Background & FAQ
(For general background information on the entire AOML Contribution to the assessment of the State of the Oceans, see also the State of the Oceans Background & FAQ.)
The Atlantic Ocean is the major ocean basin involved in large-scale northward transports of heat typically associated with the MOC, where warm upper layer water flows northwards, and is compensated for by southward flowing North Atlantic Deep Water. This large-scale circulation is responsible for the northward heat flux through the entire Atlantic Ocean. Historical estimates of the net northward heat flux in the vicinity of its maximum, which occurs in the North Atlantic roughly at the latitude of the center of the subtropical gyre, range from 0.9 PW to 1.6 PW, while estimate in the 30S to 35S band are even more uncertain, ranging from negative to more than 1 PW. While much of this variability may be a consequence of the different methods used to estimate the heat transport, natural variability cannot be ruled out. The importance of this heat transport to the world climate together with the possibility of monitoring its variability motivates this project.
AOML collects XBT data on two lines spanning the subtropical oceans: in the North Atlantic since 1995 (twice in 1995 and quarterly thereafter) along AX7 running between Spain and Miami, Florida and in the South Atlantic since 2002 (twice per year until 2004 and quarterly thereafter) along AX18 between Cape Town, South Africa and Buenos Aires, Argentina. These data capture the upper limb of the MOC transport. In the North Atlantic much of the northward transport is confined to a strong boundary current through the Florida Straits, where XBT data can also be usefully augmented with other data from the NOAA/OCO funded Florida Current transport program.
Heat transport has already been successfully computed using XBT data (Roemmich et al, 2001), however the methodology for estimating the transport can be improved. In particular, as density is essential for the flux estimates, results depend on how well salinity profiles can be estimated to complement the XBT data and on how well the profiles can be extended to the bottom of the ocean. Improving these estimates to achieve more accurate fluxes is an essential part of this project, as is a careful quantitative assessment of the accuracy of the resulting fluxes. (Read more in Uncertainty Estimates...)
Northward mass (M), volume (V), and heat transport (H) through a vertical (xz) section can be estimated directly from observations.
The northward velocity v can be treated as a sum of three terms: (i) a geostrophic contribution (thermal wind equation) relative to a prescribed reference level, (ii) an ageostrophic part modeled as Ekman flow, and (iii) a barotropic part define as the velocity at the reference level. Density, and hence geostrophic velocity can be obtained from XBT data if salinity is accurately estimated and data are extrapolated to the ocean bottom.
XBTs are small expendable probes that fall freely through the water column measuring resistance with a small temperature thermistor. Two very small wires transmit the temperature data to the ship where it is recorded for later analysis. The probe is designed to fall at a constant rate, so that the depth of the probe can be inferred from the time after launch (Hanawa et al 95). XBT profiles are obtained from an automated system mounted on the stern of the vessel, with a Sippican communications circuit board and GPS receiver. The automatic launcher is remotely controlled from a computer installed in a stateroom and holds 6-8 XBTs at a time. Sippican T-7 XBTs are deployed which record temperature time series collected every 0.1 second, to depths that extend to about 850 meters. The computer performs crude quality control which automatically deploys a new XBT if specified tolerances are exceeded. If required, station spacing is adjusted to sample interesting features.
Preliminary estimates of mass and heat transport have been obtained from temperature profiles collected along AX07 and AX18 high-density lines. Salinity was estimated for each profile by linearly interpolating the closest of Levitus' climatological mean salinity and temperature profiles to the XBT temperature and the climatological profiles were used to extend the data to the bottom. In computing geostrophic velocities, a reference level, based on previous work in the literature and on what is known about the circulation, was prescribed just below the northward flowing Antarctic Intermediate Water (σ0=27.6 kg m-3 in the North Atlantic and σ0=27.4 kg m-3 in the South Atlantic). Within strong flows such as the Florida Current or the Malvinas Current where no level of "no motion" can be found, the transport must be specified (e.g. by the mean value of the Florida Current, etc.) The velocity at the reference level is adjusted so that the net mass transport across the section is zero using a single velocity correction for each section. Typically, values of this correction ranged from 10-4 to 10-6 m s-1.
Refining uncertainty estimates for heat transport is in itself a critical component of ongoing heat transport research at AOML. Please refer to the "Heat Transport Uncertainty Estimates" page for more details...
In the North Atlantic, the heat transport was found to vary on inter-annual time scales from 0.8 ± 0.2 PW at present to 1.2 ± 0.2 PW in 1996 with instantaneous estimates ranging from 0.6 to 1.6 PW . Heat transport due to Ekman layer flow computed from annual Hellerman winds was relatively small (only 0.1 PW). This variability is entirely driven by changes in the interior density field; the barotropic Florida Current transport was kept fixed (32 Sv). Improvements to these estimates should include: (i) time varying Florida Current transports and wind fields, (ii) improved salinity estimates and extrapolations to the bottom, and (iii) improved uncertainty estimates that reflect the effect of the initial reference level, the wind field variability, the importance of barotropic flows, and the uncertainties of the salinity estimates and the extrapolations
During preliminary tests assessing the validity of the methodology in the South Atlantic, several of these error issues were considered more closely. Simulated data from a general circulation model indicated that obtaining heat transport using a level of no motion instead of the total velocity results in a bias error. At 30°S (35°S), the heat transport calculated from the velocity field was 0.55 PW (0.51 PW), the heat transport calculated from T and S was 0.85 PW (0.80 PW), and the heat transport calculated from T and S adjusted by a bottom velocity at the boundary provided by the literature was 0.60 PW (0.50 PW). The conclusion was that the transports must be adjusted for a barotropic component in the Malvinas Current region and that the "level of no motion" assumption does not hold near strong western boundary currents. Note that in the North Atlantic a good estimate for the barotropic western boundary current is readily available through the Florida Current monitoring program (see Western Boundary Current Time Series Project). Current meter moorings deployed by French scientists in the Malvinas Current will provide corresponding values for the South Atlantic. Data collected along the A10 WOCE line located near 30°S were used to estimate the errors induced by using a climatological deep T, S field from the Levitus product instead of actual data. Results indicated that this procedure can introduce an error of up to 0.2 PW and that in order to reduce the errors a better climatology and better T/S relation are needed. Additional results indicate that the use of different wind products to calculate the Ekman component of the flow induced an uncertainty of less than 0.1 PW. Particularly in the South Atlantic, the Ekman transport estimates vary with latitude and season. One of the main challenges to providing an accurate heat transport is the lack of accurate information on the South Atlantic boundary currents.