Ocean Acoustic Tomography Has Heat Content, Velocity, and Vorticity in the North Pacific Thermohaline Circulation and Climate

Home , Internal tide

B. Dushaw, G. Bold, C.-S. Chiu, J. Colosi, B. Cornuelle, Y. Desaubies, M. Dzieciuch, A. Forbes, F. Gaillard, Brian Dushaw, Bruce Howe A. Gavrilov, J. Gould, B. Howe, M. Lawrence, J. Lynch, D. Menemenlis, J. Mercer, P. Mikhalevsky, W. Munk, Applied Physics Laboratory and School of Oceanography I. Nakano, F. Schott, U. Send, R. Spindel, T. Terre, P. Worcester, C. Wunsch, College of Ocean and Fisheries Sciences A Strategy for the Role of Acoustic Tomography in Ocean Climate Observation. OceanOcean AcousticAcoustic Tomography:Tomography: 1970–21st1970–21st CenturyCentury University of Washington in the 21st Century, C.J. Koblinsky and N.R. Smith (eds), Bureau of Meteorology, Melbourne, Australia, 2001. http://staff.washington.edu/dushaw

ABSTRACT PROCESS EXPERIMENTS Deep Convection—Greenland and Labrador Seas ATOC—Acoustic Thermometry of Ocean Climate Observing the Ocean in the 2000’s: Oceanic convection connects the surface ocean to the deep ocean with important consequences for the global Since it was first proposed in the late 1970’s (Munk and Wunsch 1979, 1982), ocean acoustic tomography has Heat Content, Velocity, and Vorticity in the North Pacific thermohaline circulation and climate. Deep convection occurs in only a few locations in the world, and is difficult The goal of the ATOC project is to measure the ocean temperature on basin scales and to understand the evolved into a remote sensing technique employed in a wide variety of physical settings. In the context of In: Observing the Ocean to observe. Acoustic arrays provide both the spatial coverage and temporal resolution necessary to observe variability. The acoustic measurements inherently average out mesoscale and internal wave noise that long-term oceanic climate change, acoustic tomography provides integrals through the mesoscale and other The 1987 reciprocal acoustic tomography experiment (RTE87) obtained unique measurements of gyre-scale deep-water formation. (Worcester et al., 1993; Pawlowicz et al., 1995; Morawitz et al., 1996; Sutton et al., 1997; contaminate point measurements. (The ATOC consortium, 1998; Dushaw et al., 1999; Worcester et al., 1999) high-wavenumber noise over long distances. In addition, tomographic measurements can be made without temperature and currents. (Dushaw et al., 1993, 1994) Send and Kindler, 2001) risk of calibration drift; these measurements have the accuracy and precision required for large-scale ocean The 0–1000-m 80 climate observation. The trans-basin acoustic measurements offer a signal-to-noise capability for observing 80¡N˚ 60° 0.4 TOPEX ATOC California to l N depth-averaged ATOC GE ocean climate variability that is not readily attainable by an ensemble of point measurements. BER TS 0.2 SPI W temperature measurements 78¡ EST 0

CURRENT l On a regional scale, tomography has been employed for observing active convection, for measuring changes SPI 59˚N -0.2 WOA'94 (red, with error bars) are TSBER convection expected k Pioneer NLAND -0.4 1995 1996 1997 compared with historical data 76¡ E 30° in integrated heat content, for observing the mesoscale with high resolution, for measuring barotropic 76˚ E Seamount R G n G 1 EN N T d (WOA ’94) and the S C 0.4 currents in a unique way, and for directly observing oceanic relative vorticity. The remote sensing capability 5 U f California to k GREE EA R r NLAND R ENT 74 C) temperature derived from ¡ 6 2 0.2

has proven effective for measurements under ice in the Arctic (in particular the recent well-documented o o 4 Kauai 0 TOPEX/POSEIDON (T/P) J AN 3 temperature increase in the Atlantic layer) and in regions such as the Strait of Gibraltar, where conventional vla1 -0.2 72 MA 57˚N ¡ altimetry (blue, assuming sea 72˚ YEN in-situ methods may fail. CURRE -0.4 1995 1996 1997 NT 0° surface height variations are K11/K21/K31 7 0.4 0¡ vla2 California to n caused only by thermal 150 As the community moves into an era of global-scale observations, the role for these acoustic techniques will JAN MA 90 20¡W YEN I. 180 120 E ° 0.2 10¡ 150 20˚ E ° expansion). The means have W 10 10˚ ¡ 0¡ be (1) to exploit the unique remote sensing capabilities for regional programs otherwise difficult to carry out, ° ° 0 10˚ 0˚ ° been removed from all time (2) to be a component of process-oriented experiments in regions where integral heat content or current data K12/K22/K32 -0.2 55˚ N series. Note the smoothness are desired, and (3) to move toward deployment on basin scales as the acoustic technology becomes more 30° -0.4 1995 1996 1997 600 and small error bars of the

