Abstract

Deployment of Deep Argo regional pilot arrays is underway as a step toward a global array of 1250 surface-to-bottom profiling floats embedded in the upper-ocean (2000 m) Argo Program. Of the 80 active Deep Argo floats as of July 2019, 55 are Deep Sounding Oceanographic Lagrangian Observer (SOLO) 6000-m instruments, and the rest are composed of three additional models profiling to either 4000 or 6000 m. Early success of the Deep SOLO is owed partly to its evolution from the Core Argo SOLO-II. Here, Deep SOLO design choices are described, including the spherical glass pressure housing, the hydraulics system, and the passive bottom detection system. Operation of Deep SOLO is flexible, with the mission parameters being adjustable from shore via Iridium communications. Long lifetime is a key element in sustaining a global array, and Deep SOLO combines a long battery life of over 200 cycles to 6000 m with robust operation and a low failure rate. The scientific value of Deep SOLO is illustrated, including examples of its ability (i) to observe large-scale spatial and temporal variability in deep ocean temperature and salinity, (ii) to sample newly formed water masses year-round and within a few meters of the sea floor, and (iii) to explore the poorly known abyssal velocity field and deep circulation of the World Ocean. Deep SOLO’s full-depth range and its potential for global coverage are critical attributes for complementing the Core Argo Program and achieving these objectives.

1. Introduction: The value of Deep Argo

The Argo Program (Roemmich et al. 2009; Riser et al. 2016) has transformed ocean observation by deploying and sustaining a global array of nearly 4000 autonomous floats collecting temperature and salinity profiles from the sea surface to 2000 m every 10 days. Argo was designed (Argo Science Team 1998) to meet targets of spatial coverage, depth, and data quality that were practical at the time, and to do so in a cost-effective manner. The targets have evolved over Argo’s 20-yr history as the float and sensor technologies have matured and improved. Argo’s increasing capabilities, including the possibility of sampling to 6000 m, are enabling new applications in basic research, climate assessment, education, and ocean reanalysis and prediction that were not possible in Argo’s early years.

A compelling case exists for extending Argo to the sea floor. Argo was conceived as an upper-ocean observing system (Argo Science Team 1998) only out of practical necessity and historical context. Twenty years ago, profiling floats were not capable of operation below 2000 m, and profiling float CTDs were not sufficiently accurate or stable to measure abyssal property fluctuations. Argo was designed to replace and improve on the broadscale XBT networks (Smith et al. 1999), which measured temperature to about 800-m depth along shipping lanes. Relative to the XBT networks, Argo’s global coverage, 2000-m depth range, and its addition of salinity were all major steps forward. However, Argo’s scientific and operational objectives, including closure of the global heat and sea level budgets, clearly apply to the full depth of the ocean (e.g., Zilberman 2017). Half of the ocean’s volume is below 2000 m, and half of the ocean’s area is deeper than 4000 m. Less than 1% of the ocean’s area is deeper than 6000 m. The variability of temperature and salinity, including their multidecadal trends, extends to the ocean bottom, and is of particular interest in newly formed abyssal waters. Closure of Earth’s energy budget requires sampling to the sea floor, and trends in temperature and salinity are bottom intensified in some ocean basins (Purkey et al. 2019). The meridional overturning circulations have elements extending to the bottom in all oceans (e.g., Talley 2013), and observations of their transport and variability are among the ocean’s key challenges. The deep and abyssal oceans are sampled primarily by shipboard repeat hydrography (Talley et al. 2016), but with very sparse spatial coverage of track lines, and decadal repeat cycles. A full-depth Argo array will strongly complement the repeat hydrography program, filling the spatial gaps and observing the variability down to subseasonal time scales. There has never been a question of whether the Argo array should sample the complete ocean, but only of when it would become technologically feasible.

The Scripps Instrument Development Group (IDG) began the design of the Deep Sounding Oceanographic Lagrangian Observer (SOLO) 6000-m profiling float in 2011. This was motivated by the high scientific value of systematic deep ocean observations, by the technology advances in Argo profiling floats, and with support from the National Oceanic and Atmospheric Administration. Simultaneously and collaboratively with the Argo community, Sea-Bird Scientific began work on a deep ocean low-power CTD, the SBE-61, for profiling floats.

Deep SOLO has progressed from design through prototype to regional- and basin-scale deployments (Table 1 and Fig. 1). It is one of four different models of Deep Argo floats, including the 4000 m Deep Arvor (Le Reste et al. 2016) and Deep New Profiling Float of Japan (NINJA) and the 6000 m Deep SOLO and Deep Autonomous Profiling Explorer (APEX) instruments. There are 80 Deep Argo floats of all models presently operating and tracked by the World Meteorological Organization Intergovernmental Oceanographic Commission’s Joint Technical Commission for Oceanography and Marine Meteorology in situ Observing Programmes Support Centre (WMO-IOC’s JCOMMOPS) (Fig. 1). These include 47 Scripps Deep SOLO floats, 8 commercially built (MRV) Deep SOLOs, 20 Deep Arvors, 3 Deep APEXs, and 2 Deep NINJAs. The present floats have been deployed mainly in regional pilot arrays in the southwest Pacific basin, the South Australian and Australian Antarctic Basins, and the North Atlantic (Fig. 1). Each of these deep basins is not far downstream from formation regions of abyssal water masses, and each basin has open questions of variability and change in abyssal temperature and salinity. The regional deployments are the first phase of a global 5° × 5° Deep Argo array of about 1250 floats (Johnson et al. 2015). The pilot arrays will demonstrate the scientific potential of Deep Argo and, in collaboration with commercial partner Sea-Bird Scientific, will determine the accuracy and stability of Deep Argo CTDs. The ability of the Argo Program to deploy and sustain Deep Argo on a global scale will be developed and tested in the pilot arrays.

Table 1.

Deep SOLO prototype and regional pilot deployments.

Deep SOLO prototype and regional pilot deployments.
Deep SOLO prototype and regional pilot deployments.
Fig. 1.

Location of 80 operational Deep Argo floats as of July 2019. Red upright triangles are Deep SOLO. Red inverted triangles are MRV Deep SOLO. Blue circles are Deep Arvor. Green squares are Deep APEX. Orange pluses are Deep NINJA. (Source: JCOMMOPS). Shading indicates land (black) and ocean depths <2000 (blue-gray), 2000–4000 (gray), and >4000 m (white).

Fig. 1.

Location of 80 operational Deep Argo floats as of July 2019. Red upright triangles are Deep SOLO. Red inverted triangles are MRV Deep SOLO. Blue circles are Deep Arvor. Green squares are Deep APEX. Orange pluses are Deep NINJA. (Source: JCOMMOPS). Shading indicates land (black) and ocean depths <2000 (blue-gray), 2000–4000 (gray), and >4000 m (white).

