of 19
1. Introduction
The global ocean absorbs most of the excess thermal energy trapped on the planet by anthropogenic greenhouse
gases (IPCC,
2021
). Its heat uptake plays a key role in setting the rate of global warming, sea level rise, and many
other aspects of climate projections for the next century. Accurate observations and a detailed understanding of
the oceanic heat uptake are thus vital for such projections. The estimates of global and basin-scale ocean warm-
ing trends at various depths, however, remain uncertain due to a paucity of observational data. In particular, it is
difficult to separate the long-term warming trend from other large-amplitude local transient signals (e.g., seasonal
variations, mesoscale eddies, internal waves, etc.).
Observational data for estimates of deep-ocean warming prior to the mid-2000s primarily come from ship-based
repeat hydrographic surveys. Reaching back to the 1990s (in some cases 1980s), these surveys now allow esti-
mates of multi-decadal changes, but the necessarily sparse sampling, with an average spacing of more than
1,000 km between sections, leads to large uncertainty of warming trend estimates (Desbruyères et al.,
2016
;
Gille,
2002
; Purkey & Johnson,
2010
; Talley et al.,
2016
). Since the mid-2000s, the global fleet of autonomous
Argo floats has provided us with unprecedented spatial (on average ∼300-km spacing between floats) and tempo-
ral (typically a 10-days sampling interval) resolutions (Roemmich et al.,
2015
). The current Argo array, however,
still aliases mesoscale eddies, which have scales of O (100 km). In addition, most Argo profiling floats can
reach only a depth of 2,000 m, leaving about half of the ocean's volume unsampled. Deep Argo floats have been
designed to measure the ocean to 6,000 m, but a global array is not yet available (Chelton et al.,
2011
; Johnson
et al.,
2020
; Roemmich et al.,
2019
).
Abstract
Due to limited observational coverage, monitoring the warming of the global ocean, especially
the deep ocean, remains a challenging sampling problem. Seismic ocean thermometry (SOT) complements
existing point measurements by inferring large-scale averaged ocean temperature changes using the sound
waves generated by submarine earthquakes, called
T
waves. We demonstrate here that Comprehensive
Nuclear-Test-Ban Treaty Organization (CTBTO) hydrophones can record
T
waves with a higher signal-to-noise
ratio compared to a previously used land-based
T
-wave station. This allows us to use small earthquakes
(magnitude <4.0), which occur much more frequently than large events, dramatically improving the resulting
temporal resolution of SOT. We also find that the travel time changes of
T
waves at the land-based
T
-wave
station and the CTBTO hydrophone show small but systematic differences, although the two stations are only
about 20 km apart. We attribute this feature to their different acoustic mode components sampling different
parts of the ocean. Applying SOT to two CTBTO hydrophones in the East Indian Ocean reveals signals from
decadal warming, seasonal variations, and mesoscale eddies, some of which are missing or underestimated
in previously available temperature reconstructions. This application demonstrates the great advantage of
hydrophone stations for global SOT, especially in regions with a low seismicity level.
Plain Language Summary
As the largest heat reservoir of the Earth, the ocean takes up most of
the excess heat introduced into the atmosphere due to anthropogenic greenhouse gas emissions. Quantifying
ocean warming is critical for our understanding of climate change. Despite progress, sensing the vast deep
ocean is still challenging. In the mid-last century, it has been known that sound waves can travel a long distance
in the ocean and that its speed is faster in a warmer ocean. Decades ago, this fact was used to monitor the
ocean
temperature changes by tracking the human-made sound wave arrivals. Natural submarine earthquakes
can also produce loud sound waves. These sounds can be well recorded by underwater microphones, which are
built for listening to waves produced from nuclear tests. This study uses natural sounds from small earthquakes
to estimate the ocean temperature changes in the East Indian Ocean.
WU ET AL.
© 2023. American Geophysical Union.
All Rights Reserved.
Seismic Ocean Thermometry Using CTBTO Hydrophones
Wenbo Wu
1,2
, Zhichao Shen
1,2
, Shirui Peng
2
, Zhongwen Zhan
2
, and Jörn Callies
2
1
Woods Hole Oceanographic Institution, Woods Hole, MA, USA,
2
California Institute of Technology, Pasadena, CA, USA
Key Points:
CTBTO hydrophones improve
T
-wave
observations from small submarine
earthquakes at a teleseismic distance
T
-wave travel time changes from small
repeating earthquakes significantly
improve the temporal resolution of
seismic ocean thermometry
Seismic ocean thermometry applied
to the East Indian Ocean shows clear
decadal-to-seasonal temperature
changes and mesoscale eddy signals
Supporting Information:
Supporting Information may be found in
the online version of this article.
Correspondence to:
W. Wu,
wenbo.wu@whoi.edu
Citation:
Wu, W., Shen, Z., Peng, S., Zhan, Z.,
& Callies, J. (2023). Seismic ocean
thermometry using CTBTO hydrophones.
