Center for Arms Control, Energy and Environmental Studies at MIPT

New !!!: Rescue Of "Kursk"

Submarine Collision off Murmansk: A Look from Afar

by Eugene Miasnikov (as submitted for publication)
This paper was published in DACS Breakthroughs magazine (Defense and Arms Control Studies Program at M.I.T., Winter 92/93, v. 2, # 2, pp 19-24) and reprinted in The Submarine Review (April, 1993, pp. 6-14). The paper analyses the plausible reasons, which lead to a Russian (of "Sierra" class) and a U.S. ("Baton Rouge") submarines collision incident in February, 1992. The author comes to the conclusion, that most likely, the reason was very short detection range of submarines, because of that both submarines operated covertly and did not use other means of detection except passive acoustics. There are no reliable means available to submarines that would allow them to operate both covertly and safely in such a complex environment as shallow waters. The collision illustrates that covert operations of foreign submarines close to Russian naval bases can create dangerous situations that may result in undesirable outcomes.

On February 11, 1992, two Russian and American submarines collided in or near Russian territorial waters off the port of Murmansk. While such submarine collisions have occurred before, the most recent collision seems to have generated more response in the press than previous ones. Differing and contradictory press accounts of the collision, published in both the US and Russia, raise two questions: What was the US submarine doing so close to the Murmansk naval facilities? And how could such a collision have occurred?

These two seemingly simple questions, however, raise a third question that points to issues of prime importance for security planning in both Russia and the US: Does this collision tell us anything important about the capabilities of US submarines to conduct antisubmarine operations against the newest generation Russian submarines? If so, could this incident indicate that current and future generations of Russian ballistic missile submarines could be held at risk by US Navy undersea forces? In this paper, we examine these questions using both Russian and American sources, and a technical review of the capabilities and limitations of submarine sensors in shallow coastal waters.

In particular, our analysis indicates that in the shallow northern seas, even under the best environmental conditions, the technical capabilities of modern covertly operating submarines do not allow the detection of other modern covert submarines at distances of more than a couple of hundred meters.

What is known about the incident?

The USS Baton Rouge, a Los Angeles class nuclear attack submarine1 and the Russian Sierra class submarine collided at 20:16 Moscow time, at 69o38.7' North and 33o46.9' East, roughly 4.7 miles from the line connecting Tsypnavolok Cape and Kildin Island (see map below).2 The US Navy stated that the collision occurred more than 12 miles from the shore, at a location in international waters. However, Russia uses a different set of rules for defining the boundary between territorial and international waters,3 and the Russian rules put the collision site inside their territorial waters.

The dispute over exactly what areas of coastal ocean can be considered international waters is important since international coastal waters can be used for a wide range of activities and many questions of rights and "rules of the road" are effected by whether or not naval operations occur within territorial or international waters. In the particular case of Murmansk there are obvious additional concerns about the security and operation of Russian warships near their home port. The Murmansk area contains the largest base of the Northern Fleet.

According to US officials, the collision occurred when the Sierra was surfacing beneath the Baton Rouge, which was at a periscope depth of 22 yards.4 US reconnaissance photos of the Sierra reportedly showed a large dent in the front section of her sail structure and indicate that the Russian submarine's sail may have hit the underneath aft section of the Baton Rouge.5 According to Soviet reports,6 the Sierra incurred slight damage to her sail, where substantial bits of the U.S. submarine's skin - ceramics, plastics, and other components - were found. Reportedly, after the Baton Rouge had returned to her base in Norfolk two weeks later, divers conducting an underwater inspection found scrapes, dents and two minor cuts to her port ballast (one of two on the submarine).7 Fortunately, the accident did not cause any injuries or deaths.

Why was the Baton Rouge there?

The U.S. Navy has not released an explanation of why the Baton Rouge was operating so close to the Russian coast. However, sources within the Pentagon have reportedly said that the Baton Rouge was on an intelligence gathering mission at the time of the collision.8 This explanation has also been brought up and discussed in other press accounts of the collision as well.