There are varying approaches being taken in the design of 4000- and 6000-m floats, but issues of high pressure and the need for high data accuracy are encountered by all models. The 4000-m models are extended versions of 2000-m Core Argo floats, built into cylindrical pressure housings and with a 4000-m version of the Core Argo SBE-41 CTD mounted on the float top cap. The 6000-m models use a spherical glass pressure housing inside a plastic shell, with the self-contained SBE-61 CTD mounted in a separate titanium pressure case (Fig. 2), cabled to the glass sphere.

Fig. 2.

A Deep SOLO float is seen descending from the sea surface off San Diego. The spherical glass pressure housing is protected by a plastic shell. The Iridium/GPS antenna, SBE-61 CTD, and passive “bottom finder” wire can all be seen. The float’s pancake bladder is inside the cowling, behind the CTD. (Photo: IDG)

Fig. 2.

A Deep SOLO float is seen descending from the sea surface off San Diego. The spherical glass pressure housing is protected by a plastic shell. The Iridium/GPS antenna, SBE-61 CTD, and passive “bottom finder” wire can all be seen. The float’s pancake bladder is inside the cowling, behind the CTD. (Photo: IDG)

The focus of the present work is the Deep SOLO float model, owing to the novel aspects of its design and operation, and because it accounts for the majority of presently active Deep Argo floats (Fig. 1). Section 2 describes the lineage of Deep SOLO as the product of over 25 years of profiling float development and experience. In section 3, key aspects of Deep SOLO design are discussed. Section 4 details logistical, deployment, and operational characteristics of Deep SOLO. Finally, section 5 describes the suitability of Deep SOLO for the scientific vision of global full-depth Argo missions. The still-improving accuracy and stability of the SBE-61 CTD are important but peripheral here, and will be addressed in detail in a later analysis. The perspective of this work is both technical and scientific, to argue that a Deep Argo global array is practical to implement, and that it will have a transformative impact on ocean observing, continuing the Argo revolution of the past 20 years.

2. The pedigree of the Deep SOLO float

a. ALACE

The most common of the profiling floats deployed during the 1990s’ World Ocean Circulation Experiment was the Autonomous Lagrangian Circulation Explorer (ALACE; Davis et al. 1992). This instrument, including its profiling version (P-ALACE), was developed collaboratively by the IDG under R. Davis, and the Webb Research Corporation under D. Webb. ALACE operated between the sea surface and about 1500 m, using a reciprocating pump to adjust its buoyancy by transferring oil between internal and external bladders. These instruments had a few shortcomings, most significantly that the pump sometimes failed when air bubbles formed in the hydraulic system and were drawn into the pump. Also, the float’s buoyancy could be increased but not decreased under high pressure.

b. SOLO

The limitations of ALACE floats led to reconsideration of the hydraulic system, resulting in the IDG’s SOLO (Davis et al. 2001) and separately in the Webb APEX. The ALACE reciprocating pumps were replaced by single cylinder pumps, eliminating the problems with air bubbles, replacing the internal oil reservoir with the cylinder volume, and enabling buoyancy reduction under pressure. Low-pressure air pumps were added, to inflate an external air bladder and increase the instrument’s freeboard on the sea surface for data transmission. The principal limitation of APEX and SOLO is that the volume of the pump cylinder defines the capacity for buoyancy adjustment. Larger pump cylinders made the floats heavier and less efficient. Without an additional means of buoyancy adjustment, it was not practical to profile from Argo’s target pressure, 2000 dbar, to the sea surface in warm tropical waters having a large density difference over that pressure range. Moreover, the added air pump and external air bladder introduced new vulnerabilities. In spite of these limitations, the APEX float remains in wide use today, accounting for nearly 40% of all active Argo floats in mid-2019.

c. SOLO-II

The tradeoffs between the ALACE and SOLO designs for use in Argo were revisited by R. Davis and IDG engineers, and in 2010 deployment began of the next-generation SOLO-II Argo float. A return to reciprocating pumps, custom designed for this application and fabricated in-house, was attractive for achieving global 2000-m capability and to reduce the mass and volume of the floats for greater energy efficiency and easier handling and transport. At 19 kg, The SOLO-II is about 30% lighter and smaller than SOLO or APEX. The original ALACE’s problems with air bubbles were mitigated through a reconfiguration of the hydraulic system’s geometry, utilizing gravity for passive bubble removal at the pump intake. In addition, an active system uses an optical bubble detector and a low-pressure pump for bubble removal, if needed. The addition of a hydraulic fuse in SOLO-II makes it possible to reduce the float’s buoyancy under high pressure. In a typical Argo mission, this means the float can descend from the sea surface to a parking pressure of 1000 dbar, make depth adjustments in either direction while drifting at this depth, and then descend farther to 2000 dbar to begin profiling during ascent. Profiling on ascent enables a near-real-time profile to be transmitted on surfacing. SOLO-II has no air bladder and expends only a modest amount of energy in surface pumping. The high pumping capacity of SOLO-II allows it to cycle from 2000 dbar to the sea surface anywhere in the World Ocean. Moreover, the float’s large buoyancy range enables identical ballasting for all floats deployed on a cruise, simplifying ballasting, operational planning, and deployment sequencing. The total energy consumption of the present SOLO-II, including pumping, communications, and continuous CTD profiling throughout ascent from 2000 dbar, is 8.1 kJ per cycle. With a (hybrid lithium) battery energy capacity of 5.4 MJ, of which about 75% can be utilized operationally, SOLO-II has sufficient energy for more than 10 years of 10-day Argo cycling, plus reserve energy for potential secondary missions such as profiling under extreme weather events. Following early teething problems with external oil bladders in 2010–11 initial deployments, SOLO-II has become an extraordinarily reliable platform. During the most recent three calendar years, 2016–18, the Scripps Argo Program deployed 265 SOLO-II floats, of which 263 remained active at the end of 2018 (Source: JCOMMOPS).

d. Deep SOLO

Given the compelling scientific and operational value of sampling the full water column, and the capability of the SOLO-II hydraulic system for high pressure and high pumping volume operations, it became apparent by 2011 that a 6000-m Deep SOLO float should be developed to satisfy Deep Argo requirements. Following design work on Deep SOLO beginning in 2011, a prototype Deep SOLO (SN 6002, WMO ID 5904318) was deployed off California in January 2013. This float profiled daily to the 4000-m bottom depth before being recovered in September of that year, and reconditioned in the laboratory. Float 6002 (WMO ID 5902341) and a second prototype, SN 6003 (WMO 5902342), were deployed in June 2014 by R/V Tangaroa in the southwest Pacific basin (Table 1). They profiled to about 5500 dbar every few days. Each float completed more than 100 deep cycles until being recovered by R/V Kaharoa in September 2015 for examination.

