MYS0D9C766T24XP·Mystery · signal

The Bloop and NOAA's unexplained ocean sounds

Nojhan · CC BY-SA 3.0

Recorded by: NOAA PMEL hydrophones: US Navy SOSUS from 1991 and the eastern equatorial Pacific autonomous array from May 1996⁠[2]
Bloop: Summer 1997; broadband; heard at over 5,000 km; attributed to icequakes⁠[3]
Upsweep: Since at least August 1991; seasonal; near 54°S 140°W; origin unresolved⁠[11]
Train: 4 March 1997; 32–35 Hz; attributed to an iceberg grounded near Cape Adare⁠[7]
Slow Down: 1997 (file dated day 139); attributed to an iceberg grounding⁠[8]
Whistle: 7 July 1997, one hydrophone at 8°N 110°W; resembles an erupting arc volcano⁠[10]
Julia: 1 March 1999, 21:05 GMT; attributed to an iceberg running aground off Antarctica⁠[9]
Raw data: Not publicly released; only sped-up example clips and spectrograms are online⁠[15]
Status: Partly explained
Questions: 0 of 5 answered

People Christopher G. FoxPMEL acoustic monitoring scientist; first author of the array design paper Phil LobelBoston University marine biologist quoted on a biological source Haruyoshi MatsumotoPMEL acoustics engineer; co-author of the array design paper Robert P. DziakPMEL acoustics scientist; led the iceberg A53a study

Catalog keyMYS0D9C766T24XP

The deep ocean carries low-frequency sound for thousands of kilometres through the SOFAR channel, a layer where temperature and pressure trap sound waves. From 1991 NOAA's Pacific Marine Environmental Laboratory (PMEL) used the US Navy's SOSUS hydrophones to listen for undersea earthquakes and eruptions, and from May 1996 it ran its own array of autonomous hydrophone moorings in the eastern equatorial Pacific for the same purpose⁠[1]⁠[2]. Among the earthquakes, whales and ships, the analysts heard a handful of loud signals they could not place and gave them names.

The best known is the Bloop, a broadband sound recorded in the summer of 1997 and heard on sensors more than 5,000 km apart⁠[3]. In 2002 a New Scientist report, relayed by CNN, said it bore the hallmarks of an animal but was louder than any known creature, and the idea of an unknown giant animal took hold in popular culture⁠[4]. NOAA's own scientists had already suspected ice: in 2001 Christopher Fox of PMEL's acoustic monitoring project said it might be ice calving because "it always comes from the south"⁠[5]. PMEL now describes the Bloop as consistent with icequakes from large icebergs cracking, based on the similarity of its spectrogram to icequakes recorded in the Scotia Sea and on its arrival direction⁠[3]; NOAA's Ocean Service dates the identification to 2005⁠[6].

The other named sounds have been assigned in the same way. Train, Slow Down and Julia are attributed to icebergs grounding on the seafloor; Whistle to an erupting volcano too distant to locate; Upsweep, the oldest and most persistent, remains unresolved on PMEL's own page⁠[7]⁠[8]⁠[9]⁠[10]⁠[11]. The case is a good test of how far a spectrogram and a bearing can carry an attribution. The explanations are plausible and fit a body of peer-reviewed work showing that Antarctic ice is one of the loudest natural sources in the southern oceans⁠[12]⁠[13]⁠[14]. But no paper has published the original 1990s recordings, their arrival times or their azimuths, so the specific attributions cannot be checked by anyone outside NOAA. Of the 5 questions on this site, 0 are answered. The most likely answer to “What made the loud unidentified sounds recorded by NOAA's Pacific hydrophones in the 1990s?”, not yet confirmed: all of them are natural geophysical or cryogenic sounds (icequakes, grounding icebergs, volcanic eruptions….

Key questionsset by the editors

What made the loud unidentified sounds recorded by NOAA's Pacific hydrophones in the 1990s?Most likely: all of them are natural geophysical or cryogenic sounds (icequakes, grounding icebergs, volcanic eruptions…

What produces the Upsweep?No answer ahead yet

All 5 questions

What happened

Spectrogram of The Upsweep recording — The Upsweep recordingNOAA Pacific Marine Environmental Laboratory, 1991. Public domainArtifact

PMEL began recording the Navy's SOSUS arrays in August 1991 and was using them to locate undersea eruptions by 1993⁠[11]⁠[1]. In May 1996 it deployed its own array of autonomous hydrophones in the eastern equatorial Pacific, moored in the SOFAR channel to monitor the East Pacific Rise between 20°N and 20°S and the Galápagos Ridge; the array recorded continuously and was recovered and redeployed at intervals⁠[2].

