Main Perth Thermometry Science People Engineering The Experiment Epilogue
Useful Links
Site Search

Acoustic Thermometry of Ocean Climate

While HIFT was a successful experiment in that it answered the basic questions concerning the propagation of man-made sound over antipodal distances, it was soon realized that the ocean's basins were undergoing different kinds climactic variations. The antipodal ranges of HIFT therefore integrated across these different regimes, which is not what one wants to understand the ocean's climate system. The thinking therefore became that basin-scale transmissions were a more appropriate way to measure the ocean's climate changes.

ATOC in the Pacific

Two sound sources were installed for the first phase of the ATOC feasibility study, one on Pioneer Seamount off central California and one north of Kauai (Figure 1-1, Location of the ATOC Sound Sources and Receivers). The Pioneer source began transmitting in late 1995 and continued transmissions in accordance with MMRP protocols until it was turned off at the end of 1998. The Kauai source began transmitting in late 1997 and continued transmissions until October 1999, again in accordance with MMRP protocols. The signals transmitted by the sources were received on SOSUS receiving arrays in the North Pacific and, for part of the time, on a vertical receiving array located at Ocean Weather Station Papa (50°N, 145°W). The transmissions from the Pioneer Seamount source were also recorded at various times on vertical receiving arrays located near the Big Island of Hawaii and near Kiritimati (Christmas) Island. A small number of the Pioneer Seamount transmissions were recorded by a receiver off New Zealand. The signals from the Kauai source were also recorded by Russian scientists at a permanent, bottom-mounted receiver located off Kamchatka.


The primary objectives of the first phase of the ATOC feasibility study were to determine (i) the precision with which acoustic methods could be used to measure large-scale changes in ocean temperature and heat content and (ii) the effects, if any, which the acoustic transmissions would have on marine mammals and other marine life. The longer-range goals of ATOC were to use acoustic thermometry data to study seasonal and interannual temperature variability associated with a variety of oceanographic phenomena, such as El Niño/La Niña and the PDO, and to test and improve computer models of ocean circulation. The ultimate goal was to test and refine climate models in order to gain a better understanding of the Earth's changing climate, including the link between global warming and sea level rise.

A quick summary of acoustic tomography/thermometry

The basic idea of acoustic thermometry is simple. Sound travels faster in warm water than in cold water. The travel time of a sound signal from a sound source near Hawaii to a receiver near California, for example, will decrease if the intervening ocean warms up, and will increase if the ocean cools down. Acoustic thermometry is feasible because:
• The ocean is nearly transparent to LF sound, so that relatively weak acoustic signals can be detected over distances of many thousands of kilometers using appropriate signal processing techniques; and
• The speed at which sound travels in the ocean depends primarily on temperature. (Sound speed also increases with an increase in salinity, but in the open ocean deep water, salinity normally has only a small effect on the speed (Urick, 1983).)

Acoustic thermometry takes advantage of an acoustic waveguide deep within the ocean that traps and carries sounds over long distances. This waveguide, known as the "sound channel" or SOund Frequency And Ranging (SOFAR) channel, is centered on the ocean depth where the speed of sound is at a minimum. Above the sound channel axis, sound travels faster because the water is warmer; below, sound travels faster because the pressures are greater. Acoustic energy within the sound channel that would otherwise spread outward to higher or lower depths is refracted (bent) back toward the sound channel axis by this difference in speeds. The net effect is that the sound channel serves as a waveguide that transmits underwater sounds efficiently over long distances. The sound speed minimum varies in depth based upon the temperature profile at a given location. Since surface temperatures tend to decrease toward the poles, the sound channel axis generally is deepest in tropical waters and shallowest in Arctic waters. Typical depths of the sound channel in the Gulf of Alaska, for example, are 100-200 m (330-660 ft), but in warmer areas it is much deeper, on the order of 750-1000 m (2460-3280 ft). On the north shore of Kauai, the sound channel axis is nominally at 800 m (2625 ft), approximately at the depth of the Kauai sound source. Not all of the acoustic energy travels straight down the axis of the sound channel. Instead, some of the sound waves cycle up close to the ocean surface, where they are bent back down, cross the axis of the channel, and reach close to the ocean floor before being bent back once again toward the surface. By measuring the difference in travel time between sound that traveled a straight course down the axis of the sound channel and that which cycled in waves through various depths of the ocean, scientists can measure how ocean temperatures vary with depth. Acoustic travel times provide direct 3D measurements of the horizontally and vertically averaged temperature along the paths traversed by the sound, suppressing the effects of small-scale ocean variability that dominate measurements at a point. The great advantage of acoustic thermometry compared to other types of temperature measurements is that such averages are just what are needed to study large-scale ocean variability and long-term trends in ocean temperature. The information obtained is similar to that which would be obtained for the atmosphere by averaging data from many separate weather stations. In addition, mathematical techniques referred to as inverse methods are used to infer the horizontal and/or vertical structure of the temperature field by combining travel time data from acoustic signals that have traveled along different paths through the ocean. Information on the structure of the ocean temperature field is needed to understand, for example, how the atmosphere and ocean interact to determine our weather and climate and to study the effects of environmental variability on marine life.