Since January 2016, 62 Deep SOLOs (not counting the prototypes; see Table 1) have been deployed in regional pilot arrays in the southwest Pacific basin, South Australian and Australian–Antarctic Basins, and western North Atlantic. A total of 55 Deep SOLO floats are active as of July 2019, with four others (6005, 6018, 6028, and 6053) having been recovered and returned to the laboratory for examination or repair. A fifth (6004) exhausted its batteries and two (6055, 6058) have likely failed, possibly having snagged the sea floor. The Deep SOLO’s hydraulic system has worked exceptionally well. Commercial production of the Deep SOLO float has been initiated by MRV Systems, with 8 MRV Deep SOLOs deployed as of July 2019 (Table 1) and over 28 more planned.

3. Design aspects of Deep SOLO

a. Glass sphere

Glass spheres provide greater strength and payload in seawater than cylindrical metal pressure housings and are more economical than titanium spheres. Carbon fiber cylinders (Le Reste et al. 2016) were also considered, but thought to be risky given reports of reliability issues in the field at that time. Based on payload buoyancy, cost, and reliability, glass spheres are the pressure housing of choice for deep ocean applications. Deep SOLO uses the borosilicate glass Vitrovex (Nautilus Marine Service) 33-cm-diameter glass sphere. This sphere, made of 1.2-cm-thick glass, is rated to 7000 m, and with its protective plastic shell, has a mass of 11.1 kg. Its buoyancy is approximately 11 kg, and the interior volume and total volume of the sphere are 15.0 and 18.8 L, respectively. When the sphere is loaded with hydraulics, batteries, electronics, and other internal components (Fig. 3), plus the external 2.8 kg SBE-61 CTD, the plastic hardhat, and other external hardware and parts, the mass of Deep SOLO is 27 kg. For the fully assembled Deep SOLO, the compressibility coefficient is α = 2.64 × 10−6 dbar−1, and the thermal expansion coefficient is β = 1.2 × 10−4°C−1. For a neutrally buoyant Deep SOLO at 6000 m, with T = 0.7°C, S = 34.7 PSS-78, ascent to the sea surface in tropical waters (T = 30°C, S = 33.4 PSS-78) requires an increase in buoyancy by 360 g (340 mL).

Fig. 3.

Cutaway schematic diagram of Deep SOLO showing the arrangement of components.

Fig. 3.

Cutaway schematic diagram of Deep SOLO showing the arrangement of components.

Glass spheres have been used in the deep sea as mooring flotation for many years. In those applications they are subject to extended periods at high pressure but are not repeatedly pressure cycled as is the case with profiling floats. With pressure cycling, microcracking and spalling can begin at the corners of the glass, that is, at the drilled holes and the faces of the two hemispheres that are mated together. During the development of Deep SOLO, glass spheres were pressure cycled hundreds of times and examined for cracks. Some spalling was observed, but the spheres were found to be capable of many more cycles than would be seen in Deep Argo missions. This conclusion was reinforced by examination of the prototype floats following their recovery. A couple of factors were found to affect the rate of spalling. One of these was the width of the chamfer, or beveled edge, on the inside of the sphere’s equator. Tests over 200 cycles, with a chamfer of 0.5 mm resulted in less than 20 mg of spalled glass, an order of magnitude less than for a 1.5-mm chamfer. The other preventative measure is to closely recreate the initial alignment of the two glass hemispheres relative to one another, each time they are mated.

New glass spheres are routinely inspected during float assembly and are rejected when irregularities are found. Another dimension to this problem is vibration, which can accelerate the growth of microcracks. Therefore, care has been taken to isolate the glass spheres from motor and hydraulic system vibration.

b. Hydraulic system

Deep SOLO requires 340 mL of mineral oil to be pumped from the internal bladder to the external pancake bladder in order to ascend from 6000 m to the ocean surface. Sufficient freeboard is needed for reliable Iridium and GPS reception and transmission. Deep SOLO’s internal reservoir is therefore filled with 600 mL of mineral oil to provide the reserve buoyancy on the sea surface or in case the float entrains sediment on the sea floor. The same custom-designed reciprocating pump is used for IDG’s Deep SOLO, SOLO-II, and the Spray glider. In the case of Deep SOLO, the pump bearings and other elements are strengthened for operation at 6000 dbar. The external pancake bladder has a capacity of 800 mL and is mounted with a titanium fitting on the bottom of the glass sphere. The external bladder can be directed either to the pump outlet or to the polyurethane interior reservoir by means of a motor-controlled high-pressure ball valve. The interior of the glass sphere is evacuated to 10 psia (7 dbar). When the valve is opened at the sea surface, the partial vacuum allows the oil to be pushed from the external bladder to the internal reservoir, decreasing the total displaced volume and causing the float to sink.

The pump’s energy consumption is roughly linear with pressure, pumping 0.75 mL s−1 and consuming 6 W of power at the sea surface and 0.5 mL s−1 and 60 W at 6000 dbar. Due to the high energy consumption at high pressure, pumping at depth is minimized.

c. Bottom detection

A crucial difference between Core and Deep Argo is that Core Argo floats are deployed and usually remain in waters deeper than their 2000-m profile depth while Deep Argo floats should descend to the sea floor repeatedly. Most Core Argo floats complete their life without contacting the bottom. Deep Argo floats should sample to the ocean bottom without being trapped by sediments or having the CTD fouled with sediments. Deep SOLO’s CTD is mounted horizontally on the bottom cowling of the float (Figs. 2 and 3), where the weight of the CTD helps keep the float upright, and the conductivity cell is always below the sea surface. With this CTD location, if the float rests on the bottom the CTD intake is very close to the sediments.

To mitigate problems of bottom contact, Deep SOLO uses a passive “bottom detection” system. A 3-m plastic-covered stainless steel wire hangs below the float. The float descends until it reaches a preset “profile depth” or the float controller determines from pressure data that it is no longer descending. The profile depth is typically set below the estimated bottom depth. In that case, descent stops when the weight in water of the “bottom detection” wire lying on the sea floor renders the rest of the float neutrally buoyant. This passive system reliably allows floats to get within 3 m of the ocean floor but keeps the CTD sensors safely above the bottom. The 3-m wire has sufficient weight in water (20 g m−1) to increase the depth at which the float achieves neutral buoyancy by 1.5 km.

