Isopycnal Float Studies of the Subpolar Front:
Preliminary Results

(Presented at the WOCE Conference in Halifax, Nova Scotia, Canada, 25-29 May 1998)

Tom Rossby, Mark Prater, Huai-Min Zhang,
Peter Lazarevich, and Paula Pérez-Brunius

Acknowledging the support of
Sandra Anderson-Fontana and Jim Fontaine

Graduate School of Oceanography
University of Rhode Island
Narragansett, RI 02882

Supported by NSF OCE-9531878, ONR/NSF N0001492J1651, and CONACYT

Introduction. As part of the Atlantic Climate Change Experiment (ACCE) of WOCE, we are deploying isopycnal floats in the Subpolar Front (SPF) just west of the Charlie Gibbs Fracture Zone (CGFZ) to study pathways of the mean flow and processes of cross-frontal exchange. This region is of particular interest. The Subpolar Front has traditionally been characterized as a zonal flow across the Mid-Atlantic Ridge (MAR), then branching to feed the thermohaline circulation to the north and the wind-driven circulation to the south. Yet the distinct pattern of temperature, salinity, and oxygen on isopycnal surfaces throughout the region indicates that some waters retroflect to the northwest just east of the CGFZ and flow along the Reykjanes Ridge toward Iceland. Our program is designed to address the nature and degree of the organized flow over the ridge and the subsequent splitting of waters. We are using unique, oxygen-measuring isopycnal RAFOS floats ballasted for the 27.5 sq surface.

The RAFOS Float. Our RAFOS floats are neutrally-buoyant, acoustically-tracked, sub-surface drifters which measure position, temperature, and pressure as they flow along with the ocean currents. In addition, these floats are the first ever to measure dissolved oxygen concentration. The float is made isopycnal by the addition of a "compressee" which allows the float to match the compressibility of seawater.

click for a larger image
(A) Argos Radio Antenna; (B) Argos Radio Transmitter;
(C) Main CPU, Memory, and Electronics; (D) Batteries;
(E) Endplate; (F) Hydrophone; (G) Oxygen Sensor; (H) Compressee.

fig1.jpg - 49147 Bytes Regional bathymetry of the study area from the GTOPO30 global digital elevation model (DEM) dataset (Smith and Sandwell). The white vertical line represents the CTD and float deployment line from the November 1997 R/V Håkon Mosby cruise. Also indicated ('*') are all sound sources located in the study area; R=University of Rhode Island, IM=Institut für Meereskunde, AR=Service Hydrographique et Océanographique de la Marine (SHOM). Source R7 failed a few days after deployment in June 1997; source AR7 was recovered in September 1997. Also shown: IRB=Irminger Basin, RR=Reykjanes Ridge, RP=Rockall Plateau, ICB=Iceland Basin, NB=Newfoundland Basin, FC=Flemish Cap, MAR=Mid-Atlantic Ridge.

fig2.jpg - 53823 Bytes The mean SST in March (Fig. 2), when the mixed layer is at its deepest, shows how the warm waters from the subtropics spread toward the northeast by the North Atlantic Current (NAC) and the SPF. The SPF shows up clearly as a zonal front at about 52 oN and crosses the MAR near the CGFZ. Notice how successive isotherms peel off to the north, going from west to east across the MAR. East of the CGFZ the isotherms fan out more broadly, suggesting a more diffuse flow of warm subtropical waters to ward the northeast.

Mean march SST data are from 1987-1990 NOAA/NASA AVHRR Pathfinder, courtesy of Paulo Polito and Ken Casey.

fig3a.jpg - 53514 Bytes The mean hydrography at 27.5 sq shows a similar fanning of water properties east of the CGFZ, with a sharp turn to the west-northwest of the isotherms and isohalines north of 54 oN, suggesting an intrusion of warm and salty waters into the Irminger Basin (Fig. 3a,b). This results in a nearly congruent retroflection of the streamfunction contours toward the MAR (Fig. 3e), with a flow aligned with the ridge toward Iceland (the Irminger Current) and a weaker flow in the Iceland Basin toward the Nordic Seas. In striking contrast, the oxygen field does not show the intrusion of high oxygen waters into the eastern basin nor the penetration of low oxygen waters into the Irminger Basin. This disparity raises the questions of possible diapycnal fluxes and of the relative role of advection and diffusion in the area. Using the oxygen and temperature measurements of the floats will help us address these questions in better detail.

fig3b.jpg - 51405 Bytes fig3c.jpg - 50218 Bytes
fig3d.jpg - 54043 Bytes fig3e.jpg - 52136 Bytes

Mean hydrography is derived from forty years of bottle data. Source: World Ocean Atlas 1994, NODC. Nominal pressure of sq = 27.8 is 2000 dbar. Black areas represent the areas where the 27.5 sq (Fig. 3a-d) and 27.8 sq (Fig. 3e) do not exist.