There are several kinds of intelligence missions that could, at least in principle, have brought the Los Angeles class submarine so close to the Russian coast. One type of mission is simply aimed at gaining experience operating in shallow waters as close as possible to the Russian coast. Although one press account stated that "there is little if any tactical reason these days for American submarines to operate so close to Russian shores,"9 the US Navy may not yet have reached the same conclusion.10 The gathering of intelligence with submarines may also be able to provide information on aspects of Russian Naval operations that could be useful in helping to predict the actions and movements of the Northern fleet.11 According to knowledgeable sources in the Russian Navy: "Intelligence gathering is a routine activity of American subs near our coast. Typically, there are one to two American or British submarines operating close to the Soviet coast off Murmansk, 1-3 off the Kamchatka peninsula and 1 off the coast of Vladivostok. During naval exercises this number can increase by a factor of two."12

Gathering intelligence either inside or just outside Soviet territorial waters has been a long-term program of the American Navy, and has been given names such as Holystone, Pinnacle, Bollard and Barnacle.13 These activities apparently included close-up photography of the undersides of Soviet ships and submarines; plugging into Soviet underwater communication cables to intercept high-level military and other communications considered too important to be sent by radio or other less secure means; observation of Soviet SLBM tests including monitoring of the various computer checks and other signals that preceded test launchings; and the recording of "voice autographs" -- the noises made by operating Soviet submarines. One possibility, consistent with reports that her mission was intelligence gathering, is that the Baton Rouge was on the mission to install (or recover) intelligence gathering devices from the seabed near the shore.

Press speculation that the "American submarine, which...was sitting at 'periscope depth', may have been using secret interception equipment to monitor communications at nearby military bases"14 is implausible for two reasons: First, continuous radio interception of military communications can be accomplished without the use of a submarine. Communications intercepts can be accomplished with surface ships operating from international waters, and intermittent interceptions can be accomplished with satellites orbiting in space. Second, a submarine periscope or antenna sticking out of the water can be observed by a variety of relatively long-range sensor systems - in particular by modern radars.15 As a result, unless there is a compelling reason to argue otherwise, it is unlikely that a submarine commander would be willing to keep a periscope mast deployed for a substantial period of time in such close proximity to a potential enemy's surveillance systems and forces.

Could the Baton Rouge have been trailing the Sierra?

According to one Soviet press report,16 the collision was the result of a cat-and-mouse game between the US and Russian submarines. This possibility deserves detailed study, since it could indicate that the US might still retain substantial capabilities to trail Russian ballistic missile submarines.17

Since the Baton Rouge was operating covertly close to major Russian Naval facilities, it is highly unlikely that she would have used her active sonar during normal operations. Such a practice could greatly increase the likelihood that her presence would be detected. It is also unlikely that she would have been using her long towed array. The length of such an array with towing cable is more than 1 km, many times the water depth at the place of collision.18 In addition, it is difficult to control an array's orientation in the water without severely constraining the motion of a submarine. Such constraints are highly undesirable when operating in close proximity to potentially hostile forces. Thus, even though using her towed array would substantially improve her detection capabilities, it is unlikely that the Baton Rouge would have this type of sensor system deployed in shallow waters. Hence, she was almost certainly using only her fixed passive sonar systems while she was operating in the shallow waters off Murmansk. This conclusion allows us to estimate her detection and counterdetection capabilities against the Sierra submarine she was allegedly tracking.

There are three reasons why the submarine passive acoustic detection ranges would have been short in the place of collision.19 First, in shallow waters, the ambient noise levels from wind-generated breaking waves are typically 10-100 times higher than those in deep water. This noise generates background signals that can mask the presence of a signal from even a nearby submarine.20 Second, in shallow water, the acoustic signals from a target submarine will arrive at the acoustic detectors of a hunting submarine from many different directions and at many different times. This is because sound waves generated by a submarine will be reflected from the constantly shifting ocean surface and from numerous locations on the ocean bottom many times before they arrive at the face of an acoustic detector. Since the signals from a target and from breaking waves both unpredictably come from many directions, there is no way to enhance the signal from a target submarine relative to that from interfering wind-noise by increasing the number of receivers and the size of array.21 Third, there are no sound focussing effects in shallow waters, as there are in deep water, that can make it easier to detect the submarine against the background of ocean noise. Such deep water effects can strongly focus the sound from a localized sound source like a submarine at well defined distant ranges in the ocean. At the same time these effects only weakly focus the diffusely generated sounds from wind-generated noise.22

Quantitative analysis shows that the shallow water detection range of the fixed sonar of the Baton Rouge against a quiet23 Sierra class submarine would only be a couple hundred meters - even if the acoustic conditions for detection were nearly ideal and the submarine was oriented so that its sensors could achieve maximum sensitivity. Near ideal acoustic conditions could occur only in extremely calm seas. For environmental conditions that are much more typical of the waters off Murmansk, like those associated with a 10 knot surface wind,24 noise levels would be high enough to result in the same short detection range of even if the Baton Rouge were using a long towed array. Making matters worse, it is likely that the Sierra may have encountered the American submarine from behind. In this circumstance, the Baton Rouge would have had no ability to detect the approaching Sierra, as the fixed sonar on the submarine cannot detect the signals within a cone 60 degrees to the rear of the submarine.