The bottom detection system works well where the bottom depth is known to within the tolerance provided by the wire rope, which is roughly 800 m after discounting the 1.5-km value to allow for the inertia of the sinking float. The depth error tolerance is most likely to be exceeded in rough topography, where the float may occasionally contact the sea floor. Figure 4 shows a Deep SOLO (SN 6019) record of bottom approaches, including the bottom depth from bathymetric data (Smith and Sandwell 1997, updated 1-min bathymetry, https://topex.ucsd.edu/marine_topo/, in black), the maximum depth reached by the float (red), and the float’s profile depth parameter (green). When the profile depth is shallower than the bottom, for example around cycle 25, no bottom contact is made. When the profile depth is deeper than the bottom, the float finds the bottom, stopping within 3 m of the ocean floor. The bottom depths reported by the float are typically tens of meters above or below the depth from the gridded bathymetry. These differences are usually greater than the error of the float’s Kistler pressure sensor (<0.1%), so they may represent a corrected bottom depth. When the float’s maximum depth is much less than the gridded bathymetry (Fig. 4, cycle 64), it indicates the likelihood that a previously uncharted seamount has been discovered by Deep SOLO. At present most Deep SOLOs are operating over flat deep abyssal plains. More experience is needed to understand and weigh the trade-off between missing the deepest part of the water column if the profile depth is too shallow or having the float cowling run into the bottom if the profile depth is much too deep. The profile depth can be changed whenever the float is on the sea surface, and this is being done manually until an automated procedure is developed. A possible failure mode occurs when the bottom is unexpectedly shallow, and the parking depth is not shallow enough for the float to lift off the sea floor. It is conjectured that floats 6055 and 6058 may have failed by having their bottom detection wires drag along the rough bottom during the parking phase until becoming snagged. This failure mode will be addressed in firmware to ensure that the float does not park in contact with the sea floor.

Fig. 4.

Deep SOLO bottom detection. Time series for Deep SOLO SN 6019 (WMO ID 5902523) of target profile depth (green is the depth where descent would stop if there is no bottom contact), gridded bathymetry (black; Smith and Sandwell 1997), and float’s maximum depth (red; from pressure measurements). The float cycle time was 10 days through cycle 31 and is 15 days since cycle 31. It was deployed on 20 Oct 2016. The inset map shows the regional gridded bathymetry in the South Australian Basin and the float’s trajectory (light blue line) from deployment (green star) to the most recent position (red star).

Fig. 4.

Deep SOLO bottom detection. Time series for Deep SOLO SN 6019 (WMO ID 5902523) of target profile depth (green is the depth where descent would stop if there is no bottom contact), gridded bathymetry (black; Smith and Sandwell 1997), and float’s maximum depth (red; from pressure measurements). The float cycle time was 10 days through cycle 31 and is 15 days since cycle 31. It was deployed on 20 Oct 2016. The inset map shows the regional gridded bathymetry in the South Australian Basin and the float’s trajectory (light blue line) from deployment (green star) to the most recent position (red star).

d. Ice avoidance

Since the regional pilot arrays of Deep Argo floats include the Southern Ocean, near the Antarctic continent, it was important to implement an ice avoidance algorithm (Klatt et al. 2007) in Deep SOLO in order to protect its Iridium/GPS antenna from damage due to ice contact. At the heart of the algorithm is a sequence of temperature measurements made on ascent between (typically) 40 and 20 dbar. If the median temperature is colder than a specified value (typically −1.65°C), ice avoidance is activated. The ascent is stopped, and a new descent phase is initiated, with the previous profile(s) stored for later transmission. Two Deep SOLOs deployed near the Antarctic continental slope (SN 6041, 6042) in early 2018 remained below the surface to avoid possible ice during the subsequent winter and then surfaced again in the spring. Profiles with ice avoidance lack surface data and position information. The profiles are stored and transmitted on surfacing. Without communications, new mission parameters cannot be received by the float.

4. Operation of Deep SOLO

a. Logistics and deployment

Deep SOLO is shipped and deployed upright in a biodegradable fiberboard box measuring 53 cm wide × 85 cm long × 102 cm high (Fig. 5). The float is held immobile inside the box. The box provides protection for the float’s antenna and CTD, which could be damaged by contact during transit or on a rolling ship. The box is strapped (yellow straps in Fig. 5) onto a tray that is removed at deployment time. An inner strap (orange in Fig. 5) remains in place to hold the box closed during deployment until either water contact triggers a spring-loaded release or a manual release is pulled. Once the release lets go (and is recovered), the float separates from the box in a matter of seconds and immediately begins sinking.

Fig. 5.

A single box for shipping and deployment of Deep SOLO protects the float’s fragile antenna and CTD against breakage while being carried on a rolling deck and facilitates containerized shipping and efficient stowage at sea. The yellow strap and fiberboard tray are removed prior to deployment. The cutouts in the box are for ease in carrying (near top) and for accessing the float to initiate a self-test with a magnet swipe.

Fig. 5.

A single box for shipping and deployment of Deep SOLO protects the float’s fragile antenna and CTD against breakage while being carried on a rolling deck and facilitates containerized shipping and efficient stowage at sea. The yellow strap and fiberboard tray are removed prior to deployment. The cutouts in the box are for ease in carrying (near top) and for accessing the float to initiate a self-test with a magnet swipe.

To avoid rough handling during shipping, Deep SOLOs are typically shipped in batches as a full container load. This way, boxed floats are hand loaded into the container and secured by laboratory personnel at Scripps and hand unloaded either by laboratory personnel or collaborating partners at the destination. Once unloaded for shore-side storage or directly onto the deployment vessel, a final float self-test is initiated by a magnet swipe on the glass, while the float is sitting in a location with unobstructed sky. The float’s magnetic switch is accessed without opening the box, through aligned cutouts in the box (lower cutout in Fig. 5) and in the float’s cowling (Fig. 2). The pump, GPS, electronics, and CTD are all checked. At the conclusion of the self-test, engineering data are transmitted back to the SIO float laboratory via Iridium. Optionally, more thorough testing or firmware modification can be carried out by short-range serial radio communication (Xbee) through the float glass. After self-testing, the float is left negatively buoyant, with the exterior bladder emptied of oil. The emptied bladder signals “test passed” to the self-test operator. The float is then left in a low-power state, sometimes for several months, while awaiting deployment. Periodic sampling of the float’s pressure sensor, typically every 10 min, is used to detect deployment and initiate the float mission.

b. Deep SOLO mission parameters

Deep SOLO mission parameters, including profiling depth, parking depth, cycle time, and CTD sampling characteristics are easily adjustable from shore via Iridium. Some mission parameters are in common with SOLO-II and others are unique to the Deep Argo operation. During operation, after leaving the sea surface when the external bladder is emptied of oil, and sinking to about 50 dbar (or sometimes deeper on the initial cycle), the float pumps part of its oil back into the external bladder. The amount of oil is estimated so that the float will reach its profile pressure while still sinking at about 5 cm s−1. The initial sinking rate after pumping is about 20 cm s−1, slowing through the water column as the float passes through increasingly dense water. The total fall-time to 6000 dbar is about 13 h. The descent is arrested by pumping if not by bottom contact, and the float rises to its parking depth for its set parking duration, followed by the ascent back to the surface. The variable ascent speed is controlled by successive pumping intervals, with a typical rise time of 17 h from 5000 dbar. The first cycle is a shallow dive to about 200 dbar for collection of engineering data. This is followed by dives to 1000 and 4000 dbar before descending close to the sea floor. The staged increase in profile depth is for ballasting adjustment and early detection of pressure-dependent problems.

