The distributions and transports of deepwater masses at the western bo
undary in the tropical Atlantic off Brazil have been studied on three
surveys along 35 degrees W and 5 degrees S and one at 10 degrees S. Tr
ansports are obtained from direct measurements of the velocity fields
(Pegasus profiling system and lowered acoustic Doppler current profile
r) and from geostrophic computations. Using chlorofluoromethane (CFM)
and hydrographic distributions, four water masses could be identified
forming the North Atlantic Deep Water (NADW) system. Two of these have
a high CFM content, the ''shallow upper NADW'' (SUNADW) and the ''ove
rflow lower NADW'' (OLNADW). These exhibit the highest velocity signal
s at 35 degrees W, where distinct flow cores seem to exist; most of th
e southeastward flow of the SUNADW (centered around 1600 m) occurs 320
km offshore between 3 degrees 09'S and 1 degrees 50'S (9.7 +/- 3.3 Sv
); farther north in that section, a highly variable reversing flow is
found in a second velocity maximum. The transport of OLNADW (centered
around 3800 m) of 4.6 +/- 2.6 Sv is guided by the Parnaiba Ridge at 1
degrees 45'S, 35 degrees W. The water masses located between the two C
FM maxima, the Labrador Sea Water (LSW) and the LNADW old water mass (
LNADW-old), did not show any persistent flow features, however, a rath
er constant transport of 11.1 +/- 2.6 Sv was observed for these two la
yers. The total southeastward flow of the NADW at 35 degrees W showed
a transport of 26.8 +/- 7.0 Sv, if one neglects the reversing SUNADW n
orth of 1 degrees 50'S. At 5 degrees S the flow of all deepwater masse
s shows vertically aligned cores; the main southward transport occurre
d near the coast (19.5 +/- 5.3 Sv). The boundary current is limited of
fshore by a flow reversal, present in all three surveys, but located a
t different longitudes. At 10 degrees S a southward transport of 4.7 S
v was observed in November 1992. However, the section extended only to
32 degrees 30'W, so that probably a significant part of the flow has
been missed. An important result is the large transport variability be
tween single cruises as well as variability of the spatial distributio
n of the flow at 35 degrees W, which could lead to large uncertainties
in the interpretation of single cruise observations. Despite these un
certainties we suggest a circulation pattern of the various deepwater
masses near the equator by combining our mean transport estimates with
other observations.