Even if some unexpected combination of events led to the detection of the Sierra, the fixed sonars on the Baton Rouge could provide very little useful information about the direction and range of the Sierra. These detection and tracking limitations are a consequence of the relatively small size of fixed hull arrays and the highly unpredictable and variable transmission efficiency of underwater sound.

The arguments above do not support the speculation that Baton Rouge was trailing the Russia submarine. It would have been very hard for the Baton Rouge to do so in this particular situation.

Could the Sierra have known that the Baton Rouge was there ?

Clearly, the conditions for the Sierra to detect the Baton Rouge by passive acoustic means were no better. As we have learned from informed Russian sources, there are specific safety rules for the safety of submarine operations in such a complex environment. The area of a submarine operations is closed to any shipping. The submarine is supposed to "look" around with using her sonar every hour and every time when her depth of operation is changed. In order to do this a submarine has to move along a loop shaped trajectory, because her sonar is deaf in the aft direction.

It is possible also that the Russian submarines may use their sonars in an active mode as a standard operational procedure, when they do not need to be covert in their home waters. The use of an active sonar could allow the Sierra to increase her detection range to a couple of kilometers and to obtain much more detailed target location information. It appears, however, that she was not using her active sonar, since the American submarine would have heard the approach of the Sierra and determined the bearing of the Russian submarine25 at a distance of at least several kilometers, which is enough to take care to avoid a collision. In fact, in this situation, the Baton Rouge would have detected the Russian submarine long before she was detected by the Sierra.26

It is also interesting to consider two related questions: What were the potential capabilities of the Russian ASW forces to detect the Baton Rouge? Would it have been possible to avoid the accident if the Northern Fleet ASW forces acted properly?

Most likely, the American sub had to pass a Russian seabed passive (or active) sonar system near the approaches to the Russian shore. The detection range against a quiet Los Angeles class submarine of such a passive system might be about 1-5 kilometers in the best conditions of a calm sea. According to the Russian Chief Navigator Valery Alexin, several fishery ships were present, the screws of which generated noise similar to that of the American submarine.27 These fishery ships might substantially mask the Baton Rouge, although it seems improbable that such ships were actually present, since the Russian Navy is typically very cautious about permitting civil ships in the operating areas. Using a passive seabed system the location of the submarine could be determined more accurately than by using a towed array, but probably not better than 1 or 2 km. If stationary active sonar techniques were employed,28 the detection range would be restricted by surface and bottom reverberations and might not exceed 5 km.

The Baton Rouge could also have been tracked either by a ship's active sonar at a range of no more than 1 or 2 km or by sonobuoys which could be deployed by aircraft. The detection range of active sonoboys most probably was not more than 1 km for this particular case.29 A possible means of locating a submarine more precisely could be airborne magnetic anomaly detection (MAD) and lidar sensors.30

As a result, it appears unlikely that the Russian ASW Forces detected the American submarine before the collision. As the estimates above indicate, substantial capabilities would be required to keep even a relatively small area safe from invasions by foreign submarines. Most probably, on routine duties, less capable forces are deployed, and this might be the reason why the American submarine was not detected.


In spite of the claims in the news reports, it appears likely that neither submarine heard the other before the collision and that the collision was a basically an accident. Playing an extended cat-and-mouse game would have been impossible there because of very short detection ranges that would be possible against quiet submarines. Moreover, there are no reliable means available to submarines that would allow them to operate both covertly and safely in such a complex environment as the shallow waters of the Barents sea.

The circumstances of this collision suggests that, at least in some environmental conditions, if carefully operated, modern Russian submarines are almost impossible to detect by passive acoustic methods, even by the highly capable ASW forces of the United States.31 If true, this has important implications for the options available to Russian policy makers as they decide how to implement the nuclear reductions called for by the START agreement as well as possible future deep reductions in strategic nuclear forces.

