Atlantic Meridional Heat Transport

South Atlantic Section: AX18

Methodology & Results

Methodology Used

Uncertainty Estimates


For general background information on the entire AOML Contribution to the assessment of the State of the Oceans, see the State of the Oceans Background & FAQ

For broader information on AOML Meridional Atlantic Heat Transport estimations, see the Heat Transport Background & FAQ

Methodology Used for South Atlantic:

The methodology used to obtain the baroclinic component of the heat transport from the XBT data collected along AX18 has been published by Baringer and Garzoli (2007). it can be summalized as follows: Salinity (S) is estimated for each XBT profile by using two-dimensional fields of S(T,p) relationships at different locations created by Thacker et al., (2004) for the South Atlantic using ARGO profilers and CTD data available in the region. (Please see these figures of S(pressure, longitude) and T(pressure, longitude) in the upper 850m of the ocean.) More information on the Salinity estimation can be found here. Where insufficient CTD and ARGO data are available, the World Ocean Atlas 2001 (WOA01) gridded annual climatology (Stephens et al. 2002) is used to estimate S(T). In order to obtain the heat transport across the section, the total mass transport must be zero (i.e., mass must be conserved). The XBT probes sample the ocean only up to about 850m, hence the data are extended to the bottom using the WOA01 gridded climatology by interpolating the data to the location of each XBT to generate an annual mean climatology for the deep ocean. The bottom is determined to be the depth at the location of the XBT from the Smith and Sandwell (1997) 2-minute database of bathymetry.

Geostrophic velocities are determined using the dynamic method where a level of no motion was chosen at a depth just below the northward flowing Antarctic Intermediate Water at σ0=27.4 kg m-30 defined as potential density relative to the surface) and σ2 = 37.09 kg m-32 defined as potential density relative to 2000 dbar) to compare with previous results. Ekman transports are determined using NCEP climatological monthly mean reanalysis winds by interpolating the NCEP values to the location of the XBT observation. (Please see this station map with wind field.) Transports are computed in layers and summed for the entire water column. A simple constant velocity for each section is then applied uniformly at the reference level so that the net mass transport across the section was zero. Typically, values of this velocity range from 10-4 to 10-6 m s-1. Additional corrections to the net transports are made when needed to account for barotropic motions and for XBT sections that failed to terminate at the 200 m isobath or shallower on each side of the Atlantic. Results had been published by Garzoli and Baringer 2007.

Back to Top of Page...

Uncertainty Estimates:

Possible uncertainty in the findings can result from the assumptions applied, including: the simple mass balance geostrophic method, the uncertainty in salinity assigned to each temperature observation, the representativeness of Levitus data below 800 m, and the variable latitudes that each XBT section crosses the Atlantic. Sensitivity to these factors and others such as the specific wind climatology used and its inherent uncertainty are discussed in detail at this link...

Details on determining the overall uncertainty can be found in Baringer and Garzoli (2007) Part I and Part IIavailable here. Based upon the tests made with the A10 full water column CTD data, our best estimate of the uncertainty is about 0.36 PW which includes a possible bias of 0.21 PW due to questionable representativeness of the deep density field used. Reducing and understanding these errors is an active and challenging field that requires much further study.

Back to Top of Page...

Ekman fluxes from different wind products

The total Ekman heat flux is define as the difference between the Ekman temperature transport (in the Ekman layer) and the section average temperature times the Ekman mass flux (so that the Ekman transport is mass-balanced).The following figure shown Ekman heat flux between 38oS and 28oS using Levitus temperature and NCEP winds.

However several different areal temperature averages were compared (defined as 'cases' below). The temperature field used was either from the Levitus Climatology alone or a combination of the XBT observations (0-850 meters) and Levitus data (below 850 meters). Results are shown in following figure. The different average of total heat transport is less than 0.06 PW. However there are interesting variations over time linked to mesoscale variability in the region, Brazil Current meandering etc that the NCEP winds were used for the heat transport estimates.

Case Wind Product Average Temperature
case 112 NCEP monthly climatology XBT-Levitus temperature field
case 122 NCEP monthly XBT-Levitus temperature field
case 142 Hellerman annual mean XBT-Levitus temperature field
case 152 ECMWF monthly climatolog XBT-Levitus temperature field

a) The Ekman flux is determined from wind stress values. The EKman fluxes differ by less that 0.06 PW.

b) Differences between the Ekman fluxes.

c) The total heat transport using the five different Ekman flux estimates.
The difference average of total heat transport is less than 0.06 PW The geostrophic heat transport is determined from the XBT section (see Baringer and Garzoli, 2007).

All calculation theme produced these results are made by Qi Yao. For more information, contact

Back to Top of Page...

The original AX18 High density XBT line, was done from Cape Town to Buenos Aires, using the Evergreen Container ships. After the cruise conducted in March/April 2007, Evergreen decided to eliminate that transect. After several attempts to reinstall the line, the observations were resumed in October 2007, using first CMA CGM, then Hamburg Sud ships, and different ports.The line is from Santos to Cape Town or Cape Town to Rio. Arrival and departure from Brazil was necessary for the deeper draft requirements.

Data collected along AX18 was used to estimate the oceanic heat transport across nominally 35oS. The methodology used, as well as the error of the estimates were publish by Baringer and Garzoli (2007) and applied to the South Atlantic by Garzoli and Baringer (2007). A summary of the methodology is described in the previous pages.

Figure 1: Different transects for AX18

In order to compare the estimated obtained from both lines, (Figure 1) the northward heat transports estimated from the AX18 line after March 2007 need to be adjusted to make the estimates equivalently to the heat transports across 35oS (the previous AX18 location), that is, the difference between the heat gain from the atmosphere and the ocean heat storage rate (Figure 2, green line) should be added to the new estimates.

An estimate of the heat balance in the triangle area was obtained as the difference between the heat gained from the atmosphere and the ocean heat storage rate. The first term is obtained from NCEP air-sea heat fluxes, the heat storage rate was estimated using monthly temperature/salinity climatologies from the World Ocean Atlas 2005.

Figure 2: Heat balance in the triangle formed by the two different AX18 transects.


Table 1:

Results of the heat transport from line AX18 Case 122 using reference level σ2 = 37.09 kg m-3.

MonthYearEkman Flux (1015W)Geostrophic Heat Transport (1015W)Total Heat Transport (1015W)Normalized to 35oS

(*): 1 PW = 1 petawatt = 1015 Watts

Latest Quarterly Report:

Click on the image below to see full details about the latest Quarterly Report... Latest Report

Back to Top of Page...