Since the Deep SOLO can increase but not decrease its buoyancy under high pressure, the parking phase occurs during ascent. To obtain an uninterrupted full-depth profile, the profile is therefore collected during the descent phase. Optionally, if a near-real-time profile is required, a secondary profile can be obtained on ascent after parking. The ducted inlet for the pumped SBE-61 is offset from the float hardhat (Fig. 2) to receive an undisturbed stream of water while either ascending or descending. Collection and comparison of paired descending and ascending profiles can be valuable for improving data quality (e.g., thermal mass or sensor offset issues).

Continuous and discrete CTD sampling modes are used. Typically, the upper ocean is sampled in continuous mode while the float is descending most rapidly, and then switched to discrete mode at slower descent speeds to conserve energy. A typical profile could include continuous mode from 0 to 2000 dbar, then 10-dbar discrete samples except 5-dbar samples in the bottom 200 dbar. The CTD energy expenditure in that case is about 5.2 kJ per 6000-dbar profile.

c. Data communications and management

Data communication by Deep SOLO utilizes Iridium short-burst data (SBD) messaging. SBD has proven reliable on SOLO-II and Deep SOLO in all sea conditions and latitudes including in the Southern Ocean. SBD provides data via gateways to the user in the form of emails, thus obviating the need to maintain a dedicated “always on” server. When Deep SOLO surfaces, it pumps for an additional ~140 s to increase its freeboard for data transmission. It obtains a GPS fix and then transmits about 10 data messages of length up to 340 bytes each, including profiles, trajectory data, and engineering and technical data. Depending on sea conditions and satellite coverage, the data transfer takes 100–200 s. Queued messages from shore, for example containing new mission parameters, are transmitted to the float as part of the packet exchange. Just before the float leaves the sea surface a second GPS position is obtained and stored for transmission on the subsequent surfacing. Total time on the sea surface is typically less than 10 min. The SBD messages from Deep SOLOs follow the same processing pathway as for Core Argo floats. At present they are decoded at Scripps into ASCII files that are forwarded to the U.S. Argo Data Assembly Center (DAC). There, Argo’s real-time quality control procedures are applied. The data, including quality flags, are formatted as Argo netCDF files and forwarded to the Argo Global DACs to be made immediately available to users.

Procedures for delayed-mode quality control of Deep Argo data are still under development, but will be similar to Core Argo. In the examples of Deep SOLO SBE-61 data shown in section 5 of this paper, the only adjustment to the data is for correction of the conductivity cell compressibility coefficient. The SBE-61 applies a default value (cpcor = −9.57 × 10−8 dbar−1) that was inherited from the shipboard SBE-4, which uses a different cell encapsulant. The default value has been found to produce a pressure-dependent fresh bias, of about 0.004 PSS78 at 5000 dbar, relative to nearby recent shipboard hydrographic data. An interim corrected value of cpcor = −11.66 × 10−8 dbar−1 (Murphy and Martini 2018) has been applied in postprocessing to remove the fresh bias.

d. Float recovery

Although not part of routine operation, recovery of Deep SOLO floats is possible and has been carried out on numerous occasions already (Table 1), mostly for diagnosis and correction of defects. As noted above, the prototype floats (SN 6002 and 6003) deployed in 2013 and 2014 were recovered to inspect the glass spheres and the hydraulic system components for wear. In the first “production run” of 7 Deep SOLOs (6004, 6005, 6006, 6007, 6009, 6010, 6011) deployed in January 2016, an impedance mismatch in the float-to-CTD connection resulted in occasional serial communication errors. Symptoms included truncation of some salinity values and some “long vacations” during which floats remained at their parking depth for about 5 months rather than 15 days. A firmware solution for the miscommunication problem was developed and five of the seven floats were recovered in June 2016, with the other two being on “long vacation” at the time of R/V Kaharoa’s recovery voyage. Of the five recovered floats, one (6005) had its antenna damaged and was returned to Scripps. The other four were successfully reprogrammed onboard Kaharoa and immediately redeployed. Two more recoveries were carried out by Kaharoa in October 2018. Floats 6008 and 6015 had rapidly drifting (uncorrectable) salinity measurements. Following recovery, these CTDs were removed from the floats and returned to Sea-Bird Scientific for diagnosis. New CTDs were installed onboard Kaharoa and the floats were redeployed in a matter of minutes. Float 6028, whose CTD had failed shortly after deployment in March 2017, was successfully recovered in the western North Atlantic by Royal Research Ship (RRS) James Cook in November 2018. Finally, float 6053 and 6018, both with possibly failed CTD cables, were recovered in February of 2019 by the R/V Tangaroa in the Southern Ocean and by a charter vessel belonging to Albany Seafoods, Ltd., near Albany, Australia, respectively.

The remarkable record of Deep SOLO recoveries has demonstrated important attributes of the instrument, its operation, and its handling. First, GPS messaging sent every few minutes to the recovery vessel enables floats to be located quickly and reliably. A strip of reflective tape, for visibility and to indicate the orientation of the instrument, is placed on the hard hat opposite to the CTD. Second, a plastic-encased synthetic rope halter on Deep SOLO (not present in Fig. 2) can be snared with a pole and hook from the deck of a research vessel in moderate 2–3-m swell and windy conditions. Third, new float controller firmware can be installed at sea through the glass housing in a matter of minutes. Fourth, the SBE-61 CTD can be replaced quickly on deck by crew members provided with brief instruction. Fifth, Deep SOLO’s fragile antenna is vulnerable during recovery, although with practice, antenna damage is preventable. A hardened antenna is being tested. Finally, a recurrent problem with the CTD serial communications cable or connectors has been detected, and a fix is being implemented collaboratively by Sea-Bird Scientific and SBE-61 users. Looking forward, it is important for Deep Argo to have the capability to recover a fraction of floats for CTD recalibration, for diagnosis of platform and sensor faults, and potentially for recycling of the expensive instruments. The capability to recover Deep SOLO has been demonstrated.

e. Energy budget

As with Core Argo, the viability of the Deep Argo Program depends on deploying floats with sufficiently long life expectancy to permit periodic reseeding of the array. Average float lifetimes in Core Argo of 4 years or longer have proven sufficient to sustain the array. The average float lifetime depends on both battery lifetime and on the probability of premature failure in the float’s hydraulics, CTD, and other systems. Deep Argo is especially challenging since the energy requirement for buoyancy adjustment is much greater than in Core Argo, and because the high pressure operating environment can increase wear in pump components or cause leaks and cable or connector failures.

