Abstract

 

                  Thirty-two RAFOS floats were launched at the depth of intermediate water, near 750 m, in the Benguela Current along 30S and its extension along 7W.  The floats were tracked acoustically for two years during 1997­–1999.  Seven floats looped in three Agulhas Current rings, which drifted west northwestward at a mean velocity of around 5 cm/sec.  Floats not in Agulhas rings tended to drift westward at around 2 cm/sec in the latitude band 22S–35S.  North of 22S three floats drifted eastward.  This report describes the float trajectories and summarizes the main results.  These are the first subsurface long-term Lagrangian data in the Benguela Current.

 

 

 Introduction

 

            The overall objective of the Benguela Current Experiment is to measure the northward flow of intermediate water in the eastern South Atlantic.  This water comprises a large part of the upper layer of the thermohaline conveyor belt or meridional overturning circulation in the Atlantic.  Upper layer water flows northward across the equator into the northern North Atlantic where the water is cooled and transformed into deep water, which returns southward as a deep western boundary current.  The goal of this experiment is to obtain the first long-term Lagrangian measurements in the Benguela Current, which is the origin of the northward-flowing intermediate water in the South Atlantic and in the Benguela Current extension, which is the main conduit of intermediate water westward across the South Atlantic (Figure 1).

 

            The Benguela Current Experiment is one component of the larger Cape of Good Hope Experiments, KAPEX (Boebel et al., 1998).  The overall objective is to study the interocean exchange of subsurface waters between the Atlantic and Indian Oceans south of Africa.  Other components of KAPEX launched floats in the Agulhas Current and in the South Atlantic Current south of our floats.  The new float trajectories in these two regions have recently been described by Boebel et al. (2000).  Some of the Benguela Current floats drifted over the mid-Atlantic Ridge into the western South Atlantic and into the region of the WOCE Deep Basin Experiment (DBE) where other acoustic floats have been tracked (Boebel et al., 1999) and are being tracked (Ollitrault, 1999) in the intermediate water.  The overall goal of these experiments is to develop a circulation scheme for intermediate water in the South Atlantic that is based on direct measurements of the velocity field.  Results from KAPEX have been published in a special issue of Deep-Sea Research II which includes results from the Benguela Current Experiment (Richardson and Garzoli, 2003).

 

Deployment Cruise

 

            Thirty RAFOS floats, two ALFOS floats, and two moored sound sources were launched from the R/V Seward Johnson during a cruise from Cape Town to Recife, September 4–30, 1997 (Figure 2).  In addition to the float work, 44 CTDO-LADCP stations to 2000 m were obtained, XBT surveys mapped three Agulhas rings, and seven surface drifters were deployed in the rings (Roubicek et al., 1998; Garzoli et al., 1999).  The rings were previously identified by satellite altimetry and tracked by altimetry (Garzoli et al., 1999) back to their formation near the Agulhas Current retroflection located near 40S 18E.

            The floats were launched along two lines, one roughly along 30S that cuts across the Benguela Current and the other along 7W that cuts across the Benguela Current extension.  Seven floats were launched in the three Agulhas rings that were located near the cruise track (see Garzoli et al., 1999).

 

 

 

 

Figure 1:  Schematic circulation diagram of currents at the intermediate water level (~750 m) in the vicinity of South Africa showing the general location of the Benguela Current (after Boebel et al., 1998).  The new float data described in this report reveal that intermediate water in the Benguela Current flows westward across the mid-Atlantic Ridge between 22S–35S versus flowing northwestward as indicated in this figure.  Two anticyclonic Agulhas Current rings, which form near 40S 18E and drift northwestward are also indicated.  Depth contours are 1000 m and 3000 m.

 

 

 

 

 

 

 

Figure 2:  Location of KAPEX cruise tracklines, RAFOS float deployments (dots) and sound sources (large circles, where  R  is for Rhode Island,  K  for Kiel, and  M  for WHOI) (see Boebel et al., 1998 and Boebel et al., 2000).  The Benguela Current experiment launched floats along 30S (nominal) and 7W on R/V Seward Johnson cruise SJ9705 in September 1997 from Cape Town to Recife.