) Labrador

2 400 robust and simplified. 200 0.4 Kauai to n 0 acoustic data, especially for (W/m nz Heat Flux -200 0.2 100 60˚W56˚W52˚W 0 the long (5 Mm) paths. In the ° 50 60 (%) )

0-1000 m Average Temperature ( -0.2 lower two panels, the annual Ice Cover -2 0 4 -0.4 1997 1998 1999 cycle has been removed from 0 Horizontal Integral from Tomography J m BASICS Local Estimate at Boundary Mooring the T/P data; the acoustic 9 Geometry of the RTE87 gyre-scale The low frequency (daily averaged) Comparisons of the acoustically 0.4 3 Local Estimate at Central Mooring Kauai to d

10 2 ATOC acoustic paths in the North rays on these particular paths reciprocal acoustic tomography barotropic currents for each leg and the determined heat content change Surface Heatflux Integral + 70 W/m 0.2 Sound travels faster in warm water than in cold x experiment, with acoustic transceivers areal averaged relative vorticity. The in the top 100 m (curves between 2 Pacific. The letters k, l, etc. are 0 sample below the seasonally water. By measuring the travel time of sound over a arbitrary names of receivers such as -0.2 varying surface layers, so at locations 1, 2, and 3. The ranges of vorticity is calculated by integrating the day 150 and 260) with other 1997 1998 1999 1 -0.4 known path, the sound speed and thus temperature the acoustic paths were 750, 1000, currents around the experiment triangle. estimates (triangles from XBTs, 1 the U.S. Navy SOSUS arrays. they do not observe the Depth (km) can be determined. Sound also travels faster with a and 1250 km. The calculation of vorticity cancels much points with error bars from CTDs). 0 annual cycle. current than against. By measuring the reciprocal of the variability of the currents. travel times in each direction along a path, the -1

2 -2 absolute water velocity can be determined. Each Oct Nov Dec Jan Feb Mar Apr May Jun 0 0.1 C) 1988 1989 (˚C) ° 11.5 0-1300 m Heat Content ( acoustic travel time represents the path integral of Barotropic and Baroclinic Tides -1.6 -1.2 -0.8 -0.4 0 Jul Oct Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr Jul θ (˚C) the sound speed (temperature) and water velocity. 1997| 1998 | 1999 | 2000 Over the past several years a number of interesting observations of oceanic tidal variability have been made 26 As the sound travels along a ray path, it inherently 11

by long-range acoustic transmissions. These observations are examples of how line-integral observations N 24 averages these properties of the ocean, heavily (Top) Geometry of the tomography array in the (Top) In the Labrador Sea, the tomography mooring arrays were Acoustic paths going through eddies and fronts. provide significantly better signal-to-noise capability than point measurements for detecting large-scale 10.5 1997 1998HOT filtering along-path horizontal scales shorter than Greenland Sea during 1988–1989. A deep deployed in different years in the area where deep convection 22

In ocean acoustic tomography, data from a oceanic variability. This capability is a major advantage of acoustic tomography. Latitude 20 the path length. Over a 1000-km range, a depth- multitude of such paths crossing at many convective chimney was observed near the center activity is expected (shaded). 10 Hawaii averaged temperature change of 10 m° C is easily of the array during March 1989 (shaded region). different angles are used to reconstruct the Data from RTE87 and AMODE have shown that tomography can measure the harmonic constants of 164 162 160 158 156 154 (Bottom) A three-year time series of 0-1300-m heat content HOT TOPEX o measured as a 20-ms travel time change. sound speed (temperature) and velocity fields. barotropic tidal currents to within about 2% uncertainty. The uncertainty in the harmonic constants derived 9.5 Longitude W (Munk, Worcester, and Wunsch, 1995) (Bottom) Time-depth evolution of potential derived from tomographic data (cyan, dots) along the section 1996 1997 1998 from available current meter records appears to be about 20%, even when attempts have been made at temperature averaged over the chimney region. marked in the top panel and derived from data obtained at the ( Temperature 0-1000 Average separating the barotropic and baroclinic modes. (Dushaw et al. 1995, 1997; Dushaw and Worcester, 1998) Typical rms uncertainty as a function of depth is moorings at the ends of the section (blue: K11, red: K12). The shown to the right. Total surface heat flux and daily dashed line shows a three-year warming trend, equivalent to a 0 averaged ice cover are shown above. Convection heat flux of 10 Wm-2. The dashed green curve is the time-integral The travel time data from all the various Data from the Hawaii Ocean Time-series (HOT) site (about 100 km 21:00 45¡ N occurs in March, with cold water extending from the of the NCEP surface heat fluxes, corrected by the addition of acoustic paths can be used to reconstruct north of Oahu) are plotted with T/P data. The 0–1000-m averaged 18:00 surface to depth. It occurs when the area becomes 70 Wm-2, averaged over the section. the depth-averaged horizontal sound speed temperature is plotted on the same scale as above. The point 1 SR ± ° 40¡ ice-free and large amounts of heat are lost to the (temperature 0.125 C, 0–1000-m depth measurements have large fluctuations in temperature caused by 15:00 cold atmosphere. average) field. This map gives an indication mesoscale variability. The HOT data points show the average and 2 12:00 of the scales that can be resolved. This map rms of 10–20 CTD casts obtained during each HOT cruise. The 35¡ can be made every few days or so, and differences between the temperature inferred from T/P and the 09:00 when spatially averaged, can yield average direct measurement at HOT are comparable to the temperature 3 USA Depth (km) BASIN SCALES ocean temperature with high accuracy. signal observed in the line-integrating data. 06:00 1000 30¡ 2000