Upsweep was already present when SOSUS recording began in 1991: long trains of narrowband sounds rising in pitch over several seconds each, loud enough to be heard across the Pacific, peaking in spring and autumn⁠[11]. In 1997 the equatorial array recorded Train on 4 March, a quasi-steady tone at 32 to 35 Hz⁠[7]; Slow Down, a sound that descends in frequency over about seven minutes, in a file numbered for day 139 of 1997 (19 May)⁠[8]; Whistle on 7 July, an undulating single tone on the hydrophone at 8°N 110°W⁠[10]; and the Bloop, during the summer⁠[3]. Julia followed on 1 March 1999 at 21:05 GMT⁠[9].

Map of the worldOpen the full map

+−

Eastern equatorial Pacific autonomous hydrophone arrayHydrophone array (moorings between about 8°N and 8°S near 95°W–110°W)

A region

NOAA autonomous hydrophone at 8°N 110°WHydrophone mooring

Within 10 km

Upsweep source region, about 54°S 140°WOpen ocean (approximate location)

A region

Bransfield Strait, Antarctic PeninsulaStrait

A region

Bloop source region, between Bransfield Strait and the Ross SeaAntarctic coast (approximate location)

A region

South Georgia, Scotia SeaIsland

A region

Cape Adare, Ross SeaCape

Within 10 km

A regionNatural Earth

Timeline

Timeline

Squeezed where little happened. Each mark is a published claim, by how it stands:AcceptedMost likelyStill possibleDisputedRuled out

The accounts

There are no witnesses beyond the analysts and their recordings. The public record is a set of short PMEL web pages, one per sound, each with a spectrogram and an audio clip sped up 10 to 20 times so that infrasound becomes audible⁠[3]⁠[11]⁠[7]⁠[9]⁠[10]⁠[8]. The pages are undated and have been revised. The Slow Down clip is still labelled "Equatorial Pacific unknown source" on the program's index page while the cryogenic page calls it an iceberg grounding⁠[15]⁠[8].

Press accounts filled the gap. In June 2002 CNN, summarising New Scientist, described the Bloop as possibly a giant animal, quoted the marine biologist Phil Lobel doubting a squid but favouring a biological source, and quoted Fox that most of the named sounds had been explained, with Upsweep tracked to an undersea volcano⁠[4]. The same report said the Bloop was heard on Navy SOSUS sensors 3,000 miles apart; NOAA's Ocean Service says hydrophones more than 3,219 km apart across the Pacific; neither names the instruments precisely⁠[4]⁠[6]. NOAA's Ocean Service lists the theories that circulated: military exercises, ship engines, fishing-boat winches, giant squid, whales and unknown animals⁠[6].

What was recorded

What survives in public is about a dozen sped-up audio files and spectrogram images, released as public information that may be freely redistributed with credit⁠[16]. The Bloop clip is 16 times real speed; Upsweep 20 times; Whistle 16 times and band-passed between 8 and 20 Hz⁠[3]⁠[11]⁠[10]. The raw multi-hydrophone recordings, the arrival-time picks and the azimuth estimates behind each attribution have not been published; no 1990s equatorial array data were found in NOAA's public passive acoustic archive⁠[17].

Spectrogram of The Bloop recording — The Bloop recordingNOAA Pacific Marine Environmental Laboratory, 1997. Public domainArtifact

Two things about the sounds are stated with numbers. The Bloop was detected at ranges over 5,000 km and its bearing pointed between Bransfield Strait and the Ross Sea, or to Cape Adare⁠[3]. Upsweep was located roughly at 54°S 140°W and has declined in level since 1991⁠[11]. Whistle, heard on one hydrophone, cannot be located at all, because a fix needs at least three⁠[10].

The explanations

Ice is the explanation NOAA gives for four of the six. The supporting work is strong on the general point. Hydrophones moored in the Scotia Sea recorded iceberg A53a, about 55 by 25 km, producing harmonic tremor as it scraped the seabed and short broadband icequakes as it broke up near South Georgia on 2 February 2008⁠[18]. The breakup of icebergs C19a and B15a from late 2007 raised noise levels at mid to equatorial latitudes for about a year and a half, and seasonal noise cycles across the southern oceans track iceberg drift⁠[12]. Ice tremor off Wilkes Land reaches estimated source levels near 245 dB⁠[13], and Ross Sea icebergs were recorded as far away as French Polynesia in the same years⁠[14]⁠[19]. PMEL's case for the Bloop is that its spectrogram resembles the Scotia Sea icequakes and its bearing points at Antarctic coasts known for ice sounds⁠[3]. For Train and Julia the case is a steady or descending tone from the direction of the Ross Sea or the Antarctic Peninsula, which matches a berg dragging or settling on the seabed⁠[7]⁠[9].