The total pumping energy for a single Deep SOLO cycle to 6000 dbar is 21.1 kJ, compared with 4.2 kJ for a 2000-dbar cycle of a Core Argo SOLO-II. After parking, the pump is run intermittently to maintain steady ascent speeds that may be set to vary in different depth ranges. If the SBE-61 CTD is run in continuous profile mode from the sea surface down to 2000 dbar, and then in discrete mode from 2000 dbar to the ocean bottom at 10-dbar intervals, the SBE-61 energy consumption for a 6000-dbar profile is 5.2 kJ. Iridium communication requires 0.6 kJ to transmit 6.5 kB of data as short-burst messages. Thus, for the pump battery packs the total energy consumption is 26.9 kJ per 6000-dbar cycle, compared with 8.1 kJ for a SOLO-II 2000-dbar cycle. In addition, if data latency for the descending profile is an issue, an ascending profile for near-real-time transmission can also be obtained for any part of the water column above the parking depth. For example, if the float collects a continuous Core Argo (2000–0 dbar) profile on ascent the added energy cost for the SBE-61 CTD is 3.5 kJ.

Early versions of Deep SOLO held 4 Electrochem primary lithium battery packs consisting of 4 DD cells in each pack. However, the low duty cycle (every 10–15 days) and intermittent high pump current (4 A at 6000 dbar) results in battery passivation, with the lithium cells only delivering half of their stored energy. In Deep SOLO deployments beginning in January 2018, the previous batteries have been replaced by Tadiran (TL-6930) hybrid lithium batteries. These batteries use primary cells to continuously trickle charge rechargeable cells that power the pump with less energy loss. The efficiency is improved and about 75% of the stored energy is expected to be utilized. Moreover, rearrangement of components inside Deep SOLO created enough space for a fifth battery pack. With these battery modifications the useful energy in a Deep SOLO increased to 6.75 MJ (5 hybrid battery packs at 1.8 MJ × 0.75), and the battery life of Deep SOLO increased to 200–250 cycles to 6000 dbar, or 5–7 years of 10-day cycling. This increase in battery lifetime is a critical step toward viability for Deep Argo. Some uncertainty remains in the estimated 75% battery efficiency until many multiyear records are obtained for confirmation.

5. Discussion: Deep SOLO is designed for Deep Argo missions

The yardstick for success of the Deep SOLO is in the extent to which it contributes to implementing and sustaining the Deep Argo Program, and to fully realizing Argo’s scientific value. Here several examples are provided to illustrate ways in which Deep SOLO enables and contributes to the value of Argo, beginning with the mundane but critical issue of float lifetime.

For implementing and sustaining Deep Argo a primary factor must be float lifetime because float and sensor lifetimes define the array’s refresh time. By designing Deep SOLO from the SOLO-II architecture, many of the failure modes of a wholly new float model were obviated through the SOLO-II experience. The evidence for this is in the high reliability of recent SOLO-II deployments together with a similar success rate for the early Deep SOLOs. Following the prototype Deep SOLOs, an additional 62 have been deployed (Table 1) since early 2016. Of these, 55 are active as of April 2019. Of the other seven, the batteries were expended in one, and two may have been snagged while dragging on the bottom. The other four had failures in the CTD cable1 or the CTD, and all four of these have been successfully recovered and returned for diagnosis. Discounting the CTD-related failures, the Deep SOLO platform has had a remarkable success rate for its early life. At the same time, by utilizing hybrid lithium battery technology, the battery lifetime has increased to more than 5 years of 10-day cycling. The proof of long life will come in long records, and must include the CTD as well as the float platform. It is essential for all Argo floats to achieve long missions, and this is a high priority for Deep SOLO.

For supporting Deep Argo’s scientific objectives, it is necessary to obtain profile data that extends to the sea floor. High-latitude water mass formation is among the most difficult oceanic processes to observe directly, due to the inhospitable conditions under which it occurs, the high level of intermittency of the process, and the small horizontal scales of deep convective plumes. Argo cannot resolve individual events on these scales, but it can sample the outcome, that is the properties of newly formed waters as they accumulate and spread along the deep ocean bottom, sometimes in thin layers, moving away from formation regions.

Deep SOLO has already demonstrated the ability to advance our scientific understanding of the ocean below 2000 m. With its passive bottom detection system, the float profiles to within 3 m of the ocean bottom on most cycles. Figure 6 shows the potential temperature θ and salinity records along the track of Deep SOLO 6013 (WMO ID 5902457). This float was deployed at 22°S in the southwest Pacific basin. It drifted northward with an initial drift pressure of 5000 dbar. The drift pressure was gradually shoaled as bottom depth decreased. This float followed a western branch of the Samoan Passage northward into the central Pacific basin. Along its northward track, the erosion of the North Atlantic Deep Water salinity maximum is seen at 4000–4500 dbar. In the bottom layer, the potential temperature of Antarctic Bottom Water increases toward the north, and the figure also indicates that the float sampled close to the bottom in most cycles (red plus symbols).

Fig. 6.

(top) Potential temperature and (middle) salinity along (bottom) the path of Deep SOLO 6013 (WMO ID 5902457) from the green star to the red star. Coldest potential temperature contours are 0.63° and 0.64°C. Red plus and black cross symbols are the deepest depths reached by the float and the published bathymetry, respectively.

Fig. 6.

(top) Potential temperature and (middle) salinity along (bottom) the path of Deep SOLO 6013 (WMO ID 5902457) from the green star to the red star. Coldest potential temperature contours are 0.63° and 0.64°C. Red plus and black cross symbols are the deepest depths reached by the float and the published bathymetry, respectively.

A second example shows that Deep SOLO’s coverage of the southwest Pacific basin, although still short of the Deep Argo target of a 5° × 5° array, is nonetheless adequate for mapping large-scale abyssal property distributions. The map of potential temperature on the 4500-dbar surface (Fig. 7) is based on all Deep SOLO profiles in the southwest Pacific basin from 2016 to July 2019. The map captures the large-scale pattern of abyssal potential temperature, increasing northward and eastward (Talley 2007), and indicative of the interior basin geostrophic shear in this layer (e.g., Roemmich et al. 1996). Smaller-scale features such as the deep western boundary current along the Tonga–Kermedec Ridge are not represented in Fig. 7, but will begin to emerge as more data are obtained in the region. The Deep SOLO dataset complements decadal repeat hydrography transects by revealing the property distributions between the widely separated hydrographic transects and on time scales shorter than decadal.

