Introduction

The amount of influx of Indian Ocean intermediate water into the Atlantic is a crucial parameter in the characterization of the global thermohaline circulation since the added heat and salt from the Indian Ocean help to precondition the Atlantic for deep convective events.
 
  While the influx of Indian Ocean Central Water from the Indian to the Atlantic Ocean has been documented by observational studies [e.g. Gordon et al., 1992], the behaviour of the deeper layers is less clear.
 
  On the average, North Atlantic Deep Water and bottom waters are expected to flow in the opposite direction, i.e. to escape from the Atlantic to the Indian Ocean, while the fate of Antarctic Intermediate Water (AAIW) is least known [Shannon and Hunter, 1988].
 
 
 
 
 
 
 
  AAIW forms a layer about 500 m thick between the Central and Deep Waters and is clearly characterized by a vertical salinity minimum near the 27.2 isopycnal surface. On this surface, distinctly different minimum salinities are observed in the Indian (S = 34.40) and Atlantic (S = 34.25) Oceans
 
  Despite this unmistakable tracer signal, it is difficult to determining the inter-ocean transport. The problem is due to the strong mixing of these water types and the underlying intense mesoscale flow field which defeats a climatological description by a few sections.
 
 
 
  The main inter-ocean exchange mechanism for Central Water is undoubtedly the formation of Agulhas Rings at the Agulhas Retroflection region. [Lutjeharms, 1996], but this is arguable for the intermediate depth layer.
 
  Ring shedding involves closing of the Agulhas Retroflection onto itself. This process requires an increasing narrowing of the Retroflection loop. This can either be triggered by the intruding cold core wedges of subantarctic origin from the south or via boundary trapped pulses of the Agulhas Current.
 

Classical conceptual image of Agulhas Ring separation, taken from Schmitz (1996, his figure II-132) and Duncombe Rae (1991, his figure 5). The initiation (2), development (3), and separation (4) of an Agulhas ring from the Agulhas Retroflection (1) is caused here by the northward protrusion of subantarctic water. Our study indicates an alternative mechanism, that is the cutoff of the Agulhas Ring through the southward protrusion of water trapped within the Pulse.
 

To obtain a mesoscale resolving, multiyear estimate of the subsurface flow field, an international team of scientists collaborated within KAPEX (Cape of Good Hope Experiments; [Boebel et al., 1998]) to deploy a total of 114 RAFOS floats in the region around southern Africa. These instruments [Rossby et al., 1986] are designed to drift passively at intermediate depth, while recording their distance to moored sound sources and pressure and temperature on regular intervals. Returning to the sea-surface after pre-programmed subsurface mission periods, the floats transmit the collected data via satellite to the researcher, who can determine the floats' trajectories and hydrographic environment.