Dear Dan Lu et al.,

Thank you for your thoughtful response to the referees comments. Their suggestions will no doubt greatly improve your manuscript, potentially making it suitable for publication in Biogeosciences. It is not clear to me whether your response to the comments of Marko Laine will satisfy their concerns, so I will be asking the reviewers to have a look at your revised manuscript once submitted.

In addition, I am not convinced by your response to comment 4 from referee 2, regarding your claim that DREAM and AM can constrain the majority of parameters in DALEC. I agree with the reviewer that "simply saying the model is simple enough/doesn’t have many parameters is insufficient.", yet in your response you argue that "The DALAC model used in this study is a simple model with considering only six processes and five carbon pools, and all the 21 parameters were shown to be sensitive to the NEE data, despite that some are more sensitive than others (Safta et al, 2015)." But the DALEC model has been shown in multiple studies to be poorly constrained my NEE data alone. See for example the Fox et al. REFLEX paper cited in your manuscript, which also employs the AM approach that you use. And practically all other DALEC papers, of which there are many. The only study to my knowledge that found good constraints was Bloom and Williams, 2015 Biogeosciences, 12, 1299–1315, but they needed to impose additional 'common sense' constraints to do so. You could do a better job of citing the relevant literature regarding DALEC applications, and putting your results in the context of previous studies. You conclude that "Therefore, it is not surprising that both AM and DREAM algorithms can constrain the parameters pretty well." but given that all other studies that use DALEC and EC data alone show poor parameter constraint I find it very surprising indeed. The argument that it is due to low observational uncertainty does not make sense, as previous studies should have used the same NEE uncertainty. Unless of course you have mis-characterized the uncertainty? I really feel you need to get to the bottom of this, as it is a striking difference from most/all previous studies and the implications of such good constraint are large.

Best wishes,
Trevor

Dear authors,

Thank you for your detailed response, which was greatly appreciated by the two referees. Both referees feel the manuscript has very much improved, and suggest publication following minor revisions. Please keep their comments in mind when preparing your manuscript for final submission, particularly Jasper Vrugt's comments regarding implications for homoscedastic NEE errors, and reviewer 2's suggestion to strengthen the abstract by better emphasizing the residual analysis.

Congratulations on an excellent piece of work and best wishes for your future efforts.

Trevor