A two-dimensional, vertical-section model is described. The model takes density data along a section as input and calculates the velocity through the section relative to the bottom. We have used it to analyze changes in the circulation of the North Atlantic by taking as input the objectively analyzed density data of Levitus for the climatological annual means and for the pentads 1955--1959 and 1970--1974. Sections along 55.5¿W, 54.5¿N, and 23.5¿N have been considered, and estimates of the poleward heat transport through 54.5¿N and 23.5¿N have been made. Ekman transports are calculated using the Hellerman and Rosenstein wind stress field and wind stress analyzed by da Silva et al. for the period 1945--1989. At 54.5¿N, mass balance is achieved by combining with the absolute transport calculations of Greatbatch et al. giving values for the poleward heat transport of 0.6 PW, 0.7 PW and 0.5 PW for the climatological annual mean and the pentads 1955--1959 and 1970--1974, respectively. These values compare well with previous estimates obtained using surface heat flux calculations but are not distinguishable from each other. We have performed a similar calculation for 23.5¿N, this time by requiring mass balance through the section assuming that the northward transport and flow temperature of the Florida Current was the same in each pentad and equal to the climatological annual mean value, an assumption we believe to be justified.

The calculated heat transports are 1.2 PW, 1.0 PW, and 0.8 PW for the climatological annual mean and the pentads 1955--1959 and 1970--1974, respectively, with an estimated error of at least ¿0.3 PW. The climatological value agrees with previous estimates at this latitude. The corresponding heat transports calculated using potential temperature and geostrophic velocity zonally averaged over that part of the section to the east of the Bahamas are also 1.2 PW, 1.0 PW, and 0.8 PW. The error in these calculations is likely to be smaller, indicating that the 1970--1974 value may be less than that for climatology. We show that in the 1970--1974 case, the zonally averaged velocity is more southward above 1000 m and that this accounts for the change in the calculated heat transport.

We note the sensitivity of this change to rather uncertain differences between the climatology and 1970--1974 data sets in bottom density along the topographic slope. Other results concern the vertical distribution of the absolute transport changes diagnosed by Greatbatch et al. Along 55.5¿W, the eastward transport of the Gulf Stream relative to the bottom is some 30 Sv less in the 1970--1974 case than in that for 1955--1959, a change comparable in magnitude to that found by Greatbatch et al. However, the maximum change found by Greatbatch et al. is displaced to the south of that diagnosed relative to the bottom, indicating the importance of changes in bottom velocity. Greatbatch et al. also considered a case in which the density below 1500 m is assumed to have remained unchanged between the pentads, and diagnosed an eastward transport change for the Gulf Stream of over 20 Sv. On the other hand, the change in eastward transport relative to and above 1500 m is only 7 Sv, indicating that even in this case, changes in bottom velocity play an important role. ¿ American Geophysical Union 1993