The annual mean meridional Ekman transport across each repeat of the 25ºN
hydrographic section is computed in 1º longitude bins from the NOC global monthly mean wind stress climatology (Josey et al., 1998; 2002):
Ekman transport = -!xLx
"f (3.1)
(e.g. Gill, 1982) where τx is the zonal component of wind stress, ρ seawater density, Lx
the width of each 1º longitude bin, and f the Coriolis parameter. The NOC climatology is the only one computed from ship observations corrected for biases in observing procedure and is thus expected to be the most accurate of such available products (Josey et al., 2002). Its long term mean (1980 to 2004) Ekman transport across the Atlantic at 25.5ºN of 3.8 Sv is in good agreement with 3.6 Sv from the atmospheric model
reanalysis product of National Centre for Environmental Prediction (NCEP) across 25.7ºN. Variability of the two climatologies is also consistent, each having a standard deviation of 1.9 Sv between 1980 and 2004.
Since Ekman transport (long term mean between 1980 and 2005) at 26.5ºN is 30% lower than that at 24.5ºN it is important to follow the ship’s cruise track of each transatlantic hydrographic section as closely as possible when computing Ekman transport. The 1957 and 1992 sections therefore use wind stress at 24.5ºN across the whole section, while others use values at 25.5, 26.5 and 27.5ºN towards the eastern
boundary, and the last two use 26.5ºN in the western boundary (Figure 3.3). This is also important to account for the different section widths (Figure 3.1) and an additional reason that the NOC climatology is preferable to NCEP which has 1.9º zonal resolution compared to the 1º of the NOC climatology.
Figure 3.3 Grid points of the NOC wind stress climatology used for calculation of Ekman transport (+) (1ºx 1º resolution) with the 1998 (+) and 1957 (•) station positions
for reference. Note that the mid-ocean between 67.5 and 23.5ºW is omitted where all sections follow the 24.5ºN line. Depth contours of 1000, 3000 and 5000m are included for reference and land is shaded.
Figure 3.4 Ekman transport in 1º longitude bins computed from wind stress at the locations of Figure 3.3. The 24.5ºN transports are shown in +, while those angling northward to the eastern and western boundaries are in x.
The resulting long term mean Ekman transport as a function of longitude is shown in Figure 3.4 with the difference due to cruise track changes only amounting to 0.2 Sv between the long term annual means; 4.5, 4.6 and 4.4 Sv for the 1957/1992, 1981 and 1998/2004 tracks respectively. For calculation of the Ekman component of meridional heat transport (Section 3.3.3) the temperature of the Ekman transport at each 1º point in longitude is the mean 10 dbar temperature of the CTD stations within 1º longitude of this point. This follows Wijffels et al. (1994) who found that 80% of the Ekman transport occurs within 0.1ºC of the 5 dbar temperature and all within 0.2ºC of this.
With a 20 dbar vertical resolution dataset (Section 3.2.4), the 10 dbar temperature is therefore the most appropriate for the Ekman transport.
The annual mean Ekman transports (Figure 3.5, Table 3.1) of 1981, 1992, 1998 and 2004 computed from the monthly resolution climatological NOC wind stress
(incorporating the different cruise tracks) have a range of 1.5 Sv from 3.7 Sv in 1981 to 5.2 Sv in 1998, with 10% error (Josey et al. 2002) based on comparison of the NOC monthly mean wind stresses with WHOI meteorological buoy wind observations.
Figure 3.5 Annual mean Ekman transport from the NOC wind stress climatology at 24.5ºN (75.5ºW to 15.5ºW) (+) and following the 1998 cruise track (x). The 1981 cruise track transports agree with those plotted to within 0.1 Sv. The long term mean transport of 4.5 Sv is also plotted (black dashed line).
Year 1957 1981 1992 1998 2004 Ekman transport ±
error (Sv) 4.5 ± 0.8 3.7 ± 0.4 4.6 ± 0.5 5.2 ± 0.5 4.5 ± 0.5
Table 3.1 Annual mean Ekman transports for use in mass balance of mid-ocean geostrophic flow. Years 1981-2004 correspond to plotted values of Figure 3.5. The error is 10% of the annual mean transport reflecting the uncertainty in wind stress (Josey et al., 2002) while that of 1957 incorporates the additional uncertainty associated with using the long term mean transport.
The NOC climatology only extends back to 1980, and for the 1957 section we use the long term mean at 24.5ºN of 4.5 Sv. This is justified since although linear regression shows a trend towards increased transport over time (0.3 Sv per 10 years) this is not significant at the 95% level and is likely due to increase of the winds with the
predominantly positive state of the North Atlantic Oscillation (NAO) over the latter part of the record (Josey et al., 2002). The mean annual station based NAO index from 1980 to 2002 inclusive is 0.6, with standard deviation 2.0, which encompasses the 1957 index of –0.5 (Figure 3.6). There is therefore no justification for assuming the 1957 annual average Ekman transport to be any different from the 1980-2005 mean of 4.5 Sv at
24.5ºN but uncertainty is increased to 0.8 Sv (combining the 10% climatology error with the standard deviation of annual mean Ekman transports from 1980 to 2005 of 0.7 Sv, Table 3.1). Although the NCEP wind stress climatology extends back to 1948, we do not use this to estimate the 1957 Ekman transport since there are uncertainties
associated with the poorer longitudinal and latitudinal resolution of the NCEP reanalysis product relative to the NOC climatology.
Figure 3.6 NAO index based on the annual average normalised sea level pressure difference between Ponta Delgada, Azores and Stykkisholmur/ Keykjavik, Iceland (http://www.cgd.ucar.edu/cas/jhurrell/indices.html) in black. The 1957 value is circled and the limits of the 1980-2002 mean (solid grey line) ± one standard deviation over
this period are extended back to 1950 with grey dashed lines.
3.2.3 Florida Straits transport