The great emphasis that has recently been placed on assuring the safety and security of the nuclear weapons of the former Soviet Union stands in sharp contrast to the circumstances of this incident. The recent collision illustrates that covert operations of foreign submarines so close to Russian Naval bases can create dangerous situations that may result in unpredictable outcomes. More than half of the 54 Russian strategic submarines, each with 16-20 nuclear missiles, are still based near Murmansk. This suggests that any benefits accrued by US attack submarines operating so close to Russian port facilities may be offset by the risk that a nuclear weapons related accident might eventually result.


I gratefully thank Prof. Theodore Postol and Dr. George Lewis at the DACS Program of MIT for very helpful advice and discussions during the whole time of my work on the subject considered in this article. I would also like to thank Prof. Ira Dyer at the Ocean Engineering Department of MIT for the help in estimating of the detection capabilities against submarines by acoustical means and the high ranking officer in the Russian Navy, former captain of a nuclear powered submarine, whose name I can not mention, for valuable comments on tactics of submarine operations. I am also thankful to Dr. David Wright at Federation of American Scientists, Matthew Partan at the DACS Program of MIT and Paul Podvig at the Center for Arms Control, Energy and Environmental Studies in Moscow Institute of Physics and Technology for their comments and suggestions when this article was under preparation.

1) According to press reports, the Russian submarine had the number 689 on her sail. Numbers on the hulls of Russian submarines and warships are typically drawn in order to misinform a potential enemy and have nothing to do with the real designations, which are classified. In the following text we will call the Russian submarine 'Sierra' for the sake of brevity.

2) "Novosti" TV report (Moscow, February 24, 1992).

3)The US and Russia recognize a 12-nautical mile territorial limit, but the two countries have different methods of applying this limit. According to the Russian method, points are marked 12 nautical miles beyond the line between two pieces of land that extends farthest into the sea on either side of a bay or a gulf: in this case, Tsypnavolok Cape, on the Rybachii Peninsula; and the northern shore of Kildin Island. The U.S. method draws a line 12 nautical miles offshore that follows the general contour of the coastline. The U.S. and former Soviet governments have held bilateral discussions, most recently in 1990, aimed at resolving this and other nautical boundary disputes. These talks have so far not met with success.

4) Bill Gertz, "Russian Sub's Sail Damaged in Collision," Washington Times, Feb. 27, 1992, p.4.

5) Bill Gertz, ibid.

6) Artur Blinov and Nikolay Burbyga, "Underwater Incident in the Kola Gulf," Izvestia, Feb. 20, 1992 p.1; Nikolay Burbyga and Viktor Litovkin "Americans Not Only Helping Us, But Spying on Us. Details of Submarine Collision in Barents Sea," Izvestia, Feb. 21, 1992, p.2.

7) Bill Gertz, op.cit.

8) Melissa Healy, "Russian, US Subs Collide off Murmansk," Boston Globe, Feb. 19, 1992.

9) John H. Gushman Jr. "Two Subs Collide off Russian Port," The New York Times, Feb. 19, 1992.

10) If an unexpected political reversal results in future military conflict between the US and Russia, the US Navy might well prefer to fight, at least some of the time close to Russian Naval bases. Such concerns may well lead US Navy planners to conclude that submarine missions near the Soviet coast should be a high priority military activity.

11) Barry R. Posen, Inadvertent Escalation: Conventional War and Nuclear Risks, Cornell University Press, 1991, p. 142.

12) According to the same source, foreign submarines typically operate in the area north of the entrances to Kola and Motovsky Bays.

13) Desmond Ball "Nuclear War at Sea," International Security, Winter 1985-86, v.10, No 3, pp.3-31.

14) John H. Gushman, Jr., op.cit.

15) The Russian Navy presumably keeps substantial ASW forces on alert at Murmansk, since it is known that foreign submarines routinely attempt to operate in the waters near their naval bases.In the ocean waters near Murmansk an American submarine might have to contend with airborne, shipborne and shore-based radars capable of detecting the presence of a periscope deployed above the ocean surface. For example, a deployed submarine periscope can typically be detected at dozens of kilometers by modern airborne radars on maritime patrol planes. In addition, if a submarine is detected in highly confined coastal waters, it may have few tactical options available to evade further detection.  As it seeks to escape, there is always the potential that it could be trapped, or later be reacquired, by ships and helicopters that are executing a coordinated search of the confined area with active sonar systems. The American submarine would therefore have to operate with extreme caution, as nearby Russian ASW forces could be quickly launched in pursuit of the submarine if it is detected.

16) Nikolay Burbyga and Viktor Litovkin, op. cit.

