In this manuscript, Zhang et al. compare the atmosphere-vegetation decoupling factor omega simulated by the land surface model ORCHIDEE with observation-based estimates derived from 106 eddy covariance sites across PFTs. The motivation of this study is valuable as it attempts to evaluate the model performance of simulating an ecosystem property rather than just the simulation of fluxes themselves. This is useful as it allows deeper insights into the functioning of the model compared to more common evaluation approaches. However, I found one critical point that needs to be corrected before the article can be published. In Eq. 3, z should be the reference height (or sensor/measurement height for the eddy covariance towers), not the vegetation height. See e.g. the Monteith and Unsworth 2013 textbook (4th edition, Eq. 17.6) or Liu et al. 2007. To be consistent with the tower observations, z must be set to the height of the sensor at the flux sites in the model. If this is not done a series of biases are introduced which affect the rest of the results. For example, the interpretation of a measured wind speed at a flux site depends on the height where it’s measured (logarithmic increase with height), but in the model the assumption is made that the wind speed was always measured at vegetation height, which is incorrect and which causes biases in Ga that affect different PFTs to a different extent. 

A second major point is that more care must be taken in how ‘observations’ of omega are used in such an evaluation exercise. The omega values here are modelled products as well (observation-based at best) which come with a range of assumptions that will affect its interpretation. For example, the canopy boundary layer conductance in de Kauwe et al. 2017 from Thom et al. 1972 was derived for a bean crop but applied to all PFTs in that study. That will inevitably lead to biases in the flux-derived omega values affecting different PFTs to a different extent. Similar issues arise from a non-closure of the energy balance that lead to negative biases in Gs (probably also different in different biomes/PFTs). The authors point to these issues in the discussion, but it would be most useful to the reader if the consequences of these potential biases on the results were elaborated. Just describing that an issue causes a bias is not very useful. It would be much better to show what factors are likely to have major impacts on the comparison by conducting e.g. sensitivity analyses. Knowing that a model-data mismatch could be caused by either the observations or the model is absolutely crucial for such an analysis. 

Minor comments:

Introduction:

l. 43: add ‘of’ after simulation

l. 45: one cannot add conductances if they are in series, only resistances. Please rephrase.

L. 52: surely there must be more references supporting this statement. 

Methods:

L100: please describe how stomatal conductance (gs) is scaled to the canopy level.

L 115: please give the equation for z0h here, not only the reference. The z0h to z0m ratio is relevant for the canopy boundary layer conductance and thus Ga. Does the equation imply differences in z0h/z0m across PFTs?

L 120: add ‘radiation’ after short-wave down and long-wave down

Where in Su et al. 2001 can Eq. 4 be found?

Section 2.2 is inaccurately named as there is basically no information on the FLUXNET data used. I argue that this section needs to be split up: one part giving more information on the data used (e.g. what sites from what biomes), which could be combined with section 2.3, and  then a separate section for the simulation setup.

L 152ff: I do not understand why the integrated canopy level stomatal conductance was not used? Surely the canopy-integrated value needs to be used in order to be comparable with the flux-derived omega.

L. 158: sites

L. 209: observations

L. 214: compensation

Discussion

L 327: what other factors? Please explain.

L. 370: Limitations

L. 380: I find this point a bit senseless. If conditions with wind speed don’t occur across the data set, why should these conditions matter? They may be of interest from a theoretical point of view but not for LSM evaluation.

L. 389: what are the uncertainties and how would this affect observation-derived omega values?

L. 397: how and under which conditions would this bias the results? This needs to be explained better.

References

De Kauwe MG, Medlyn BE, Knauer J, Williams CA. 2017. Ideas and perspectives: how coupled is the vegetation to the boundary layer? Biogeosciences 14: 4435–4453.

Liu S, Lu L, Mao D, Jia L, others. 2007. Evaluating parameterizations of aerodynamic resistance to heat transfer using field measurements. Hydrology and Earth System Sciences Discussions 11: 769–783.

Thom AS. 1972. Momentum, mass and heat exchange of vegetation. Quarterly Journal of the Royal Meteorological Society 98: 124–134.