Thermometry Results

Analyses of data from the ATOC project demonstrated that acoustic thermometry is a powerful tool for making routine measurements of large-scale ocean temperature variability and heat content, as originally hypothesized. The key results obtained to date are:

Comparison of measured travel times for transmissions from Kauai (blue) with travel times calculated using NOAA's 2005 World Ocean Atlas (gray). The time means have been removed from all of the time series. Approximate estimates of temperature perturbations averaged along the ray paths inferred from the travel times are shown on the right-hand axis. The results from WOA05 are repeated each year.
(i) Acoustic travel times can be measured with a precision of about 20-30 milliseconds (msec) at 3000-5000 km (1620-2700 nm) ranges. For comparison, the total travel time for an underwater acoustic signal over 5000 km (2700 nm) is nearly an hour. ATOC data measurements proved to be more precise than originally thought possible. The initial concern that acoustic scattering from small-scale ocean structure, such as internal waves, might make accurate measurements of acoustic travel times impossible at 3000-5000 km (1620-2700 nm) ranges proved to be unfounded. Transmissions over these long ranges are needed to measure ocean gyre-scale variability, which is the scale on which ocean climate fluctuations are expected to occur. An ocean gyre is a large, ocean-basin size (on the order of a few thousand kilometers or nautical miles), roughly circular motion of surface water in response to wind forcing. The travel times can then be used to estimate the range- and depth- averaged temperature with a precision of about 0.006°C (0.01°F) at ranges of 3,000-5,000 km (1620-2700 nm) (Dushaw, 1999; Worcester et al., 1999).
(ii) Range- and depth-averaged temperature estimates made from the acoustic travel-time data are consistent with direct temperature measurements made with instruments lowered from ships (Worcester et al., 1999).
(iii) The observed travel time changes can be clearly related to known ocean processes. The ocean tides are well known from other measurements, and their effect on the acoustic travel times can be predicted, providing what is essentially a large-scale test signal. The measured and predicted travel time fluctuations at tidal frequencies are in excellent agreement out to 5000-km (2700-nm) range. One of the significant sources of LF sound transmission variability related to ocean temperature is seasonal change, with the upper ocean warming during summer and cooling during winter. The ATOC data show corresponding seasonal changes in travel times, as expected, particularly for acoustic paths that travel north of the Subarctic Front, where the seasonal temperature changes extend to significant depths, rather than being confined to a shallow seasonal thermocline (Dushaw et al., 1999).
(iv) The range and depth-averaged temperatures derived from ATOC are consistent with and complementary to related estimates derived from measurements of sea-surface height. The acoustic thermometry data from the Pioneer Seamount source have been used in conjunction with measurements of sea-surface height made by the TOPEX/POSEIDON satellite altimeter to test and constrain a computer model of the ocean circulation in the North Pacific (ATOC Consortium, 1998). Sea-surface height is related to ocean temperature because of thermal expansion. It was found that previous interpretations of sea-surface height variability as being primarily due to ocean temperature changes are inaccurate. The effects on sea-surface height of varying ocean salinity and ocean currents also appear to be significant. This result is important because it affects the way in which sea-surface height data are used to test and constrain ocean circulation models. This result is also important because it means that satellite altimetry data and acoustic thermometry data are complementary, providing independent information on ocean structure. The altimeter has excellent horizontal but poor vertical resolution, and the acoustic data provide information from the ocean interior with moderate vertical resolution but poor horizontal resolution. Both have good temporal (i.e., time-related) resolution. Consistent results for the seasonal heat storage in the ocean are found when the acoustic and altimetry data are combined with a computer model of the ocean general circulation. The two data types are both found to be important in constraining the model, with the combination providing more information than either data type alone.

