03

06:10:06

-35.13

-15.60

0

5.19

-

5.63

DBIC

T

42.791

1247912

1998/01/15

10:19:26

-29.41

-13.62

0

4.49

-

4.41

DBIC

T

36.851

1261344

1998/02/11

17:25:39

-34.00

-14.66

0

4.19

-

4.13

DBIC

T

41.500

1285207

1998/03/13

13:25:39

-1.44

-14.44

0

4.31

4.68

3.80

DBIC

T

12.510

1314521

1998/04/03

17:52:51

-6.88

-11.69

0

4.33

3.89

4.41

DBIC

T

15.087

1334240

1998/04/10

16:40:40

-1.66

-15.23

0

4.87

-

5.85

DBIC

T

13.254

1340832

1998/04/25

06:07:29

-35.37

-17.44

22

4.51

-

5.64

DBIC

T

43.467

1358647

1998/06/26

22:10:16

-59.30

-27.08

107

4.19

-

3.76

DBIC

T

68.116

19983775

1998/08/04

12:28:37

-53.02

21.87

0

4.83

-

4.96

DBIC

T

63.641

20038102

1998/08/04

17:48:21

-53.04

21.93

0

4.80

-

5.09

DBIC

T

63.676

20038287

 

Conclusions

TP phases are commonly observed from oceanic seismic events with mb > 4.6 at stations of the IMS network. They can sometimes be detected from relatively small events (mb < 4) and at stations as far as 1600 km from the coast.

Simple models of the propagation through the ocean and land often provide predictions of phase arrival times which agree remarkably well with the observed data. In other cases, however, the simple model is not adequate and it is clear that effects due to more complex propagation paths between the source and receiver and lateral heterogeneities will have to be considered in order to obtain accurate predictions of arrival time.

Additional knowledge acquisition of the behavior and relationships of TP signals may well provide enhanced monitoring of the Comprehensive Test Ban Treaty since propagation of signals of interest to the CTBT could coincide with those of TP.


Thanks to Wessel & Smith for GMT - integrally involved in the graphics.
Thanks to John Coyne for the waveform processing system Geotool.
Thanks to members of the Australian Geological Survey.


 

Cook, R W; cook@maxwell.com

Maxwell Technologies - Systems Division Reston Geophysics Office
11800 Sunrise Valley Dr., Suite 1212
Reston, VA 20191-5309 United States

 

Stevens, J L; stevens@maxwell.com

Maxwell Technologies - Systems Division
8888 Balboa Ave.
San Diego, CA 92123-1506 United States