Fig. 7.

Map of the mean (2016–19) potential temperature (°C) on the 4500-dbar surface in the southwest Pacific basin. Deep SOLO values of θ from each profile are interpolated and smoothed with a 6° Gaussian function. Bottom depths shallower than 4500 m are shaded gray. Profile locations are shown by small black × symbols. The Tonga–Kermedec Ridge is the shallow feature on the left-hand side of the plot, extending northeastward from 37°S, 180°W.

Fig. 7.

Map of the mean (2016–19) potential temperature (°C) on the 4500-dbar surface in the southwest Pacific basin. Deep SOLO values of θ from each profile are interpolated and smoothed with a 6° Gaussian function. Bottom depths shallower than 4500 m are shaded gray. Profile locations are shown by small black × symbols. The Tonga–Kermedec Ridge is the shallow feature on the left-hand side of the plot, extending northeastward from 37°S, 180°W.

A third example illustrates not only Deep SOLO’s capability to detect multidecadal changes in properties, but also the value of year-round profiling. Figure 8 shows the abyssal θS relationship for Deep SOLO 6042 (WMO ID 7900679) profiles that are within 0.5° of repeat hydrography transect SR03. These data are in the Australian–Antarctic Basin near the Antarctic continental shelf. The corresponding SR03 data are also shown for transects since 1993 and stations that are nearly collocated in space with the Deep SOLO float. The SR03 data have not been used to adjust the float salinity. Nevertheless, the float data are consistent with the shipboard summertime hydrographic time series showing the freshening of the abyssal layer by about 0.04 (PSS-78) over 25 years. Moreover, at about θ = −0.4°C, winter Deep SOLO profiles show an increase in salinity that is bottom intensified and very close to the bottom (tens of meters; Fig. 8 inset). Although not a unique interpretation, this is consistent with recent water mass formation and with brine rejection having caused the anomaly. Although this example is of one float during one seasonal cycle, it illustrates the value of systematic year-round profiling in the seasonal ice zone, and of the need to sample close to the ocean bottom.

Fig. 8.

Abyssal θS plot of float 6042 (black; WMO ID 7900679), including the float profiles within 0.5° of an SR03 station, plus SR03 data since 1993 (again, within 0.5° of the float). The profiles are in the Australian–Antarctic Basin, south of Australia near the Antarctic continent. The inset shows the progression of 6042 profiles to winter, with the higher near-bottom salinity being seen in June–August.

Fig. 8.

Abyssal θS plot of float 6042 (black; WMO ID 7900679), including the float profiles within 0.5° of an SR03 station, plus SR03 data since 1993 (again, within 0.5° of the float). The profiles are in the Australian–Antarctic Basin, south of Australia near the Antarctic continent. The inset shows the progression of 6042 profiles to winter, with the higher near-bottom salinity being seen in June–August.

Fourth, as with Deep SOLO 6013, other floats in the southwest Pacific basin array also have a parking depth a few hundred meters above the sea floor, and this is adjusted from shore via Iridium if the bottom depth changes substantially along track. There are two reasons here for departing from the conventional Argo strategy of 1000-m parking depth. First, during Deep Argo’s regional pilot phase, it is desirable to keep floats from drifting out of the regions of deployment. A deep parking depth is expected to have weaker velocities and therefore shorter displacements than at 1000 m. Second, the lack of historical velocity measurements in the abyssal ocean worldwide provides an opportunity for discovery. New and surprising findings will result from observing abyssal ocean float trajectories. The trajectory of Deep SOLO 6004 (WMO ID 5902441; Fig. 9) is an intriguing example. After an initial period of slow wandering, it encountered a minor oceanic ridge at about 35°S, 164°W, where it was entrained in a swift, and likely narrow, flow along the ridge reaching 8 cm s−1 northward for a 15-day period (cycle 78). In 60 days the float drifted more than 300 km to the north. Finally, it slowed and drifted eastward out of the current. This ridge is far to the east of the more substantial Tonga–Kermedec Ridge and Louisville Ridge, both of which are known to have northward deep boundary currents along their eastern flank (Whitworth et al. 1999). It is possible that the paradigm of broad, sluggish flow spanning the “flat” abyssal plains of the ocean should be modified, and that a more complex collection of boundary flows occurs wherever meridionally oriented long ridges are found, even if only a few hundred meters high. Depending on the strength and number of such flows, they could modify the meridional transport of the abyssal ocean and hence the deep limb of the meridional overturning circulations. When a new class of measurements is made or a new domain sampled, it is important to allow new phenomena to be discovered. The abyssal ocean circulation may hold many surprises.

Fig. 9.

Trajectory of Deep SOLO 6004 (× symbols; WMO ID 5902441), with bottom depth shaded, in the southwest Pacific basin. The float drifts close to a small ridge on the abyssal plain. It accelerates and follows the ridge for over 300 km in 60 days, then slows as it moves off to the east. The Louisville Ridge is seen in the southwest quadrant of the plot.

Fig. 9.

Trajectory of Deep SOLO 6004 (× symbols; WMO ID 5902441), with bottom depth shaded, in the southwest Pacific basin. The float drifts close to a small ridge on the abyssal plain. It accelerates and follows the ridge for over 300 km in 60 days, then slows as it moves off to the east. The Louisville Ridge is seen in the southwest quadrant of the plot.

Finally, coverage of the southwest Pacific basin by Deep SOLO floats has already been shown to be sufficient for observing, with good statistical confidence, the warming trend in abyssal temperature and heat content during the period 2014–18 (Johnson et al. 2019). Previously, deep ocean temperature trends were described on multidecadal time scales from repeat hydrographic transects. The new findings confirm Argo’s value in observing deep ocean variability and trends on shorter time scales as needed for improved estimation of Earth’s energy imbalance and the steric component of sea level rise.

The four Deep SOLO pilot arrays have demonstrated that this new model of Argo float is capable of meeting Argo’s requirements for sampling from the sea surface to the bottom. Already, the Deep SOLO has demonstrated its ability to monitor deep-ocean warming (Johnson et al. 2019) and thus deep thermosteric sea level, two of the key scientific objectives of the Argo Program. In addition, Deep SOLO has already captured a number of scientifically relevant deep ocean processes, including deep water-mass transformation along the bottom limb of the meridional overturning circulation, the seasonal cycle of new deep ventilation off the Antarctic coasts, and abyssal property distributions and patterns of flow. With the commercial production of Deep SOLO, this new model of Argo float is ready to extend Argo’s range from 2000 m to full ocean depth, and to form the backbone for implementation of the global Deep Argo array.