ATOC and Numerical Ocean State Estimates

Sound speed at 300-m depth in the North Pacific Ocean derived by various means shown with the ATOC acoustic thermometry paths. The sound speeds estimates were derived from: the NOAA World Ocean Atlas'05 (WOA'05), upper ocean temperature profiles produced by an objective analysis (OA) procedure that combines satellite altimetric height with in situ temperature profiles, the JPL-ECCO solution, and a high-resolution integration of the POP model. The WOA05 field is the annual mean, the objective analysis is for winter 2003, and the JPL-ECCO and POP estimates are for January 2003. The bottom panels show the differences between the fields derived from the various estimates and the field derived from the annual mean WOA05.
In a test of the accuracy of ocean state estimates (estimates of the ocean's temperature and salinity as a function of time by combining available measurements with numerical modeling of the oceans) obtained by various means, the measured travel times were compared with travel times computed using several independent estimates of the North Pacific. The differences between measured and computed travel times were sometimes substantial, indicating that acoustic thermometry data provides significant additional constraints for numerical ocean models beyond those provided by existing data. The computation of these differences is the first step in the process of assimilating the acoustic data as useful constraints in ocean state estimates.

Marine Mammal Research Program Results

The California and Hawaii ATOC MMRPs were designed to determine the potential effects of the acoustic transmissions on marine mammals and other marine life. They consisted of multiple components, including:

  • Aerial surveys designed to determine any changes in the abundance and distribution of marine mammals in the vicinity of the Pioneer Seamount source;
  • Elephant seal tagging studies designed to determine any changes in elephant seal migratory or diving behavior in response to the Pioneer Seamount source transmissions;
  • Playback studies to humpback whales off the Kona-Kohala coast of Hawaii designed to look for behavioral changes in response to ATOC-like sounds prior to the actual ATOC source transmissions north of Kauai;
  • Aerial surveys designed to determine any changes in the abundance and distribution of humpback whales north of Kauai when the ATOC source was transmitting compared to measurements made in previous years when the source was not transmitting;
  • Visual observations of humpback whale abundance, distribution, and behavior north of Kauai to determine if there were any changes in response to the ATOC transmissions;
  • Undersea acoustic recordings made with seafloor data recorders north of Kauai to determine any changes in humpback vocalizations in response to the ATOC transmissions;
  • Auditory measurements on small toothed whales (odontocetes) to determine their sensitivity to the frequencies transmitted by the ATOC sources; and
  • Playback studies to fish at the Bodega Bay Marine Laboratory designed to look for behavioral changes in response to ATOC-like sounds.

Abundance and distribution. During the MMRPs conducted in both California and Hawaii, there were no observations of overt or obvious short-term changes in the abundance and distribution of marine mammals in response to the transmissions of the ATOC sound sources. No species were observed to vacate the area around the sound sources during transmissions. Intensive statistical analyses of aerial survey data showed some subtle shifts in the distribution of humpback (and possibly sperm) whales away from the Pioneer Seamount source during transmission periods. No statistically significant shifts in distribution were found for any other species of marine mammal. Visual observation data from the Kauai MMRP showed a similar small shift in mean distance of humpback whales away from the Kauai source during transmission periods.

Behavioral measures. During the MMRPs conducted in both California and Hawaii, there were no observations of overt or obvious short-term changes in the behavior of humpback whales in response to the playback of ATOC-like sounds, nor elephant seals or humpback whales in response to to transmissions of the ATOC sound sources. Intensive statistical analyses revealed some subtle changes in the behavior of humpback whales in response to the playback of ATOC- like sounds and to the transmissions of the ATOC Kauai source (Frankel and Clark, 1998; Frankel and Clark, 2000). The study results showed that the distance and time between successive whale surfacings (segment length and segment duration) increased slightly with increasing sound levels. This result is not what would be predicted, in that if the animals were stressed by the sound source, it might be expected that they would remain at the surface longer because of the lower received levels there. Longer dive durations would correspond to increased exposure to the sound source. No statistically significant changes were found in any other behaviors measured.

Vocalizations. The Hawaii MMRP did not find any overt or obvious short-term changes in the singing behavior of humpback whales in the vicinity of the sound source north of Kauai. No statistically significant changes in the underwater sound output from humpback whales in one of the frequency bands in which they vocalize was found in the vicinity of the Kauai source.

Audiograms. The hearing sensitivity of two species of dolphins to the ATOC sound was measured behaviorally (Au et al., 1997). Audiograms showed that their hearing is poor at the frequencies transmitted by the ATOC sources. The animals would have to be extremely close to an ATOC source simply to be able to detect the transmissions.

Fish. Preliminary playback studies of ATOC-like sounds to fish found no statistically significant responses (Klimley and Beavers, 1998). All of the effects detected by the MMRPs were subtle and found only after intensive statistical analyses. Bioacoustic experts concluded that these subtle effects would not adversely impact the survival of an individual whale or the status of the North Pacific humpback whale population (Frankel and Clark, 2000).

Main Perth Thermometry Science People Engineering The Experiment Epilogue
© Copyright 2007 by the University of Washington. All rights reserved.
Contact information for HIFT