Hydroacoustic Monitoring of the CTBT

Lindsay H Hall

Introduction

It has been known for a long time that low frequency sound can travel for very large distances in the sea1. This is usually due to the existence of a sound channel, the SOFAR channel, which effectively restricts the sound to depths where the sound travels without any interaction with either the sea surface or the sea floor. Without any interaction there is no loss of energy due to scattering or refraction into the sea bed and the only losses are the cylindrical spreading loss and the very small absorption loss.

Explosions, chemical or nuclear, are powerful sources of broadband sound and small (0.8 kgm) charges of TNT have been readily detected for many decades over transoceanic propagation paths from the low frequency sound they produce. Such a source recently produced a 20 dB signal to noise ratio on a single hydrophone after a 10,000 km propagation path2. Nuclear explosions, releasing many orders of magnitude more energy, are even more easily detected, the only qualifications being that there be no substantial land mass or extensive tract of shallow water between the explosion and the detector. Nor need the explosion take place in the sea: nuclear explosions within small islands have been detected by hydrophones thousands of kilometres distant.

Hydrophones are the most sensitive devices for picking up sounds in the sea but they are expensive to put in place, and their maintenance can be costly. An alternative to a hydrophone is a high frequency (0.5 to 20 Hz) seismometer near a steeply shelving coast. Such a seismometer responds to T phase, a compressional seismic wave generated by conversion of the incident sound at the boundary to the land. Such sensors are not as sensitive as hydrophones but they have been routinely used to detect submarine volcanoes and earthquakes from the underwater sounds these produce3. They are much cheaper to install and maintain than hydrophones.

Discrimination between transients due to man-made explosions and those due to natural events can be relatively straightforward. Earthquake T phase has less energy at the higher frequencies and a more gradual build up and decay over a longer total duration. If the explosion does not vent at the sea surface it may be possible to measure a modulation to the frequency spectrum corresponding to a bubble pulse frequency. Explosive submarine volcanism can generate very high level transients but the duration of this activity is a good discriminant.

The CTBT

In Article 1 of the Comprehensive Test Ban Treaty each State Party undertakes 'not to carry out any nuclear weapon test explosion or any other nuclear explosion, and to prohibit and prevent any such nuclear explosion at any place under its jurisdiction or control'. Article IV of the Treaty sets out a regime to verify compliance with this and the other basic obligations of the treaty.

Over 70% of the earth's surface is covered in water, so it is fortunate that a sparse network of stations using either hydrophones or T phase seismometers has the potential for monitoring clandestine nuclear explosions in the sea. As part of the International Monitoring System a network of 11 hydroacoustic stations was defined, 6 hydrophone and 5 T phase, whose locations are shown in the accompanying figure4. These stations were proposed at a Hydroacoustics Workshop in Paris in October 1995 and again at the Hydroacoustics Experts Meeting in Geneva in December of that year and adopted by the negotiators in the Conference on Disarmament and are reflected in the hydroacoustics network in the treaty text. Two of the hydrophone stations (Wake and Ascension) and one of the T phase stations (Queen Charlotte) were in existence before the Treaty, but require upgrading to IMS specifications in the future. Work is progressing on the remaining stations and it is presently planned to have all stations contributing to the International Data Centre in Vienna within four years. Wake Island, already contributing to the IDC with existing hydrophones, is not expected to be fully upgraded until 2005.

Current Capabilities

Figure 1 Map showing the planned IMS hydroacoustic network

Stations which are contributing data to the IDC at the time of writing are shown on the map in white. None of these stations have been certified as defined in the proposed Operational Manual for Hydroacoustic Monitoring and the data must be regarded as interim, although in practise the effect of certification may be small. With this distribution of stations there is the capability for detecting nuclear explosions in the oceans in which they are placed or which they border, and a minimal capability for localising such explosions using hydroacoustic data only. Where these stations do have value is their ability to help discriminate between small natural earthquakes with subsea foci and man-made explosions in the sea. There are substantial ocean areas, the largest in the South Pacific, where the existing hydroacoustic stations provide no coverage at all.

Future Capabilities

The situation will be considerably better when all 11 hydroacoustic stations are contributing data. At the December 1995 meeting mentioned earlier it was predicted that an explosion of a well coupled 200 ton underwater explosion (equivalent to a mb=4 earthquake) would be detected by a minimum of 3 stations in most oceans except the North Atlantic, the Arctic and within the Indonesian Archipelago. The resulting predicted distribution of location errors is shown in Figure 25, together with the predicted errors from the Primary, the combined Primary and Auxiliary IMS seismic networks, and then using all available data. The benefit of using all three networks is obvious. It has been noted at at least two scientific meetings that the location accuracy of the hydroacoustic network can be improved if T phase from selected auxiliary seismic stations is also used, particularly for locations in the South Pacific.

