an der Christian-Albrechts-Universität – Kiel, IfM 318

Table of Contents

List of Figures. 3

List of Tables. 3

Introduction. 4

Description of RAFOS Floats. 4

Float Ballasting. 6

Deployment of Sound Sources. 6

Deployment of Floats. 7

Float Tracking. 9

Acknowledgements. 10

Tables. 12

References and related literature. 19

RAFOS Float Data. 20

List of Figures

Figure 1: Schematic diagram of presumed thermocline and intermediate water flow.. 8

Figure 2: RAFOS float deployment sites and sound source locations. 9

Figure 3: Sound source drift estimates. 10

Figure 4: Summary plot of isobaric (IfM-Kiel) RAFOS float trajectories. 20

Figure 5: Summary plot of isopycnal (URI) RAFOS floats. 20

List of Tables

Table 1: Launch parameters of floats deployed during ANT XIV/2 on RV Polarstern. 12

Table 2: Agulhas float launch parameters, launched from RV KUSWAG V. 13

Table 3: Launch parameters of floats set from aboard RV Fridtjof Nansen. 13

Table 4: Launch parameters of Agulhas float set from RV KUSWAG I 14

Table 5: Technical float parameters and (subjective) comments. 15

Table 6: Sound source parameters used to track KAPEX floats. 18

Introduction

This is the final data report of RAFOS float data collected by the Institut für Meereskunde, Kiel (IfM-Kiel), the University of Rhode Island (URI) and the University of Cape Town (UCT) during the 1997-1999 Cape of Good Hope Experiment (KAPEX) around southern Africa [Boebel et al., 1998a; Lutjeharms et al., 1997]. The objective of the program, funded by the National Science Foundation (USA) and the Bundesministerium für Bildung, Forschung und Technologie (Germany), was to study the interocean exchange of subsurface waters south of Africa between the Atlantic and Indian Oceans. Ninety-two floats were deployed at depths between 100 and 1200 m. The floats recorded arrival times of sound signals from moored sound sources, generally twice a day, together with measurements of the ambient pressure, temperature and in selected cases (URI floats) oxygen. Some IfM-Kiel floats acquired data only once per day to prolong their mission length, while the URI floats took additional pressure and temperature measurements every 6 hours.

Four separate float deployments took place: March/April 1997, December 1997, February 1998 and June 1998. On the first cruise on the RV Polarstern 35 IfM-Kiel floats were deployed in the Cape Basin [Boebel et al., 1998b]. The second cruise made use of SA KUSWAG V to deploy 17 URI floats off Port Elizabeth, South Africa. The third deployment was handled by scientists and crew of RV Dr. Fridtjof Nansen to deploy 4 IfM-Kiel and 6 URI floats off Walvis Bay, Namibia. The final cruise was onboard SA KUSWAG I to deploy 30 URI floats on two legs off Durban, South Africa. CTD casts were taken during all deployments from RV Polarstern and RV Dr. Fridtjof Nansen, while XBT data were collected during the Kuswag cruises. Float missions varied between several weeks and two years. The floats were tracked using a total of 11 sound sources manufactured by the Webb Research Corporation. The sources were deployed during the RV Polarstern cruises (March/April 1997 and March 1999) and two RV Seward Johnson cruises (August and September 1997).

Description of RAFOS Floats

The floats used in the project fall into two general categories: quasi-isobaric and isopycnal floats. Floats deployed by IfM Kiel are of the quasi-isobaric type and based on electronics provided by BathySystems of Rhode Island. The floats were assembled at the IfM Kiel from parts. Floats deployed by URI are of isopycnal character and based on the standard WOCE RAFOS float boards manufactured at URI. Floats were assembled at URI and supplemented with a compressee that provided a system compressibility close to that of seawater.

Float performance in terms of number of instruments returning to the surface and transmitting data versus the number instruments launched was 87% (80 out of 92 floats), which is about 5% below the usual rate. The lower rate is mostly due to the floats assigned to the Namibia deployment (4 lost out of 10), which is characterized by two handicaps; a) it used several reconstructed floats which were assembled from bits and pieces of floats that have either failed earlier rigorous testing or which had been used before and had eventually been recovered by coincidence; b) the floats were launched in fairly shallow water due to limitations of the cruise track.

