|Large-scale transport into the Arctic: the roles of the midlatitude jet and the Hadley Cell|
H. Yang et al.
This is an interesting paper aimed at understanding the cross-model variation in the transport in to the Arctic of tracers with mid-latitude sources. Following several other works that have pointed out the diversity in Arctic abundances of these tracers, this work points out convincingly the importance of transport by the zonal mean, meridional circulation for understanding this diversity in the CCMI multi-model ensemble. An important wrinkle to this is that this meridional transport differs even across models that specify the meteorology in some way. Although the role of the mean meridional circulation does not explain all aspects of inter-model disagreement regarding large-scale transport into the Arctic, it seems to be a significant contributing factor, and well worth highlighting.
I find this to be an interesting study well worth publishing. I have some questions about the methodology and some suggestions about presentation, but if these are addressed I would recommend publication. I also found the writing to be a bit difficult to follow at times.
My only more significant comment is regarding the choice of flux decomposition into a zonal mean an eddy component. I was initially somewhat surprised at this choice, rather than decomposing into a stationary (say,monthly mean) and time-varying component, as it seems the latter would be better suited for understanding the local poleward transport in the Pacific sector. The time mean would filter out transport from synoptic scale eddies just as well. I would have thought the zonal mean meridional velocity would be just as, if not more, relevant to NH50 transport as to CO50, but that doesn't seem to be the case (I appreciate the CMF argument and this may well be the story). Perhaps the zonal decomposition is easier to study with the model output available (though this isn't so clear to me). Have you tried this decomposition?
On a related note, it seems that a second way to quantify the contribution of the meridional transport (beyond inter-model correlations) would be to directly calculate the tracer flux from the time mean (or zonal mean) meridional winds. I would expect this to have substantial variations even in CO and in NH50, even if there are confounding effects that reduce the correlations with overall Arctic burdens. This could be quite helpful for quantifying the importance of this effect for the transport of realistic tracers. Possibly one could even try isolating the role of inter-model differences in meridional velocity by computing the flux with a reference tracer distribution.
With regards to the presentation, I wonder if it would help to better distinguish between model integrations with constrained meteorology and those with free running dynamics. I have included specific suggestions below.
p3 l29 - Are the source data for this interpolation really on the model native grid? I would be surprised if any model runs in isobaric coordinates rather than hybrid pressure (this seems to be confirmed by Table 1). I suspect it's more likely that the source data have been interpolated by the modeling centers.
p6 l10-11: I was confused by the sentence starting 'However, it is difficult....' Is the point that these integrations are different because of how they implemented the source of CO50, but that they may also differ in their transport?
p6 l15: I can see how this source variability increases the model range in winter, but the EMAC models are right in the middle of the model distribution for summer Arctic burdens so the latter assertion wasn't so clear to me.
p7 l20-24: I wasn't so convinced by this argument - even if the transport into the Arctic is weaker in the summer, if the sources are at midlatitudes, this transport still seems essential for determining summer burdens, no? Unless a large fraction of the summer is left from winter-time transport, in which case this is probably worth discussing, since transport in the winter and spring would be more important than summer time transport.
p9 l12: This is inconsistent with the caption of Fig. 8 that gives an upper bound is 200 hPa.
p12 l15: By 'unchanged' do the authors mean that the model anomaly from the ensemble mean NH50 concentrations at high latitudes is consistent with its anomaly at midlatitudes?
For all figures I found myself wishing there was a clearer sorting in the legends by specified versus free dynamics. I don't think it is said anywhere explicitly that the open circles are specified dynamics integrations. Also, where correlation coefficients are given it would be informative to show coefficients for only the specified/free dynamics runs. Some of the regression lines (e.g. Fig. 10c) seemed to be obscuring clear differences in behavior; similarly in panels 12d,e,h, for instance, the two sets of integrations seem to follow rather different regression lines. This is discussed at some points in the text, but the presentation could be made clearer.
Fig. 2,4, and similar panels in the following: Would it be helpful to show ensemble means separate for specified dynamics and free running models?
Fig. 5: In this figure especially, it is hard to take in these 16 panels - sorting them by specified versus free dynamics would be more useful than by modeling center.