Figure 2 The synergy between seismic and hydroacoustic networks

The equivalent source level of a well coupled 200 ton underwater explosions is so large that it is accepted that provided there is no bathymetric blockage, such an explosion will be detectable by at least one hydroacoustic station. It is usually assumed that a shallow or even near surface atmospheric test over water will also be readily detectable. This is not immediately obvious, for hydrophones in the SOFAR channel are not particularly noisy with the din of the hundreds of ships that are at sea every day. This is because ship noise does not couple well into the SOFAR channel and requires a favourably sloping bottom before it is trapped6. The best empirical evidence for coupling into the duct by a source outside the duct is provided by the undoubted existence of abyssally generated T Phase from earthquakes under the deep ocean7. At the time of the atmospheric nuclear tests at Mururoa, data from hydrophones off New Zealand were examined for signs of an explosion, but none was found.

The Performance Limits of the Hydroacoustic Network

The minimum detectable explosion in the sea is a complex issue which will depend on the individual station and the propagation path between it and the explosion, and is not one which can be approached other than in generalities. A key factor is the background noise against which the detection must be made, and this will only be accurately known when the stations are on line. Four of the hydroacoustic stations are in regions where there has been recent volcanic activity. Of more concern is explosive submarine activity, particularly in the Pacific. In 1987-8 the MacDonald Seamount (28.99S, 140.26W) was active for weeks at a time. Another important factor is the propagation loss when not all the sound propagation is via the SOFAR channel. This depends critically on the nature of the sea bed and the sound velocity profiles along the propagation path, and may be best estimated after propagation loss experiments have been conducted to 'fine tune' the acoustic models.

Source location is presently done by triangulation using travel times to the various detecting stations. For hydrophone sensors care is needed selecting the appropriate arrival time: in the North Atlantic an explosion on the SOFAR axis builds to gradual climax and abruptly drops away, whereas in the South Pacific the onset is abrupt and there is a gradual fall off in level8. The travel time for an axis travelling ray is therefore best determined by the end of the sequence in the one case and the beginning in the other. Source localisation may be improved by timing the reflections from known bathymetric features. In this regard recently active volcanoes were found to be better reflectors of explosions from the Mururoa test site than were other islands, presumably because they have less sediment cover. Timing of explosions at T phase stations will be subject to difficulties because of the uncertain location of the point where the underwater sound is converted to the terminal island P phase. Calibration of the stations over a range of azimuths would seem necessary.

On a more positive note, computer processing power and hard drive capacity continue to grow at an amazing rate. Compared to voice recognition systems, the task of processing hydroacoustic data to discriminate between explosions from naturally occurring transients is relatively straightforward. The very high signal to noise ratio expected for any nuclear explosion raises the possibility of using the computer intensive techniques of matched field processing to determine source location.

References and Notes

1 Urick, R.J. Principles of underwater sound / 3rd edition. McGraw-Hill Book Co., 1983. Section 6.2. pp 159 - 164.

2 Bannister, R.W. et al. ATOC---New Zealand receiver site survey and acoustic test. ASA 125th Meeting Ottawa May 1993

3 Talandier,J. Submarine volcanic activity. Detection, monitoring and interpretation. Eos Vol. 70, No. 18, 1989. pp 561, 568 - 569

4 This figure copied from Lawrence, M. et al. The hydroacoustic network, International Monitoring System: status and plans. CTBT Seismic Research Symposium, New Orleans September 2000.

5 This figure copied from Hydroacoustic/seismic working paper 'Possible Benefits of Comprehensive Synergy between Hydroacoustics and Seismics' presented by the French Delegation to the Ad Hoc Committee on a Nuclear Test Ban. December 5 1995.

6 Dashen, R. and Munk, W. Three Models of global ocean noise. J. Acoust. Soc. Am. Vol. 76, (2) pp 540-554.

7 Johnson, R.H. et al. Abyssally generated T phases. Geophysical Monograph No.12. The Crust and Upper Mantle of the Pacific Area. Edited by Knopoff et al. American Geophysical Union. 1968.

8 Guthrie, K.M. Wave Theory of SOFAR signal shape. J Acoustic. Soc. Am., Vol. 56, pp 827-836