R. Steinfeldt et al.: Anthropogenic carbon in the Atlantic 3853 Table 5. Decadal changes in Atlantic Cant inventories (1tCant, in Pg C). The inventory differences are obtained from the decadal-data-only inventories listed in Table 3. They are compared with the values given in Woosley et al. (2016) (based on the eMLR method, the Cant increase from 1990–2000 in Woosley et al., 2016, is adopted from Wanninkhof et al., 2010) and Müller et al. (2023) (eMLR(C?) method), where the increase is calculated for the periods of 1994–2004 and 2004–2014). This study Woosley et al. (2016) Müller et al. (2023) 1990–2000 North 4.2 ± 1.4 1.9 ± 0.4 4.8 ± 0.2a South 2.2 ± 1.7 3.2 ± 0.7 3.9 ± 0.5a Total 6.4 ± 2.8 5.1 ± 1.0 8.7 ± 0.7a 2000–2010 North 5.1 ± 1.6 4.4 ± 0.9 3.9 ± 0.4b South 3.6 ± 2.1 3.7 ± 0.8 5.4 ± 0.6b Total 8.6 ± 3.4 8.1 ± 1.6 9.3 ± 1.0b 2010–2020 North 5.5 ± 1.8 South 4.6 ± 2.3 Total 10.2 ± 3.5 a Value for the 1994–2004 period. b Value for the 2004–2014 period. Table 6. Cant accumulation anomalies for the Atlantic Ocean (1tCanomant ), i.e., deviations between the Cant increase based on tracer data from the actual period and the predicted Cant increase based on tracer data from the previous period. Cant2000 ? Cant1990?2000 Cant2010 ? Cant2000?2010 Cant2020 ? Cant2010?2020 North 0.1 ± 1.5 0.2 ± 1.6 ?0.1 ± 1.8 South ?1.8 ± 1.8 ?0.8 ± 2.1 ?0.2 ± 2.3 Total ?1.7 ± 2.9 ?0.6 ± 3.4 ?0.4 ± 3.6 increase above the detection limit, at least in some places. These also contribute to the increase in the column inven- tory of the deep and bottom waters south of 40°S shown in Fig. 10f. The differences between the decadal Cant storage rates are discussed in more detail in Sect. 3.3 on decadal vari- ability. 3.2.2 Comparison of local Cant changes from the modified TTD method with dilution with other publications We now compare our inferred local Cant changes in the At- lantic with other published results. For the subpolar North Atlantic, the area with the highest increase in Cant col- umn inventory, Pérez et al. (2010) find similar storage rates (1.74 mol m2 yr?1 in the Irminger Sea and 1.88 mol m2 yr?1 in the Iceland Basin) to those shown in Fig. 10 but only from 1991–1997, where the North Atlantic Oscillation (NAO) was in a high phase. Afterwards, in the low-NAO period be- tween 1997 and 2006, their rate is less than a quarter of the previous value (0.3–0.4 mol m2 yr?1). For the northeastern Atlantic, Pérez et al. (2010) also yield lower storage rates (0.72 mol m2 yr?1 for 1981–2006) compared to our analyses (> 1.0 mol m2 yr?1, Fig. 8a, c, d). In contrast to Pérez et al. (2010), our results are averaged over a larger region and also a longer time period (1 decade compared to 6 and 9 years in Pérez et al., 2010, for the Irminger Sea and Iceland Basin), which may lead to a damping of sudden, regional changes in the Cant storage. However, the low Cant increase in Pérez et al. (2010) after 1997 also points to methodological differ- ences between the ?C0T method used in Pérez et al. (2010) and the modified TTD method with dilution used here. One of these is that the ?C0T method takes into account changes in the ocean air–sea CO2 disequilibrium over time. On the other hand, a comparison of Fig. 11a and c indicates that in the 2000–2010 decade, the Cant storage in the deeper part of the LSW is indeed very small (due to a reduction in the convection depth). The other water masses, however, i.e., the Overflow Waters and the waters above 1000 m, do not show a decrease in the Cant uptake, in agreement with the ongoing renewal of these water masses. Thus, the small increase in the Cant column inventory after 1997 in Pérez et al. (2010) seems to be unrealistic. Directly within the Labrador Sea, Raimondi et al. (2021) find a mean storage rate of 1.8 mol m2 yr?1 for https://doi.org/10.5194/bg-21-3839-2024 Biogeosciences, 21, 3839–3867, 2024