Two other technical aspects of the IfM Kiel floats stand out. First, floats that failed to transmit the whole data vector (marked XMS in Table 1 and Table 5) carry without exception float numbers <= 200. This could indicate that some of the transmitter batteries used for this batch of floats were defective or past their shelf lifetime, and had less energy available than the batteries used for floats with higher serial numbers. The second aspect concerns the floats' resistance to corrosion and subsequent leakage. Such a problem was experienced by 8 of the 37 floats produced in Kiel. The end plates and release plugs for these floats were of varying materials or had undergone different surface treatments. Endplates were usually made of German V4A grade stainless-steel, but endplate surfaces were electro-polished after the machining process in some cases while not in others . The release plug was either made of German V2A stainless-steel and painted, or from German V4A grade stainless-steel, which was explicitly shipped to the US for the manufacturing of the releases. From all floats that showed rapid sinking due to corrosion (marked PPP in Table 1, and Table 5), most featured an unpolished end-plate (6 floats with unpolished versus 2 floats with polished endplates) in combination with a German V4A release plugs (6 V4A versus 2 V2A). This should not lead to the conclusion that the V4A release plugs trigger corrosion, since we experienced corrosion with German V2A plugs as well in earlier experiments. The more plausible conclusion is that the unpolished (German V4A) end-plate is more prone to corrosion than the electro-polished version.

URI floats featured different materials ( an anodized aluminum end plate for one) which mostly avoided the problems listed above (only one sinking float), but other problems occurred. An unusually large number of floats lost their dropweights (10 out of 53) after approximately 1 month, continuing their mission at the surface. Fish bite at the monofilament connection between compressee and float could be a possible explanation for this behavior, especially in light of the fact that all but one of these floats were ballasted for the shallower isopycnal (26.8 s_q), where fish-bite is more likely.

A substantial number of floats (15 out of 53) surfaced early due to the fact that the floats went to a greater than expected pressure which triggered the emergency release procedure. This behavior does not reflect a malfunctioning of the float (only 1 float URI shows signs of leakage), but the fact that the 27.2 s_q isopycnal, for which all these floats were ballasted, reached greater depths then originally anticipated from historic hydrographic data. In fact, this might be interpreted as scientific result, rather than as a grievance.

Float Ballasting

All IfM-Kiel floats were pre-ballasted at the Institut für Meereskunde an der Universität Kiel and fine-tuned to the depth of a salinity minimum onboard RV Polarstern after taking a CTD cast at the site of deployment. URI floats were ballasted for two density surfaces (s_q = 26.8 and 27.2) at the Graduate School of Oceanography, URI. These floats were not fine tuned at sea but used as ballasted at URI. For details of the ballasting procedures please refer to Rossby et al. [1986], König and Zenk [1992], Swift and Riser [1993], Boebel et al. [1995] and Anderson-Fontana et al. [1996].

Deployment of Sound Sources

The sound sources were built by Webb Research Corporation and have proven a reliable component over many years. The source were moored at depths between 800 and 1000 m, depending on the local depth of the sound channel. Most sourced functioned reliably for the duration of the study, but two sources failed prematurely. Source R3, deployed from RV Seward Johnson in September 1997 failed to transmit from the very beginning. No signals at all were received by any of the floats from this source. Sound Source K7 failed after a deployment period of 18 months. Signals from its ARGOS watchdog were received, indicating the at least the top flotation of the mooring broke loose, either causing the source to either sink beyond its crush depth or, if the failure occurred below the source, to rise with the flotation package to the surface. To compensate for the resulting loss of acoustic coverage an additional sound source (K11) was moored from RV Polarstern in March 1999. Mooring K10 was meanwhile successfully recovered in May 1999 by RV Polarstern.