 

Floats

 

            The RAFOS floats (see Rossby et al., 1986) were purchased from Seascan Corporation in Falmouth, Mass., and assembled, calibrated (temperature, pressure) and ballasted at WHOI (Tables 1, 2).  The floats recorded temperature, pressure, and times of arrival (TOAs) from moored sound sources.  At the end of their missions the floats dropped weights, rose to the surface and transmitted data to WHOI via the Service Argos satellite system.

            The floats are quasi-isobaric and were ballasted for a depth of 750 m which lies near the center of the intermediate water layer in the Benguela Current region.  The initial float depths ranged from 660 db to 800 db and the mean initial depth was 737 db (Table 1).  Ten floats were programmed to record TOAs twice per day during 18-month missions; 20 floats recorded TOAs once per day during 24-month missions.  Two additional ALFOS floats (ALACE–RAFOS) obtained from Webb Research Corporation were used to monitor the sound sources.  These floats returned to the ocean surface for two days at monthly intervals by means of active ballasting similar to an ALACE float (Davis et al., 1992).  At the surface they transmitted acoustic data like the RAFOS floats.

            Twenty-eight (93%) of the 30 RAFOS floats successfully surfaced and transmitted data.  Two floats (375, 408) were never heard and we do not know why they failed.  Float 407 went too deep, dropped its weight and surfaced after two days.  Floats 385 and 392 surfaced before the end of their missions because of low battery voltage.  The amount of data obtained compared to the amount attempted is around 88%.  Overall we obtained 46 float-years of data from the RAFOS floats.  One ALFOS float ceased after 130 days, but the other continued to work successfully to February 2001, 3 1/2 years from launch.

 

Float Clock Corrections

 

            In principal float clock drift can be estimated by comparing the float clock to time recorded by the Argos data system when the float surfaces.  In practice we found residual errors (of unknown source) of around 20 seconds for the 18-month floats (Table 2).  We calculated a further correction to the float clocks by assuming the sound source clocks were correct and by using the distances between the floats and sound sources to estimate when the last TOAs should have been recorded.  The small drift of the source clocks made this possible.

 

Sound Sources

 

            Three sound sources were purchased from Webb Research Corporation and were moored at depths near 800 m (Table 3).  Two sources were launched from the R/V Seward Johnson and the third from the R/V Polarstern.  This help from our German colleagues allowed us time to survey the three Agulhas rings.  One of the three sources (M10) ceased transmitting on January 7, 1999; the other two continued to the end of the experiment and are presently aiding French float tracking in the western basin (M. Ollitrault, personal communication).

            Our three sources were part of an extended KAPEX source array to track floats over a large area around South Africa (Figure 2, Table 3).  Some of the Benguela Current floats drifted across the mid-Atlantic Ridge into the western South Atlantic where tracking was supplemented by the DBE acoustic array (Hogg and Owens, 1999).

            Most sound sources were not retrieved so that their clock drifts could not be accurately measured.  Fortunately the drift rate of the source clocks is small (< 0.01 sec/day) so that source drift is usually not a significant problem.  Source clock corrections (Table 3) were estimated from the surface positions of some floats and their last recorded TOAs (see Boebel et al., 2000).  In addition a time series of source clock corrections was generated using TOAs from the monthly surfacings of the ALFOS float.  Due to a 10 sec ALFOS clock jump (for unknown reasons) and other possible errors only the relative source clock drift rates (one source relative to the others) could be estimated using the ALFOS data.  The various estimates confirm small source clock drift rates.

 

 

Float Tracking

 

            Floats were tracked using Matlab-based tracking software developed by Martin Menzel in Germany and Olaf Boebel at URI (based on URI programs) and modified by Heather Hunt Furey at WHOI.  An average speed of sound of 1484 m/s was estimated from the float launch locations and first TOAs.  This was then used to calculate distances between floats and sound sources using TOAs.  Float positions were calculated by least square triangulation using distance time series from three sources.  Gaps less than 10 days long were linearly interpolated.  In general the time series data and trajectories were good quality and few problems were encountered.  Position errors are estimated to be around 4 km based on this and earlier float experiments (see for example, Richardson and Wooding, 1999).  Position time series were smoothed by means of a Gaussian-shaped filter (σ = 1 day) of weights 0.054, 0.244, 0.404, 0.244, and 0.054 for positions spaced at one-day intervals.  An equivalent filter was applied to positions spaced at one half day (.028, .066, .124, .180, .204, .180, .124, .066, .028).  This reduced random position errors, tidal and inertial fluctuations and gave nicer looking trajectories.  The smoothed time series were subsampled once per day for the 18-month floats.  Finally a cubic spline function was passed through the position series to calculate velocity along the trajectories.