17) This question is very important for survivability of the CIS strategic forces at sea. If quieter American attack submarines could covertly trail them, this could cause an unstable situation in a conflict between these two countries.

18) Several different sources all agree that the depth is 100-200 m there.

19) For detailed quantitative estimations of the detection range see: Future of Russia's Strategic Nuclear Forces: Discussions and Arguments, (by Eugene Miasnikov, Center for Arms Control, Energy and  Environmental Studies at MIPT, Dolgoprudny, 1995) and Can Russian Submarines Survive at Sea? The Fundamental Limits of Passive Acoustics, (by Eugene Miasnikov, Science and Global Security, 1994, v. 4, pp. 213-251).

20) Wind generated noise is created by the action of breaking waves at the ocean's surface. Since sound waves in the ocean are also strongly reflected by the ocean bottom, sound waves generated in shallow water are confined to very small volumes of water relative to sound waves generated in deep water. This mechanism results in markedly higher wind-generated sound power densities relative to those in deep water.

21) In technical terms, the coherence of sound in shallow water is considerably lower than that in deep water. This means that it is not possible to achieve shallow water array gains as high as those in deep water.

22) This deep ocean phenomenon is known as "convergence zone" formation. In relatively narrow (less than 4-6 km) convergence zones, submarine generated sound level may be 100 times larger than in between them. As a rule, in these convergence zones the correlation length of the signal is also larger.

23) Sierra class submarines were first commissioned in 1984 (Norman Polmar, The Naval Institute Guide to the Soviet Navy, 5th ed., 1991) when the quieting technology of the Soviets was substantially improved. The Russian submarine probably moved at a low speed of 4-7 knots, since motion with a higher speed would increase the flow noise in the hydrophones and decrease her detection capabilities, which might be dangerous in a shipping area. We assume in our calculations that the source noise level of the "Sierra" is 120 (dB) decibel with respect to 1 microPa at a distance of 1 m. For World War II submarines, the noisiness at low speeds is 140-160 dB ( R. J. Urick, Principles of Underwater Sound, McGraw-Hill Book Company, 1983), which is 100-10000 times higher. We assumed also that the source noise level of the "Los Angeles" class submarine is 110 dB, since the U.S. has more advanced submarine quieting technology.

24) The wind speed exceeds 15 m/s 25 percent of the time in this area. See Tom Stefanick, Strategic Antisubmarine Warfare and Naval Strategy, Lexington Books, 1987, p.332.

25) In an active mode, a fixed sonar of a submarine operates at a frequency of 3-10 kHz which allows it to get much more precise information on the bearing of a target. The lower the frequency, the less accurate the target bearing estimation. However, the source noise level of a target submarine at such high frequencies is small and therefore difficult to detect passively. For this reason, when passive acoustic data obtained with a fixed sonar, a narrowband processing technique is typically applied at lower frequencies.

26) It is also interesting to note, that even if the Sierra approached Baton Rouge from behind and for some reason the American submarine did not hear Sierra's active sonar, the target strength of the American sub would be 10-100 times less than if it were seen from the side. The Sierra would not have detected the target in this case even at relatively close distance.

27) Valery Alexin, "The Hunt for "Red February"? Once More about the Collision of Two Nuclear Powered Vessels," Krasnaya Zvezda, February 27, 1992.

28) It should be noted that an active system could be much more expensive to deploy and maintain than a passive one. Given the area to be surveyed and the detection range of the individual array of receivers it can be easily found that a large number of receivers would be necessary to detect submarines within 12 mile zone at Murmansk area either by passive or active means.

29) The Navy of the CIS said also that a foreign submarine was chased out of Russia northern waters on March 26, 1992 just six weeks after the collision approximately at the same location. According to Krasnaya Zvezda, a 100 km long antisubmarine barrier of sonoboys was deployed from the air (A. Pilipchuk "Antisubmariners Were Ready to Use Weapons," Krasnaya Zvezda, March 28, 1992, p.2). Given the detection range of sonoboys, this operation must have been very expensive.

30) These means of detection could only be used to locate more precisely an already detected submarine because of they have a very small detection range (200-500 m) and, therefore, the aircraft that carry them have a low search rate.

31) The vulnerability of Russian ballistic missile submarines to detection by passive acoustic means (at present, by far the most important means for detection) in different types of ocean areas will be discussed in our separate paper which is in preparation now.

Center for Arms Control, Energy and Environmental Studies at MIPT, 1993-1998.