Please revise according to the new comments of the reviewers, which I agree with.

In re-reading the draft, I also found these additional issues (all somewhat minor):

I agree with Referee #2 that some additional discussion is needed of role played by CONTRAIL data in yielding changed estimates of the Asian sink. What feature in the CONTRAIL data is responsible for this change?

Page 13, line 12. "ppm" is missing on the first number.

Figure 2 caption: "These observation data download from the NOAA-ESRL and WDCGG network in Figure 2a ". The meaning of "these observation data" is unclear, and note the recursive reference to Figure 2a.

Figure 3: The meaning of the solid blue points in the airborne plots is unclear. They need to be identified in a legend.

Figure 5 and related text: Please explain how it is possible to use atmospheric data, which provides only large-scale information, to resolve land carbon fluxes by ecosystem type. Referee # 1 made a similar point.

I have heard that CarbonTracker cannot be used in its normal configuration to assimilate airborne data because the several week window used for the optimization is too short compared to vertical transport time scales. If I'm wrong about this, you can ignore, but otherwise please explain how this obstacle was overcome in this study, thereby allowing the use of the CONTRAIL data.

The document needs editing by a facile English speaker (several coauthors are suitable) to fix some obvious language issues.

Figure resolution was poor on the submitted drafts. Higher resolution images are needed for final publication.

I'm satisfied with all significant issues except one, which is the handling of the fluxes by ecosystem type, raised by Referee # 1, and as reflected in Table 4 and in the abstract. The discussion on pages 21-22 doesn't fully address the issue, because it mainly concerns the relationship to other studies or a few specific methodological issues. To my understanding, the posterior partitioning within regions is tied to the priors, and therefore independent of atmospheric data. If this is true it needs to be emphasized. What is especially missing is a discussion of the uncertainties in the partitioning that are intrinsic to the calculation itself. Another way to state this concern is that the fluxes by ecosystem type in Table 4 and the abstract lack error bars. I appreciate the value in showing the complete results, as in Table 4, but I am concerned with the lack of uncertainty analysis, and I am therefore also concerned that the fluxes for individual ecosystem types might be too uncertain to merit being highlighted in the abstract and discussion.

My guess would be that most of the fluxes by ecosystem type are, in fact, very uncertain, with perhaps a few exceptions. This leads to a suggestion, which is to add strong caveats in the text and Table 4, and then to specifically defend those few results which might have merit. The defense needs to be crafted in the context of the method itself (not other studies). Only results which can be defended should appear in the abstract or be discussed at length in the text.

Another minor point is that the dark blue points in Figure 2b,c,d still are not clearly labeled. These panels show four kinds of points, but only three are identified in the legend.

I'm prepared to accept but am not completely happy with two new sentences in the abstract, which I'd like you to revisit:

"These inverted ecosystem-type-related carbon fluxes have large Gaussian uncertainties. This may be because (i) the detailed CO2 flux 20 partitioning in the assimilation system relies on the ecosystem-depended prior model set, and (ii) the available observations for individual ecosystem types in Asia are limited."

What I find odd is the need to speculate ("This may be..") about formal sources of error. Can the two sentences not be joined ".. have large Gaussian uncertainties because the detailed..." Or if this is not correct, please find another solution. Also, it seems to me that a major issue issue involves uncertainty in the priors. While the second sentence does indeed mention the dependency on priors, it does not adequately emphasize their contribution to the uncertainty.