The Kiel moorings were designed by Dieter Carlsen (IfM-Kiel), all other (WHOI and URI) moorings by John Kemp (WHOI). Table 6 lists the deployment dates and locations of the sound sources mooring , and their transmission times. The parameter "-offset" refers to any offset that would have been directly measured prior to the source deployment, while "-add_offset" indicates clock offsets at launch time that have empirically been determined and proved to provide the most consistent tracking results. Note that these additional offsets have been used with the IfM-Kiel floats only, while they were set to 0 for tracking of the URI floats. The "-drift" parameter indicates sound source clock drifts empirically determined by comparing expected and measured time of arrivals prior to the float surfacing.

Deployment of Floats

The deployment strategy of the project was to capture the main constituents of water flowing into the Cape Basin (Figure 1). Water was expected to enter the Cape Basin from the south via the South Atlantic Current and from the east via the Agulhas Current leakage. The first KAPEX cruise departed Cape Town on March 21, 1997 [Boebel et al., 1998b] on RV Polarstern. This cruise served three main objectives: a) to capture the presumed inflow of Antarctic Intermediate Water (AAIW) from the western Atlantic into the Cape Basin across its southern and western boundaries (the Agulhas Ridge and southern Walvis Ridge, (Figure 2); b) to reveal the meridional extent of the South Atlantic Current (Figure 1) and the presumed associated strong eastward intermediate depth flow along the Subtropical Convergence; and c) to study the early kinematic evolution of an Agulhas Ring near its formation site, the Agulhas Retroflection, and the influence of these rings on the intermediate water layer.

To capture the Cape Basin inflow (topic a) and the South Atlantic Current (topic b), 30 floats were distributed evenly along the southwestern and southern perimeter of the Cape Basin and along 3 quasi-meridional sections across the Subtropical Front/South Atlantic Current regime (Figure 1). Five additional floats were reserved for deployment into an Agulhas Ring (topic c), which was searched for and identified in upper layer velocity records from shipboard ADCP and thermocline depth measurements using 152 XBTs. CTD casts to depths between 1500 and 2000 m were taken prior to each RAFOS float deployment to enable the individual adjustment of the float's density and, accordingly, its initial drift depth, which was chosen to coincide with the local core depth of the AAIW layer.

For the purpose of capturing the inflow with the Agulhas Current, isopycnal RAFOS floats were launched into the core of the Agulhas Current. The first launch position was close to the city of Port Elizabeth (Figure 1), at the eastern end of the Agulhas Bank (KV, Figure 2) on December 3, 1997. Here the shelf is much wider than at the more northerly launch sites where subsequent deployments took place. The exact location of the current was first established by contemporaneous satellite images in the thermal infrared and subsequently in situ by carrying out an XBT (expendable bathythermograph) section roughly at right angles to the coast. The inshore edge of the Agulhas Current was found more than 60 km offshore and the floats launched at distances between 62 and 80 km, i.e. well within the high speed jet of the current.

Figure 1: Schematic diagram of presumed thermocline and intermediate water flow around southern Africa. PE indicates the location of Port Elizabeth while the location of other relevant cities is given in Figure 2. Isobaths represent 0, 1000 and 3000m depths, with areas shallower than 1000m and 0m respectively hatched lighter (blue) and darker (ochre/yellow), for b/w or color display of the plot.

The next launch into the Agulhas Current took place on 13 and 15 June 1998. The first of these (KI, Figure 2) was at about 31 °S, south-west of Durban. Using XBTs, the landward border of the current was established using the definition used by Gründlingh [1983]. It is the intersection of the 200 m depth level and the 15°C isotherm. This was found at a distance of about 40 km from the shore on 13 June (Figure 2). Floats were then launched in 3 groups of 5 floats each at distances of 42 to 58 km offshore, where the 10° C isotherm intersected the 350 m, the 450 m and the 600 m isobath, to make sure they were well into the current.

Figure 2: RAFOS float deployment sites and sound source locations. Sound sources are indicated by 3 concentric circles and straight lettering, float launch positions by solid dots. Cruise tracks are indicated by thin lines and labeled with italic letters. Isobaths as in Figure 1.>

The launch positions on 15 June 1998 were about 100 km to the northeast of the previous site, directly off Durban and at a wider section of the shelf (Figure 2). The launch sites of 3 groups of floats were defined by the intersection of the 10° C isotherm and the 520 m, 620 m and 700 m isobath. Since the shelf slope is less steep here, the launch sites were farther offshore than at the previous location. The launch position at 700 m is believed to have been close to the maximum velocity core [Beal and Bryden, 1999] whereas the others were closer to its shoreward edge.

