Interpretations of tracer derived ages are strongly dependent on the models used [e.g. Weiss et al., 1985], and on surface boundary conditions peculiar to each tracer [e.g., Doney and Jenkins, 1988; Wallace et al., 1992]. Several of the recent tracer papers discuss the difference between DWBC component spreading rates derived from tracer ages versus direct velocity measurements. Tracer derived spreading rates in the DWBC vary from 0.7 to 2 cm/s, while direct velocities up to 50 cm/s have been measured. The consensus is that tracer velocities inherently reflect the effects of two processes acting on the DWBC transport. The first is mixing [e.g. Jenkins and Clarke, 1976; Musgrave, 1990; Thiele and Sarmiento, 1990]. When the DWBC mixes with ``older'' tracer bearing interior waters, the process will cause the tracer ages to be biased toward the older component [e.g. Smethie, 1993]. The second is recirculation. Transport in recirculation gyres causes aging of water parcels by increasing the distance traveled since leaving the source region. Another consideration is the amount of time an individual particle actually spends in the DWBC's maximum velocity core, as compared with time spent in the lower velocity shoulders [ Abell et al., 1994]. Just how much each process affects the tracer derived age and spreading rate remains unknown. For these reasons the tracer derived DWBC ages are usually interpreted as an upper bound, and the spreading rates as a lower bound.
Models have been used to examine the effect of mixing on
tracer derived velocities of the DWBC. Pickart et al. [1989]
showed that the F11/F12 ratio is altered by the processes of
formation and overflow. The F11/F12 ratio was lowered by
increasing the residence time of water before entering the DWBC,
and by mixing with water surrounding the DWBC core. They found
model DWBC velocities of 5-10 cm/s were required to match derived
velocities from CFC observations between 40
and 45
N.
Rhein [1994] extended the Pickart et al. [1989] model
to 10
S, and included tritium data to allow validation prior
to 1982. She took into account the contribution to the LNADW of
three water masses, the DSOW, ISOW and Northeast Atlantic Water,
entering the DWBC at different locations downstream. In addition,
she used the tracer boundary conditions for these water masses as
reconstructed by Smethie and Swift [1989] and Rhein
[1991]. Rhein [1994] estimated that the maximum possible
model velocity compatible with the data was 5 cm/s. Thus, both the
Pickart et al. [1989] and Rhein [1994] model results
suggested that when accounting for the effects of mixing, it was
possible to increase the tracer velocity to 5 cm/s. However, there
is still a formidable gap between 5 cm/s and direct velocities of
up to 50 cm/s.
Tracer derived spreading rates and direct velocity measurements both provide different and useful information. There appear to be several distinct modes of response by the DWBC downstream to changes in high latitude forcing. These modes include waves that respond on the order of a few months, while water flowing in a direct pipeline at the maximum velocity responds on the order of one-two years, and tracers spreading on the order of decades [ Abell et al., 1994]. Recent observations suggest that much of the difference between tracer derived and direct velocities is attributable to recirculation gyres, but this remains to be confirmed. What is remarkable, is that the float data agreed with both the direct and tracer derived velocities. Floats deployed in the DWBC moved at high velocities similar to the Eulerian measurements, although often in circles. However, the float derived equatorward average spreading rates of about 2 cm/s [ Leaman et al., 1994] were similar to the high end of the tracer derived rates. It is important to recognize that both methods are providing different information. The direct velocity measurements give information on the temporal and spatial variability in the flow field, and the dynamics of the forcing. The transient tracer derived velocities integrate the effects of many processes giving an ``effective'' spreading rate.