Hydrographic section across the SPF along 37 oW, Nov.2 - Nov.4, 1997. Thirteen CTD stations and water samples were taken at each of the RAFOS float deployment sites between 49 oN and 53 oN. The sq vs. P plot (Fig. 4a) clearly shows the major dynamic SPF at 51.5 oN and a weaker branch of the SPF south of 49.5 oN. The shear velocity (50 m relative to 1150 m) computed from the hydrography has a maximum of 45 cm/s at 51.5 oN and 25 cm/s at 49.5 oN. However, the baroclinic transports (relative to 1150 m) are similar: 8.5 Sv for the major SPF branch and 8.3 Sv for the southern branch. The property vs. sq plots (Fig. 4b-d) show that two water masses are separated by the major SPF: relatively warm, saline, oxygen-poor subtropical waters in the south and cooler, fresher, oxygen-rich subpolar waters in the north.
fig4.jpg - 100954 Bytes

fig5.jpg - 60228 Bytes
The water mass distinction is also clearly shown in the T vs. S and T vs. O2 plots (Fig. 5). The S vs. sq plot also shows a core of fresh (and colder) water core at 51.8 oN, indicating eastward intrusion of Labrador Sea water and supporting the idea of possible retroflection of the SPF east of the MAR as revealed by the SST (Fig. 2) and historical hydrography (Fig. 3).

Selected NAC Experiment Float Trajectories (Fig. 6a). We launched 100 isopycnal RAFOS floats as part of the NAC study during 3 deployment cruises from 1993 to 1994. Most of these floats had data records of 8 to 10 months and were deployed on two density levels: 27.2 and 27.5 sq. Trajectories from several 10-month floats that drifted east along the SPF while on the 27.5 sq surface are shown. Two floats (between 100 and 200 m depth) drifted north along the ridge to 62 oN, and two continued into the eastern basin south of Rockall Bank, consistent with the fanning out of the temperature fields as seen in Figure 3a.Preliminary ACCE Float Trajectories (Fig 6b). Prototype float pair 432 & 434 and pair 431 & 433 were launched in the Spring of 1997 during the sound source deployment cruise on the R/V Knorr. The floats stayed together as pairs for 30 days, separating only when one float from each pair surfaced. The other trajectories shown are from four short-mission floats among the 37 floats launched in the Fall of 1997 from the R/V Håkon Mosby. Note that floats 479 and 480 eventually ended up in the same eddy at 51 oN. The details of these two floats can be seen below.
fig6.jpg - 93437 Bytes

Closer Examination of Two ACCE Floats. We highlight the data from two floats, 479 (Fig. 7a) and 480 (Fig. 7b), that were deployed for 5 month missions along 37.5 oW in November 1997. The two floats were deployed about 190 km apart (Fig. 6b). The southern float (480) was initially deeper yet warmer than the northern float (479), but both moved in parallel to the east-southeast at 8 to 10 cm/s. When float 480 approached the ridge, it turned to the north and moved toward a point where float 479 had been 20 days earlier. From that point both floats made a looping meander (albeit 20 days apart). This meandering is also evident in the pressure records, whereby the floats are deeper and warmer in the trough (subtropical side) and shallower and cooler in the crest (subpolar side). Both floats were then caught in the same cyclonic eddy at 30.5 oW and 51 oN. Float 480 experienced a drop in T of 0.5 oC and an increase in O2 of 0.3 ml/l when it entered the eddy, matching existing values of float 479. The floats remained in this stationary eddy for 3 months, at times having orbital speeds up to 20 cm/s, before surfacing on schedule.
fig7.jpg - 98447 Bytes