Cuba 4000 03:00 Midway 80 4 0 K 0 1 ACOUS—Arctic Climate Observations Using Underwater Sound 40 Turning 75 5 25¡ Longitude 4 Bermuda 00:00 Hour HAWAIIAN RIDGE Latitude 6 P 1 70 .R. Trench Puerto 3 5 Hawaii 2 5500 Rico 5500 1.50 1.52 1.54 0 100 200 300 400 500 600 Ray Prediction o The Arctic is ideally suited to measurement by acoustic tomography. It has low acoustic noise levels and no W 65 20¡ THE FUTURE Sound Speed (km/s) Range (km) 441 442 443 444 445 30 internal waves so that the acoustic propagation is very clean. Travel times of the first several acoustic modes are 60 5500 25 Travel Time (s) o N 180¡ 170¡ W 160¡ 150¡ 140¡ 20 readily resolvable, and these modes naturally sample the layers in the water column that are of oceanographic Latitude Ocean acoustic tomography will continue to be used to study ocean processes that can only be addressed An ocean sound speed profile and a set of acoustic rays for interest. Finally, it is difficult to access the Arctic water column by conventional techniques; acoustics provide perhaps the only way to remotely sense the sub-surface variability. (Mikhalevsky et al., 1995, 1999) using the unique sampling of the method (e.g., integral heat content and current velocity) or remote sensing 620 km range. A sound speed minimum occurs at about 1 km The acoustic receiver detects the arrival of multiple Schematic diagram showing the relation of the The AMODE array detected diurnal tidal signals (K and O capability (e.g., the Arctic and Strait of Gibraltar). The present ATOC basin scale array will continue to be depth. This deep ocean sound channel, or waveguide, causes 1 1 pulses, and each pulse can be identified with a mode-1 internal tide radiating from the Hawaiian frequencies) of the lowest internal wave mode. The pre- expanded, and new arrays are in the planning process for the Atlantic and the Indian Oceans. acoustic energy to be trapped in the water column, so that it can predicted ray path. These are recorded and Ridge to the acoustic array. The dashed lines dicted horizontal variation of the displacement associated propagate great distances without interacting with the ocean predicted arrival patterns for a 620-km path south of (http://deos.rsmas.miami.edu/; SCOR WG96, 1994; Barton et al., 2000) represent the crests of the waves with 160 km with this mode for the K1 frequency is shown as the colored SCICEX / TAP / APLIS-ACOUS Transarctic Sections bottom. The sound channel also causes multiple acoustic paths, Bermuda. The World Ocean Atlas was used to wavelength. The beam pattern shows the wave. The energy density of this resonant wave is about 9 called "eigenrays", between an acoustic source and a receiver. 99 make the predictions. Recorded arrivals are slightly 2.6 1 (a) directional sensitivity of the acoustic sampling. twice that of its barotropic-tide parent. The 40 m°C ril R Ap (a) t The different turning depths of the paths cause the acoustic en earlier than predicted, suggesting that temperatures im amplitude wave is trapped between the Puerto Rico shelf er xp ct E e 2.4 ns APLIS-ACOUS '99 pulses along those paths to have different travel times. were above their climatological mean. US ra 4 ° T 9 and the turning latitude at about 30 N. D 9 /ACO T 1 S C il SCICEX '98, '99 60 LI pr R EX A AP C t I en SC rim e SCICEX '95 2.2 p R Ex 50 C) P A o T Greenland

Control Points—Strait of Gibraltar

2 40 AMODE-MST—Mesoscale Mapping Over a Large Area 0.5o C increase The Strait of Gibraltar is a control point for the entire Mediterranean Sea. Mass, heat, and salt are exchanged TAP '94 R 30 One of the original goals of tomography was to synoptically observe the mesoscale variability of the ocean. The with the Atlantic Ocean. The intense and highly turbulent flow makes tomography an attractive measurement 1.8 (b)(b) R

