|Second review: Halloran et al, The mechanisms of North Atlantic CO2 uptake in a large Earth System Model ensemble, Biogeosciences.|
I reviewed this manuscript previously and I find it clearer and easier to follow thanks to the extra subsections and the addition of supplementary material.
My only remaining concern is regarding my previous comment about the lack of sensitivity to the box model MOC. The authors have persuaded me that portions of the MOC variability will be included due to prescribed changes in temperature (and also alkalinity) in the high latitude northern surface box (the first part of my comment about the Perez Nature Geoscience paper). This, I suppose, highlights the continuing difficulty in cleanly separating emergent properties of ESMs, which I think the authors do a good job. The new figure showing the timeseries of prescribed inputs shows that MOC and alkalinity decrease, which would make sense if there was a reduction in transport of high alkalinity low latitude waters to the north. Lower alkalinity waters have a reduced saturated carbon content and therefore uptake of CO2 from the atmosphere would be reduced. Prescribed temperature increase would exacerbate this effect.
However, as I pointed out previously (in the second part of my comment), the key process behind the hypothesized rise and fall of North Atlantic CO2 fluxes proposed in the abstract, in Figure 6, and by Volker et al. (2002) is the increased northward surface transport of DIC from low latitude to high latitude regions (i.e. by the box model MOC). This advective flux largely satisfies the North Atlantic surface saturated DIC capacity causing reduced uptake from the atmosphere despite increased atmospheric CO2 levels. DIC transport could increase, despite reduced volume transport if the concentration in the low latitude box was increased, as is suggested by increased anthropogenic CO2 uptake in that region forced by increased atmospheric pCO2.
The ESM ensemble does appear to show this characteristic rise and fall of North Atlantic CO2 fluxes. However, I am still struggling to reconcile the proposed mechanism with the results from the box model emulator. In all but one of the parameter sets shown in table 2 (#6), the south to north transport mostly occurs from the Southern Ocean surface box to the low latitude intermediate water box (small values of a). In an additional set (#3) there is considerable vertical upwelling (large values of b) into the low latitude surface box, although vertical mixing is considerable for some of the other parameter sets. Nevertheless, the only upper ocean route into the North Atlantic surface box is through the MOC (that is high values of a+b), which is only significant in two of the parameter sets (#3 and #6). Perhaps this is one reason that filtering of the MOC input from the ESM has little effect because the delivery of DIC to the high latitude North Atlantic is not occurring through the proposed mechanism in the emulator – even in parameter set #1, only 30% of the MOC transport goes from the surface low latitude to North Atlantic box. In parameter set #2 less than 10% of the MOC transport take the surface MOC route.
Mixing between northern surface and deep boxes can supply DIC to the North Atlantic but the majority of this was last in contact with the atmosphere in the Southern Ocean (i.e. sets with small a+b). So, instead of the headline mechanism through surface transports, there is outcropping of waters that have been isolated from the atmosphere into the surface North Atlantic. According to Volker et al. (2002), vertical mixing could actually increase the high latitude uptake of anthropogenic CO2 through exposure of older waters with low anthropogenic CO2 content. One could see how vertical mixing could reduce net CO2 uptake instead if natural DIC fluxes are include, through reemergence of regenerated-dic-rich waters transported from the Southern Ocean, but biological activity is not considered by the emulator.
I would suggest that the authors be clear about the fact that the box model emulator can get the right answer (rise and fall in high latitude CO2 fluxes) for the wrong reason (not advection from low to high latitudes) yet still “score” highly in R2 compared to the full ESM. As the authors state, the saturated carbonate chemistry of temperature, salinity, alkalinity and atmospheric CO2 is well established, but it still feels unsatisfactory to prescribe these parameters and then suggest that the resulting change in CO2 fluxes are due to surface advective transports of DIC, particularly because neither DIC concentration nor transports are presented for the emulation.
References: Latex formatting error “Corbi‘ere”.
Page 10, line14-15: Typo “…gives us confident that the box model…”
Page 14, line 27: Typo “...EMS temperature…”
Table 1 and Table 2: Please use consistent terms for model components: fluxsouth/fluxeq/fluxnorth vs piston (latitude), mixeq/mixnorth vs mixing/mixing2.