Float Tracking

Float tracking was performed using the newly developed ARTOA II GUI package for Matlab (contact oboebel@gso.uri.edu for a free version of this software). In a first step sound source drifts were determined from the observed difference between estimated and measured TOAs (times of arrival) at the first and last day (measurement) of the float's subsurface mission. The results were plotted and an estimate of the sound source drift made on a least square basis. Sound source drifts were small (less than 0.01 seconds per day) for all sound sources and the error of the clock drift estimates is certainly larger than 50% of the values themselves. Nevertheless, the assumption of these drifts during the tracking process led to significantly improved trajectories (fewer jumps when changing sound source combinations) over those calculated without drifts. The trajectories, with a few exceptions where unfavorable geometric situations between sound sources and floats were unavoidable, are estimated to be accurate within 5 km, while relative positions of successive data points are correct within at least 2 km. Initial sound source offsets (-add_offset) as indicated in Table 6 were used for K7, K8, K9 and K10 with the IfM floats, but not with the URI floats, where they were set to zero.

Figure 3: Sound source drift estimates. A drift and initial offset are calculated where possible "All direct" is based on the difference between measured and estimated sound signals arrival times for both, launch and surface times. "All indirect" is based on the difference between these initial and final offsets . "Final direct" is similar to "all direct", but includes only the TOA offsets at the surfacing of the float and assumes zero seconds offset between float and sound source at float launch time.

Acknowledgements

Excellent support by all officers and crews of vessels involved in the KAPEX is greatly appreciated. These vessels were, in alphabetical order, R. V. Dr. Fridtjof Nansen, SA Kuswag I, SA Kuswag V, R. V. Seward Johnson and R. V. Polarstern. Participants in the KAPEX that were not directly involved in the production of this data report, i.e. Chris Duncombe Rae, Dave Fratantoni, Silvia Garzoli and Phil Richardson provided indispensable support and valuable scientific discussions throughout the project. Essential technical and scientific contributions were made by S. Becker, R. Berger, P. Bouchard, D. Carlsen, J. Fontaine, D. Fütterer, U. Hueninghaus, H. Hunt, J. Kemp, M. Lankhorst, M. Menzel, P. Meyer, H. Schünemann, V. Smetacek, V. Strass, T. Strømme, M. Nielsen and C. Wooding. Support from the National Science Foundation, USA (grant no. OCE-96-17986); the Foundation for Research Development and the University of Cape Town, RSA; the Ministerium für Bildung, Wissenschaft, Forschung und Technologie (grant Nr. 03F0157A), the Alexander von Humboldt-Foundation as well as the Alfred Wegener Institut für Polar- und Meeresforschung, all in Germany, is gratefully acknowledged. This is a contribution to the World Ocean Circulation Experiment (WOCE).

Tables

Table 1: Launch parameters of floats deployed during ANT XIV/2 on RV Polarstern. The table indicates the float's serial number, the target pressure, temperature and density, its listening schedule and the final status (DNS = did not show, AOK = all OK, XMS= ARGOS transmission short, PPP = emergency pressure release, BAT = main mission battery low).

Table 2: Agulhas float launch parameters, launched from RV KUSWAG V on December 3, 1997. All floats listened twice a day for sound signals (00:30; 12:30). Abbreviations as in Table 1, with DWL = drop weight lost, and JAM = jammed compressee.

Table 3: Launch parameters of floats set from aboard RV Fridtjof Nansen (Namibia Floats). Abbreviation as above, XSC = transmission scrambled due to data overflow. Abbreviations as in Table 1 and Table 2.

Table 4: Launch parameters of Agulhas float set from RV KUSWAG I on June 13 and 15, 1998. All floats listened twice a day for sound signals (00:30; 12:30). Abbreviations as in Table 1