This manuscript can now be published subject to minor revisions. Reviewer #3 has raised a few minor issues - please try to address them in the revised version. However, no further review will be necessary.

Reviewer #3:

I think the authors have done a good job of addressing the comments of previous referees, and I also think that the paper presents a careful and compelling analysis of GOSAT data, with important sectoral information and comparisons to TCCON. I agree with the authors’ comment that while they might not completely satisfy the concerns of Referee #2, it is useful for the community to continue the discussion in the literature. However, I think that the paper could be more explicit in terms of how it is different from Turner et al., GRL, (2016) and how it addresses the issues raised by Bruhwiler et al., (2017). For example, it should say in the introduction that the Turner et al. paper inferred a trend from top-down emission estimates while this paper is using a statistical analysis of GOSAT methane enhancements. This means that the Bruhwiler et al. concerns about different inversion methods and the influence of the prior in poorly constrained regions would not apply to this analysis, which should be noted in the introduction.

Specific comments:
p. 4 line 29: “at least eight valid retrievals for a given year” What is the sensitivity to this choice?

p. 5 lines 5-6: “Some areas are sparsely sampled, such as California” Please show how under-sampling this region could affect the national trend given the bottom-up inventories (i.e., some estimate of trend sampling error)

Figure 1. Caption does not explain green vs. blue lines.