 

Preliminary Results

 

            The long-term float drifts can be seen in the displacement vectors, which connect launch locations and surface positions (Figures 3–6).  All floats in the band 22S–35S went westward.  Most of the floats launched east of the Walvis Ridge along 30S drifted westward across the Walvis Ridge and those launched along 7W drifted westward across the mid-Atlantic Ridge except for three floats north of 22S, which went eastward.  The longest displacement vectors were from the floats launched in the three rings.

            Float trajectories are more complicated than the displacement vectors but several distinctive patterns are apparent (Figures 7–10).  To help see the patterns the trajectories were subdivided into two groups based on looping characteristics.  In the first are looping portions of trajectories which are interpreted to be floats trapped in the swirl velocity of discrete eddies like the three Agulhas rings (R1–3, Figure 11).  In the second are nonlooping portions of trajectories which are located outside of discrete eddies (Figure 12).  The seven trajectories in Agulhas rings R1–R3 showed that they translated quite steadily west northwestward at around 5 cm/sec (Figure 13).  The mean velocity (and standard error) of the three rings calculated by combining the three individual mean ring velocities using the longest looping float in each ring is   - 5.2 ± 0.3 cm/sec,  1.4 ± 0.2 cm/sec  or  5.4 cm/sec toward 286°.  Ring 1 zonal velocity was   - 4.6 cm/sec,  smaller than the other two.  This ring appeared to slow as it passed over the Walvis Ridge which reduced its mean velocity.  West of the Walvis Ridge, Ring 1 translated at a mean zonal velocity of   - 4.9 cm/sec,  closer to the velocity of the other two rings.

            In the band 22S-35S the seven 7W line floats translated westward over the mid-Atlantic Ridge (Figure 9).  The average velocity over two years of these floats (391, 401, 403, 404, 410, 411 and 412) was   2.3 ± 0.2 cm/sec,  ,  where the standard error was calculated from the seven mean velocity values of the individual floats.  Relative to this mean velocity of the background flow at 750 m the Agulhas rings translated westward at around 2.9 cm/sec and northward at around 1.3 cm/sec.  The rings’ northward velocity component advected ring water northward across the generally westward flowing Benguela Current extension.

            North of 22S, floats 393, 405, and 409 drifted eastward in a current that clearly marked the northern boundary of the Benguela Current extension.  All three floats ended up in regions of very low mean velocity, two (393 and 409) near 21S 5W.  In this region the floats oscillated in a northeastward–southwestward direction with ~200 km characteristic displacements and a ~120 day period.  These oscillations could be caused by Rossby waves formed in the Angola–Benguela frontal zone located equatorward of the floats.