Journal of Geophysical Research: Solid
Earth
,
128
, e2023JB026687.
https://doi.
org/10.1029/2023JB026687
Received 8 MAR 2023
Accepted 6 SEP 2023
Author Contributions:
Conceptualization:
Wenbo Wu
Formal analysis:
Wenbo Wu, Zhichao
Shen, Shirui Peng, Zhongwen Zhan, Jörn
Callies
Funding acquisition:
Zhongwen Zhan,
Jörn Callies
Investigation:
Wenbo Wu, Zhichao
Shen, Shirui Peng, Zhongwen Zhan, Jörn
Callies
Methodology:
Wenbo Wu, Zhichao Shen,
Shirui Peng, Jörn Callies
Project Administration:
Wenbo Wu,
Zhongwen Zhan, Jörn Callies
Resources:
Wenbo Wu, Zhongwen Zhan,
Jörn Callies
10.1029/2023JB026687
Special Section:
Solid Earth Geophysics as a
means to address issues of
global change
RESEARCH ARTICLE
1 of 19
Journal of Geophysical Research: Solid Earth
WU ET AL.
10.1029/2023JB026687
2 of 19
Ocean warming can also be inferred indirectly by using the so-called sea-level budget (e.g., Dieng et al.,
2015
;
Llovel et al.,
2014
; Munk,
2002
; Volkov et al.,
2017
; Willis et al.,
2008
). By combining satellite altimetry obser
-
vations of the global sea level with satellite gravity measurements of ocean mass changes, the sea level rise due
to thermal expansion can be isolated. This method has much better spatiotemporal coverage than those based on
in situ point measurements, but the uncertainty remains high due to the complex sources of sea-level and ocean
mass changes (Milne et al.,
2009
) and the inferred net ocean temperature change has no depth information. It
is clear that our best estimate of ocean warming should be based on a combination of all available data (Levin
et al.,
2019
), and it appears that constraints that are orthogonal to what is presently available would be particularly
valuable.
To complement the existing ocean temperature measurements, Wu et al. (
2020
) proposed to use seismo-acoustic
waves generated by earthquakes, so-called
T
waves, to detect the temperature changes (Wu et al.,
2020
). This
seismic ocean thermometry (SOT) builds on the idea of active acoustic probing of ocean temperature changes:
“ocean acoustic tomography” first proposed by Munk and Wunsch (
1979
). The principle of basin-scale acoustic
tomography is to transmit sound over long distances (thousands of kilometers), which is enabled by the mid-depth
waveguide (the SOFAR channel, sound fixing and ranging), and to track changes in the travel time between a
source and a receiver. Sound waves travel faster in warmer water, so a reduction in the travel time can be attributed
to an increase in the average temperature encountered by the waves on their way from the source to the receiver.
These measurements thus intrinsically average out small-scale temperature fluctuations due to, for example,
mesoscale eddies or internal waves. Since the 1970s, several long-range acoustic transmission experiments have
demonstrated the high precision of acoustic thermometry, which is about 10 mK for a 1,000 km range (Worcester
et al.,
1994
,
1999
). In seismic ocean thermometry, we also measure travel time changes of acoustic waves traveling
a long distance through the ocean, but the synthetic sound sources are replaced by natural repeating earthquakes.
Wu et al. (
2020
) observed
T
-wave travel time changes from a
T
-wave seismic station installed on the Central
Indian Ocean atoll Diego Garcia and estimated the associated ocean temperature signals. Here, we extend this
idea by using two Comprehensive Nuclear-Test-Ban Treaty Organization (CTBTO) hydrophones in the Indian
Ocean, which have improved
T
-wave detection capability. We demonstrate that the hydrophone data result in
many more constraints and therefore improve temporal resolution in the inferred temperature time series.
2. Improved
T
-Wave Observations Using CTBTO Hydrophones
SOT relies on high-quality
T
-wave recordings from repeating earthquakes, so its performance highly depends on
the seismicity level of the region of interest. The Sumatra subduction zone is one of the most seismically active
regions in the world, which makes it a prolific
T
-wave maker. Wu et al. (
2020
) used the DGAR seismic station
on the Central Indian Ocean atoll Diego Garcia (Figure
1
), which is well known for its high-quality
T
waves
from earthquakes off Sumatra (Okal,
2008
), to measure the
T
-wave travel time changes and infer the temperature
changes of the equatorial East Indian Ocean in 2005–2016. The near-global existence of the SOFAR channel
would allow us to expand SOT across the World Ocean, but a global application would have to contend with lower
seismicity almost everywhere else and a lack of useable
T
-wave seismic stations. For example, a major part of
the Atlantic Ocean can only be sampled by earthquakes at the mid-ocean ridge, where earthquakes are much less
frequent than at most subduction zones (e.g., Sumatra). Although large earthquakes (
M
≥ 5.7) can excite clear
T
waves recorded by the Global Seismic Network (Buehler & Shearer,
2015
), their occurrence frequency is not
sufficiently high for a high-quality SOT—on average, only a few hundred of these events occur each year across
the world. In addition, the repeating intervals of these earthquakes are typically much longer than a year. Thus,
high-quality global SOT calls for more
T
-wave data from small earthquakes.