Eddy Kinetic Energy. The eddy kinetic energies (EKE) computed for the NB and the Subpolar Gyre show strong spatial gradients. At the surface, analysis of alongtrack TOPEX/Poseidon altimetry data (Fig. 8a) shows a core in the NAC off Newfoundland of 1600 cm2 /s2 . To the south of the SPF, surface values range from 300 to 600 cm2 /s2 , while north of the SPF values rarely exceed 100 cm2 /s2 , except along the western sides of the Reykjanes Ridge and the Rockall Plateau. Using RAFOS float data obtained in the NAC experiment (Fig. 8b), the EKE on the 27.5 sq surface peaks near 800 cm2 /s2 off Newfoundland. These peaks occur where float trajectories indicate topographically fixed cyclonic meanders which grow and decay but do not propagate. Values drop to below 100 cm22 /s2 in the interior of the NB.
fig8.jpg - 139535 Bytes

fig9.jpg - 62547 Bytes Mean Velocity. Figure 9 shows the Eulerian mean velocity computed from the NAC floats by binning the data into 1/2o non-overlapping boxes. The 100 float trajectories cover the NAC and NB regions well, making the statistics there meaningful. The trough at 47 oN is locked east of FC and grows and decays with time. The turning of the NAC at the Northwest Corner is also a "permanent" feature. After turning east, the NAC (or the SPF) is more wave-like, passing the MAR over the CGFZ. At least two ACCE floats released at 37 oW follow the above flow pattern. The velocity along the RR represents one float at each surface.

fig10.jpg - 51275 Bytes Lagrangian Properties, computed from the NAC floats in the SPF region, are shown in Figure 10. In computing the Lagrangian diffusivity, Taylor theory assumes a homogeneous background turbulence. The float motion is a combination of background advection (mean velocity) and turbulence. In regions of "strong" lateral mean velocity shear, resolving and subtracting the space-dependent mean velocity are very important for the statistical analyses. In regions where space-dependent mean velocity cannot be resolved, the "trend" is not removed thus biasing the statistics. In Figure 10 the velocity variances (< u'2 >, < v' 2 >) (a), Lagrangian integral time scale (b) and space scale (c), as well as isopycnal diffusivity (d), are shown as functions of the bin size used in computing the mean velocity. Larger bin sizes (e.g., 2o ) cannot resolve the current structures shown in Figure 9 (with 1/2o bin size), and bias the variance, integral time and space scales, and diffusivity to larger values. Computations for other regions show similar values and patterns.

fig11.jpg - 110628 Bytes The Oxygen Sensor. To better aid our study of water-mass exchange along the SPF, we have added an oxygen sensor to the RAFOS float. Using a pulsing technique, developed by Dr. Chris Langdon (LDEO), it is possible to operate a standard membrane-type sensor (as commonly used with CTD casts) over the lifetime of the floats. Along with increased longevity, another advantage of this technique is that it virtually eliminates the flow-rate dependence. Figure 11 is a top-view of the sensor showing the cathode (gold ring) and the anode (silver/silver-chloride triangular post). fig12.jpg - 29137 Bytes The oxygen measurement consists of 20 one-second pulses with a 4 minute interval between each pulse. Dissolved oxygen reduces at the cathode, resulting in a current which the float measures at the end of each pulse. The float stores in memory the median value of the last five measurements. (The first 15 measurements are taken to allow the sensor to reach a working equilibrium).
Of the floats that have already surfaced, 6 out of 8 indicated significantly high initial oxygen concentrations that dropped off nearly exponentially; some indicated initial concentrations that were well above saturation. A possible mechanism for this trend was that air was trapped between the sensor body and its protective sleeve when the float was deployed. As the floats descended from the surface, this trapped air space was flooded with water which then became hypersaturated with oxygen. If not flushed out readily, this volume of water would have produced a high oxygen reading. An exponential fitting was used to remove the above effect for each float, and Figure 12 shows the effectiveness of the fitting for float 480.

Summary. We have completed a preliminary high-resolution analysis of historical hydrographic data in the Subpolar Front region, and have examined our Lagrangian RAFOS float data from the previous North Atlantic Current Study and the present Atlantic Climate Change Experiment in this context. Evidence from AVHRR SST images, historical hydrographic data, and RAFOS trajectories all support the conjecture that at least part of the Subpolar Front retroflects in a broad loop to the northwest just to the east of the Mid-Atlantic Ridge. We have started making oxygen measurements from isopycnal Lagrangian floats. This new application is showing much promise. We will use these data to verify the above hydrographic picture (particularly the noncongruence of T and O2 north of 54 oN) and to examine the roles of advection and eddy diffusive processes in maintaining the distributions of T, S, and O2 in the retroflection region and beyond. We hope this work will contribute to an improved understanding of the factors (local and remote) that determine the distribution of warm water flow along the Reykjanes Ridge and along the eastern boundary toward the Nordic Seas. Fifty more RAFOS floats will be launched this summer from the R/V Knorr during our final deployment cruise. All will be measuring T, S, and O2 on the 27.5 sq surface, and most will complete their missions by early 2000.