            Trajectories in the region between the Walvis Ridge and Africa are more complex than the trajectories farther west (Figures 14, 15).  Note that the subsurface displacements of ALFOS float 101 were included in Figure 14.  A surprising result is that all six of the easternmost floats launched along 30S (excluding floats in Ring 1 which drifted northwestward) drifted various distances southward, counter to the mean direction of the near surface Benguela Current, before crossing over the Walvis Ridge (Figure 16, 17).  Three floats (392, 395, 398) drifted southward then turned and drifted northwestward in the Benguela Current.  During four months of its southward drift float 395 looped in cyclone C3 and during six months of its northwestward drift, float 395 looped in anticyclone R7, possibly an Agulhas ring.  Since the anticyclones are most likely Agulhas rings we label them R4–R7.  Float 392 looped in cyclone C1 on the way south and float 398 looped for four months in anticyclone R5 which was translating northwestward.  Two floats (101, 402) drifted quite far south, 880 km to 38.0°S for float 402 and 750 km to 36.8°S for float 101, but they both turned northward, passed close to their launch locations after 12 months (402) and 17 months (101) and subsequently turned westward.  Float 402 looped for nine months in cyclone C2 that went, generally, westward.  These two floats (101, 402) imply large-amplitude low-frequency fluctuations of meridional velocity east of the Walvis Ridge.  The sixth float (397) initially drifted westward around the northern side of ring R1, then dipped southward 3.2° near the Walvis Ridge and was entrained into anticyclone R5, and then turned northwestward.  Taken together these six floats imply that a large amount of southward flow occurred in the region east of the Walvis Ridge at least during the first several months of the float experiment and possibly longer.  All six floats eventually turned more westward in the Benguela Current and its extension clearly showing that the southward flow feeds into the Benguela Current.

            The floats in the Cape Basin east of Walvis Ridge revealed the general translation of eddies there, northwestward for three anticyclones and one cyclone, southwestward for two cyclones (Figure 15).  Floats in three eddies (R4, R5, C2) that encountered the Walvis Ridge stopped looping, implying a disruption of the normal eddy swirl velocity by the Ridge.  Because of the intense eddy motions and the low-frequency fluctuations in this complicated eastern region, the six floats do not give a very accurate picture of the mean circulation there.  They do show that mesoscale eddies are energetic and strongly advect the floats.

 

Summary

 

            In general the floats and sound sources worked well, and high quality data were obtained.  The floats were successfully tracked throughout a wide region of the South Atlantic using a large array of sources.  The floats were carefully ballasted as seen in their consistent initial depths.  Only one float was either ballasted too deep or leaked which caused it to sink below its safe depth, to drop its weight and to return to the surface.  Two other floats were never heard for unknown reasons.  Overall the success rate was around 90% based on obtaining trajectories from 27 of the 30 RAFOS floats.  A total of 46 float-years of trajectories was obtained.  Interesting preliminary results concern the translation of Agulhas rings and other eddies, the complicated inflow to the Benguela Current, the rather steady westward flow in the Benguela Current extension over the mid-Atlantic Ridge, and an eastward current located at the northern edge of the Benguela extension.

 

 

 

Acknowledgements

 

            Funds for this experiment were provided by National Science Foundation Grants OCE-9528574 and OCE-0236654.  Jim Valdes, Brian Guest and Bob Tavares prepared the floats and sound sources at WHOI.  Paul Bouchard was in charge of the mooring operations on the R/V Seward Johnson.  We thank the Captain, crew and scientific parties for their participation in the work.  We are indebted to our German colleagues who launched one of our sound sources from the Polarstern.  Many KAPEX colleagues helped in the scientific design and provided valuable scientific discussions, especially Olaf Boebel, Chris Duncombe Rae, Dave Fratantoni, Silvia Garzoli, Gustavo Goñi, Johann Lutjeharms, Tom Rossby, Claudia Schmid, and Walter Zenk.  Heather Hunt Furey generously helped our transition to the new float-tracking software.  MaryAnn Lucas typed the manuscript.

 

REFERENCES

Boebel, O., S. Anderson-Fontana, C. Schmid, I. Ansorge, P. Lazarevich, J. Lutjeharms, M. Prater, T. Rossby, and W. Zenk, 2000.  KAPEX RAFOS float data report 1997–1999.  Part A: The Agulhas and South Atlantic current components.  Berichte aus dem Institut für Meereskunde Nr. 318, Institut für Meereskunde an der Christian-Albrechts-Universität, Kiel, Germany, 194 pp.

Boebel, O., R. E. Davis, M. Ollitrault, R. G. Peterson, P. L. Richardson, C. Schmid, and W. Zenk, 1999.  The intermediate depth circulation of the western South Atlantic.  Geophysical Research Letters, 26(21), 3329–3332.

Boebel, O., C. Duncombe Rae, S. Garzoli, J. Lutjeharms, P. Richardson, T. Rossby, C. Schmid, and W. Zenk, 1998.  Float experiment studies interocean exchanges at the tip of Africa.  EOS, Transactions of the American Geophysical Union, 79(1), pages 1, 7–8.