Acknowledgments

Argo data are collected and made freely available by the International Argo Project and the national programs that contribute to it (http://www.argo.ucsd.edu, http://argo.jcommops.org, http://doi.org/10.17882/42182). The Scripps Institution of Oceanography’s role in Argo is supported by the NOAA U.S. Argo Program through NOAA Grant NA15OAR4320071 (CIMEC). Shipboard repeat hydrographic data were collected and made publicly available by the International Global Ship-Based Hydrographic Investigations Program (GO-SHIP; http://www.go-ship.org/, http://cchdo.ucsd.edu) and the national programs that contribute to it. The large contributions made by R/V Kaharoa and R/V Tangaroa, as well as other research vessels (Table 1), in Deep SOLO float deployment and recovery operations are gratefully acknowledged. Steffen Pausch of Nautilus Marine Service greatly assisted our early development by providing expert advice for working with glass spheres.

REFERENCES

REFERENCES
Argo Science Team
,
1998
:
On the design and implementation of Argo—An initial plan for the global array of profiling floats. International CLIVAR Project Office Rep. 21, 32 pp
.
Davis
,
R. E.
,
D. C.
Webb
,
L. A.
Regier
, and
J.
Dufour
,
1992
:
The Autonomous Lagrangian Circulation Explorer (ALACE)
.
J. Atmos. Oceanic Technol.
,
9
,
264
285
, https://doi.org/10.1175/1520-0426(1992)009<0264:TALCE>2.0.CO;2.
Davis
,
R. E.
,
J. T.
Sherman
, and
J.
Dufour
,
2001
:
Profiling ALACEs and other advances in autonomous subsurface floats
.
J. Atmos. Oceanic Technol.
,
18
,
982
993
, https://doi.org/10.1175/1520-0426(2001)018<0982:PAAOAI>2.0.CO;2.
Johnson
,
G. C.
,
J. M.
Lyman
, and
S. G.
Purkey
,
2015
:
Informing Deep Argo array design using Argo and full-depth hydrographic section data
.
J. Atmos. Oceanic Technol.
,
32
,
2187
2198
, https://doi.org/10.1175/JTECH-D-15-0139.1.
Johnson
,
G. C.
,
S. G.
Purkey
,
N. V.
Zilberman
, and
D.
Roemmich
,
2019
:
Deep Argo quantifies bottom water warming rates in the southwest Pacific basin
.
Geophys. Res. Lett.
,
46
,
2662
2669
, https://doi.org/10.1029/2018GL081685.
Klatt
,
O.
,
O.
Boebel
, and
E.
Fahrbach
,
2007
:
A profiling float’s sense of ice
.
J. Atmos. Oceanic Technol.
,
24
,
1301
1308
, https://doi.org/10.1175/JTECH2026.1.
Le Reste
,
S.
,
V.
Dutreuil
,
X.
André
,
V.
Thierry
,
C.
Renaut
,
P.
Le Traon
, and
G.
Maze
,
2016
:
“Deep-Arvor”: A new profiling float to extend the Argo observations down to 4000-m depth
.
J. Atmos. Oceanic Technol.
,
33
,
1039
1055
, https://doi.org/10.1175/JTECH-D-15-0214.1.
Murphy
,
D.
, and
K.
Martini
,
2018
:
Determination of conductivity cell compressibility for Argo Program CTDs and MicroCATs. 2018 Ocean Sciences Meeting, Portland, OR, Amer. Geophys. Union, IS24E-2622
.
Purkey
,
S. G.
,
G. C.
Johnson
,
L. D.
Talley
,
B. M.
Sloyan
,
S. E.
Wijffels
,
W.
Smethie
,
S.
Meking
, and
K.
Katsumata
,
2019
:
Unabated bottom water warming and freshening in the South Pacific Ocean
.
J. Geophys. Res. Oceans
,
124
,
1778
1794
, https://doi.org/10.1029/2018JC014775.
Riser
,
S. C.
, and Coauthors
,
2016
:
Fifteen years of ocean observations with the global Argo array
.
Nat. Climate Change
,
6
,
145
153
, https://doi.org/10.1038/nclimate2872.
Roemmich
,
D.
,
S.
Hautala
, and
D.
Rudnick
,
1996
:
Northward abyssal transport through the Samoan passage and adjacent regions
.
J. Geophys. Res.
,
101
,
14 039
14 055
, https://doi.org/10.1029/96JC00797.
Roemmich
,
D.
, and Coauthors
,
2009
:
The Argo Program: Observing the global oceans with profiling floats
.
Oceanography
,
22
(
2
),
24
33
, https://doi.org/10.5670/oceanog.2009.36.
Smith
,
N. R.
, and Coauthors
,
1999
:
The role of XBT sampling in the ocean thermal network
.
Ocean Observing System for Climate
,
Saint Raphaël, France, CNES
, http://www.oceanobs09.net/work/oo99/docs/Smith.pdf.
Smith
,
W. H. F.
, and
D. T.
Sandwell
,
1997
:
Global seafloor topography from satellite altimetry and ship depth soundings
.
Science
,
277
,
1957
1962
, https://doi.org/10.1126/science.277.5334.1956.
Talley
,
L. D.
,
2007
:
Hydrographic Atlas of the World Ocean Circulation Experiment (WOCE). Volume 2: Pacific Ocean.
M.
Sparrow
,
P.
Chapman
, and
J.
Gould
, Eds.,
WOCE International Project Office, Southampton, United Kingdom
, https://doi.org/10.21976/C6WC77.
Talley
,
L. D.
,
2013
:
Closure of the global overturning circulation through the Indian, Pacific, and Southern Oceans: Schematics and transports
.
Oceanography
,
26
(
1
),
80
97
, https://doi.org/10.5670/oceanog.2013.07.
Talley
,
L. D.
, and Coauthors
,
2016
:
Changes in ocean heat, carbon content, and ventilation: A review of the first decade of GO-SHIP global repeat hydrography
.
Annu. Rev. Mar. Sci.
,
8
,
185
215
, https://doi.org/10.1146/annurev-marine-052915-100829.
Whitworth
,
T.
, III
,
B. A.
Warren
,
W. D.
Nowlin
Jr.
,
S. B.
Rutz
,
R. D.
Pillsbury
, and
M. I.
Moore
,
1999
:
On the deep western-boundary current in the southwest Pacific basin
.
Prog. Oceanogr.
,
43
,
1
54
, https://doi.org/10.1016/S0079-6611(99)00005-1.
Zilberman
,
N. V.
,
2017
:
Deep Argo: Sampling the total ocean volume [in “State of the Climate in 2016”]
.
Bull. Amer. Meteor. Soc.
,
98
(
8
),
S73
S74
, https://doi.org/10.1175/2017BAMSStateoftheClimate.1.

Footnotes

a

Current affiliation: MRV Systems, San Diego, California.

© 2019 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).