Temperature ( 20

Acoustic Mid-Ocean Dynamics Experiment/Moving Ship Tomography (AMODE-MST) obtained high resolution, Latitude approach because it provides the necessary integration to estimate net transports. A pilot experiment has o 0.4 C increase nearly synoptic 3-D maps of the ocean over a large area. (Cornuelle et al., 1989, The AMODE Group, 1994) 1.6 been conducted and a permanent system for sustained observations is planned. (Send et al., 2000) Seasonal Max 10 o R Average historical maximum =1.44 C 1.4 T1 T2 0 0 Seasonal Min +2 -10 (Top Left) Sound speed perturbation at 700 m depth 200 1.2 Historical1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 280 300 320 340 –2 as measured during the MST experiment during July 400 (1980s) Longitude -1 ≈ ° Year 15-30, 1991; 1 ms 0.25 C. The ship with an acoustic (a) 36 10’ (a) Location of acoustic instruments at the Gibraltar N 600

receiving array steamed around the 1000-km diameter Depth (m) –1 eastern entrance of the Strait of Gibraltar. Algeciras R Two cabled ATAM moorings with shore Acoustic thermometry paths Regional acoustic tomography arrays (a) Acoustic data will be obtained on this A notional array for the North Atlantic circle, stopping every 3 hours (~ 25 km) to receive 800 terminus (Alert, CAN; Barrow, AK) Cable R 200 (b) In the 1994 Trans-Arctic Acoustic Propagation (TAP) experiment, Cabled ATAM moorings Pacific water circulation in the North Pacific: The Central minimal acoustic array for the next five consists of 4 cabled-to-shore or signals from six moored sources. The acoustic travel SPAIN (b) Horizontal and depth sampling by ray Autonomous sources transmissions were made from a site north of the Svalbard Atlantic water circulation Equatorial Pacific Tomography years. The Kauai ATOC source and U. S. autonomous acoustic sources, together time data were inverted to produce this field. T3/HLAT3//HLA 0 5 10 15 20 paths along path T1-T2. Range (km) Experiment (CEPTE) just north of the Navy SOSUS receivers are used. with about 10 receivers at SOSUS sites L Archipelago across the entire Arctic Ocean to receiving arrays located Tarifa in the Lincoln Sea and the Beaufort Sea. The figure shows the equator has recently finished. The and on DEOS moorings. Simple T2 36 00’ 36.2 36.6 37 37.4 37.8 38.2 38.6 (Bottom Left) This map was obtained using only air- 200 (c) A two-week comparison of along-strait 600 J Salinity (ppt) temperature of the Atlantic Layer obtained from historical climatology Hawaiian Ocean Mixing Experiment (b) Fairly dense acoustic sampling of the receivers mounted on moorings-of- expendable bathythermograph (AXBT) data at 700 m 400 current averaged along a deep-turning ray A future notional monitoring grid in the Arctic U 800 and from the (submarine) SCICEX 1995, 1998, and 1999 transects. Ocean exploiting synoptic acoustic remote (HOME) tomography arrays were North Pacific can be obtained with the opportunity can further increase the (July 18–22). 600 across the Strait (black) and the tidal and 200 The SCICEX transects were close to the 1994 TAP and 1999 deployed in 2001. The Kuroshio addition of acoustic sources: a replacement acoustic sampling at little additional 400 T1 low-frequency flow field determined from a sensing with in situ measurements cabled to ACOUS/APLIS propagation paths. The red arrows indicated the shore at Alert, Canada, and Barrow, Alaska. Extension System Study (KESS) source off the coast of California, a source cost. Tomographic instruments deployed The panels at right show the line-integral or point MOROCCO 50 wide range of direct current observations 200 Ceuta change in the maximum temperature inferred from the acoustic travel tomography array may be deployed in near Japan, an "H2O" acoustic source by European organizatons in the North sampling that was used to obtain the maps on the left, (red). The agreement is very good. ATAM stands for Acoustic Thermometry and 5 40’W 5 30’ 5 20’ time changes. Whether these results are due to global climate warming the next few years. The North midway between Hawaii and California, Atlantic in the coming decade will with estimated errors. The two estimates of the sound 0 Autonomous Monitoring. or a natural oscillation is an area of active research. Equatorial Current Bifurcation Region and an acoustic source that might be contribute to this array. Such a network speed are within the uncertainties of each method. -50 (c) Experiment in the Philippine Sea may attached to a DEOS mooring in the central would complement existing observation mean current [cm/s] 25 Apr 29 Apr 3 May begin in 2006. North Pacific. systems for ocean climate variability.