Davis, R. E., D. C. Webb, L. A. Regier, and J. Dufour, 1992.  The Autonomous Lagrangian Circulation Explorer (ALACE).  Journal of Atmospheric and Oceanic Technology, 9, 264–285.

Garzoli, S. L., P. L. Richardson, C. M. Duncombe Rae, D. M. Fratantoni, G. J. Goñi, and A. J. Roubicek, 1999.  Three Agulhas rings observed during the Benguela Current Experiment.  Journal of Geophysical Research, 104(C9), 20,971–20,985.

Hogg, N. G., and W. B. Owens, 1999.  Direct measurement of the deep circulation within the Brazil Basin.  Deep-Sea Research II, 46(1–2), 335–353.

Hunt, H. D., C. M. Wooding, C. L. Chandler, and A. S. Bower, 1998.  A Mediterranean Under-current Seeding Experiment (AMUSE):  Part II:  RAFOS float data report, May 1993–March 1995.  Woods Hole Oceanographic Institution Technical Report, WHOI-98-14, 123pp.

Ollitrault, M., 1999.  MARVOR floats reveal intermediate circulation in the western equatorial and tropical South Atlantic (30°S to 5°N).  International WOCE Newsletter 34, 7–10.

Richardson, P. L., and S. L. Garzoli, 2003.  Characteristics of intermediate water flow in the Benguela Current as measured with RAFOS floats.   Deep-Sea Research II, 50, 87-118.

Richardson, P. L., and C. M. Wooding, 1999.  RAFOS float trajectories in Meddies during the Semaphore Experiment, 1993–1995.  Woods Hole Oceanographic Institution Technical Report WHOI-99-05, 86 pages.

Rossby, T., D. Dorson, and J. Fontaine, 1986.  The RAFOS system.  Journal of Atmospheric and Oceanic Technology, 3, 673–679.

Roubicek, A. J., S. L. Garzoli, P. L. Richardson, C. M. Duncombe Rae, and D. M. Fratantoni, 1998.  Benguela Current Experiment:  R/V Seward Johnson SJ9705, State Department Cruise No. 97-023, Cape Town, September 4–Recife, September 30, 1997.  NOAA Data Report ERL AOML-33, Atlantic Oceanographic and Meteorological Laboratory, Miami, Florida.

 

 

>
 

Appendix A:  Summary Figures Showing Preliminary Results

 

            Figures 3–17 show float displacement vectors and trajectories, subdivided by float deployments in different geographical areas (30S, 7W, and Cape Basin) and into floats looping in eddies and floats outside of eddies.  Observed eddies include counterclockwise rotating Agulhas Current rings and clockwise rotating cyclones.

 

 

 

Figure 3: Float Displacement Vectors

 

Figure 4: 30S Float Displacement Vectors

Figure 5: 7W Float Displacement Vectors

Figure 6: Ring Float Displacement Vectors

Figure 7: Float Trajectories

Figure 8: 30S Float Trajectories

Figure 9: 7W Float Trajectories

Figure 10: Ring Float Trajectories

Figure 11: Floats Looping in Eddies

Figure 12: Floats Not Looping in Eddies

Figure 13: Eddy Displacement Vectors

 

Figure 14: Cape Basin Trajectories

 

 

Figure 15: Cape Basin Floats Looping in Eddies

Figure 16: Southward displacements of the easternmost six floats launched along 30S (excluding floats in ring 1 which translated northwestward).  Displacement vectors are drawn between the float launch locations and the southernmost points of each trajectory. The southernmost displace-ment of ALFOS float 101 is included even though its subsurface drift was interrupted when the float surfaced for 2 days every month.  The accumulated surface displacement of 101 on its way southward is 77 km northwestward, counter to the overall southward displacement of 800 km.

Figure 17:  Displacement vectors between the southernmost points shown in Figure 16 and the surface positions at the end of the float mission (after two years).  Floats 101 and 402 returned northward and passed close to their launch locations before turning more westward, implying low frequency variations in the meridional flow off the eastern boundary.