Spectrogram of The Train recording — The Train recordingNOAA Pacific Marine Environmental Laboratory, 1997. Public domainArtifact

Spectrogram of The Julia recording — The Julia recordingNOAA Pacific Marine Environmental Laboratory, 1999. Public domainArtifact

A biological source was the reading the press preferred for the Bloop. Its basis was the sound's rising frequency, which reminded biologists of animal calls, and its loudness, which no known animal approaches; it rests on no recording of any animal and no NOAA scientist has proposed it⁠[4]⁠[5]. A human source (exercises, engines, winches) was suggested but fits poorly with a source to the far south at ranges over 5,000 km⁠[6]⁠[3].

Upsweep has two readings. PMEL notes that its rough location lies near inferred volcanic seismicity but calls the origin unresolved; in 2002 Fox was reported as saying it had been tracked to an undersea volcano⁠[11]⁠[4]. Its seasonality could reflect the source or seasonal changes in how sound travels⁠[11]; seasonal ice noise is now documented across the southern oceans, which makes an ice component worth testing⁠[12].

Published claims

Published claims and their status

Year	Who	Claimed	Status	Reason
1995	Christopher Fox and colleagues (NOAA PMEL)⁠[1]	The Navy's SOSUS hydrophone arrays can detect and locate a seafloor volcanic episode, as they did on the Juan de Fuca Ridge in 1993.	Accepted	The method became routine and led to the autonomous arrays that recorded the 1990s sounds⁠[2].
2001	Christopher Fox (NOAA PMEL)⁠[5]	The Bloop may be related to ice calving, since it always comes from the south.	Still possible	The southern, Antarctic source held, but PMEL later attributed the sound to icequakes in cracking icebergs rather than calving from a shelf⁠[3].
2002	Phil Lobel, reported by New Scientist and CNN⁠[4]	The Bloop's profile resembles a sound made by a living thing, though probably not a squid.	Disputed	Icequakes recorded near Antarctica later matched its spectrogram, and NOAA now identifies the sound as an icequake⁠[3]⁠[6].
2002	Christopher Fox (NOAA PMEL), reported by CNN⁠[4]	The Upsweep has been tracked to an undersea volcano in the South Pacific.	Still possible	PMEL still calls its origin unresolved; the source lies near inferred volcanic seismicity, and an ice origin is also proposed⁠[11]⁠[12].
2002	Jacques Talandier, Olivier Hyvernaud, Emile Okal and Pierre-Franck Piserchia⁠[14]	Large icebergs in the Ross Sea make hydroacoustic signals that seismic stations in French Polynesia record thousands of kilometres away.	Accepted	Replicated for other bergs in the Indian and Pacific oceans and in the Scotia Sea⁠[19]⁠[18].
2005	Emily Chapp, DelWayne Bohnenstiehl and Maya Tolstoy⁠[13]	Drifting icebergs off Wilkes Land produce low-frequency tremor with source levels up to about 245 dB.	Accepted	Consistent with later iceberg recordings that show ice is loud enough to be heard across ocean basins⁠[12].
2005	NOAA PMEL acoustics researchers⁠[6]	The Bloop was an icequake, made by a large iceberg cracking and fracturing off Antarctica.	Accepted	The 2008 recordings of iceberg A53a breaking up gave icequakes with similar spectrograms, heard at over 5,000 km⁠[3]⁠[18].
2006	Jacques Talandier, Olivier Hyvernaud, Dominique Reymond and Emile Okal⁠[19]	Icebergs both grounded and drifting produce distinctive hydroacoustic signals recorded at long range in the southern oceans.	Accepted	Matches PMEL's attribution of Train and Julia to grounding icebergs, though those identifications rest on azimuth alone⁠[7]⁠[9].
2013	Robert Dziak and colleagues⁠[18]	Iceberg A53a gave off harmonic tremor while scraping shoals and short broadband icequakes as it broke up near South Georgia in 2008.	Accepted	The first iceberg followed acoustically from drift to breakup, and the reference for identifying the Bloop⁠[3].
2014	Haruyoshi Matsumoto and colleagues⁠[12]	Antarctic icebergs are a major source of ocean noise across the Southern Hemisphere, and seasonal noise cycles follow their drift and breakup.	Accepted	Shown with archived records of the breakup of bergs C19a and B15a; whether it also explains the Upsweep's seasonality has not been tested directly⁠[12].

What would settle it

Release of the original multi-hydrophone records, with arrival times, for each named sound. With those, the bearings can be recomputed and compared with iceberg positions on the same dates from the scatterometer iceberg database, which covers 1997 through ERS and NSCAT and 1999 through ERS and QuikSCAT⁠[20]⁠[21]. A classifier trained on modern labelled recordings of icequakes, groundings, tremor and eruptions could then score each 1990s sound against each class⁠[17]⁠[18]. For Upsweep, a multi-year record from hydrophones south of 40°S would show whether the source is fixed (a vent) or moves with the ice.