2.1.
T
-Wave Observations at CTBTO Hydrophones
In contrast to the noisy seismic stations on the land, hydrophones are deployed under water and directly record
ocean acoustic signals, such that
T
waves can be detected with a higher signal-to-noise ratio (SNR). Since the
2000s, CTBTO has operated a global hydrophone network, which was designed to detect and locate nuclear
explosions. Their sites are designed for optimal coverage of global ocean for nuclear test monitoring (Howe
et al.,
2019
). In addition, the CTBTO hydrophones are installed around the SOFAR channel axis to maximize
their sensitivity to remote acoustic signals. Thus, they also record clear
T
waves from distant small earthquakes
Software:
Wenbo Wu, Zhichao Shen,
Shirui Peng, Jörn Callies
Supervision:
Zhongwen Zhan, Jörn
Callies
Validation:
Wenbo Wu, Zhichao Shen,
Shirui Peng, Jörn Callies
Visualization:
Wenbo Wu
Writing – original draft:
Wenbo Wu
Writing – review & editing:
Wenbo Wu,
Zhichao Shen, Shirui Peng, Zhongwen
Zhan, Jörn Callies
21699356, 2023, 9, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2023JB026687 by California Inst of Technology, Wiley Online Library on [03/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Solid Earth
WU ET AL.
10.1029/2023JB026687
3 of 19
and are ideal for basin-scale SOT. For example, the CTBTO hydrophone station H08 around Diego Garcia shows
a clear
T
wave (SNR > 10) from a 2,900 km distant M3.8 earthquake in the Nias region (Figure
1
). In contrast,
the land-based station DGAR, which is only about 20 km away from the hydrophone, records the
T
-wave with
a much lower SNR. Another CTBTO hydrophone station H01 off Cape Leeuwin, western Australia, is further
away (4,400 km) and suffers from much more blocking by shallow bathymetry, especially in the southern part of
T
-wave path. Its
T
-wave record is noisier than that at H08 but still identifiable.
2.2.
Finding Missing Earthquakes With Template Matching
The high-SNR CTBTO hydrophone data enable us to use small earthquakes for SOT. Many of those events,
however, are missing from current catalogs produced by the United States Geological Survey or the International
Seismological Center (ISC) because they are remote from the seismic network. Wu et al. (
2020
) used about 4,000
events in the Nias region (western coast of Sumatra) that were cataloged by the ISC and found 2,047 repeating
pairs for SOT, which arise from a total of 901 events. Most of these repeaters have a magnitude larger than 4.0.
In order to use events missing from the ISC catalog (i.e.,
M
< 4.0) for SOT, the first step is to create a more
complete catalog. We apply the Template Matching (TM) method, which is a widely used seismological tech-
nique, to search for the missing earthquakes. The idea of TM is detecting seismic events by the waveform similar
-
ity between two events (see e.g., Anstey,
1964
; Harris,
1991
). In most applications, TM uses known waveforms
from identified events, so-called templates, to search for undetected events (e.g., Gibbons & Ringdal,
2006
; Li
& Zhan,
2018
; Peng & Zhao,
2009
; Zhang & Wen,
2015
). This method can help identify an earthquake based
on even a single station and thus is very helpful for poorly instrumented regions like Sumatra. Here, we use
the P-wave data from each of the 5374 ISC earthquakes in 2000–2016 at two reference stations (GSI and PSI;
Figure
1a
) as templates and scan through the continuous waveforms of these stations using a moving window.
The search starts in 2004, when CTBTO hydrophones and DGAR data become available. A new detection is trig-
gered if any of the two stations GSI and PSI have a waveform cross-correlation (CC) coefficient higher than 0.6
between the template and scanned waveforms. The location of the newly detected event is assigned the same as
its associated template. Then all the detected events from the two stations are merged to form a new catalog. We
discard some obviously false detections, for example, due to instrument glitches, and obtain a total of >500,000
new detections (Text S1 in Supporting Information
S1
; Figure
2
). Note that false detections may be remaining,
Figure 1.
Study area and
T
-wave from CTBTO hydrophones and DGAR. (a) Map of the earthquakes, seismic stations and hydrophones. The black stars are the
5374 ISC cataloged earthquakes around the Nias Island in 2000–2016 and the red stars represent the repeaters used to infer ocean temperature changes. The orange
star indicates the 2005 M 8.6 Nias–Simeulue earthquake. The trench axis is depicted by the black line with triangles. Inset: The cyan circles show the two CTBTO
hydrophone receivers (H08 at Diego Garcia and H01 at Cape Leeuwin). The black triangles show the
T
-wave station DGAR and four reference seismic stations used
in this study. (b)
T
-wave (1.5–2.5 Hz) from an 2014 M3.8 earthquake in the Nias region as recorded at DGAR, H01, and H08. The corresponding
T
-wave paths are
illustrated in the subset of (a).
21699356, 2023, 9, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2023JB026687 by California Inst of Technology, Wiley Online Library on [03/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License