|Review Pirk et al. BG Revised|
The authors did a great job revising their paper, but some new unrealistic statements have been introduced that need to be worked on before acceptance. The change in title is highly appreciated, it better reflects the content of the paper.
The new Fig. 2b really looks appropriate and gives me the adequate impression of the conditions your paper is focusing on.
Double-checking: in the abstract the bias is -46 gC m–2, and the optimzed flux is -82 gC m–2. This means that the conventional method had -82 + (-46) = -128 gC m-2 (OK - number given on p.8, line 14).
Introduction now starts smoothly into the topic.
"One conceivable mechanism is that a greater roughness length could break down larger turbulence into smaller turbulence, thus shifting some of the co-spectrum toward higher frequencies." - I do not agree!
14/8-10: "A greater roughness could break down larger turbulence into smaller turbulence, thus shifting some of the flux co-spectrum toward higher
10 frequencies": this is unrealistic. Turbulence is generated at the low frequency end and then is transported from larger eddies to smaller eddies and on, until turbulence dissipates in the very high frequency. In my view it is physically impossible that "greater roughness could break down larger turbulence into smaller turbulence". There is a continuity in the TKE transport from large eddies to the inertial subrange, no breakdown. Please rephrase!
2/14: "the averaging operation and coordinate rotation commonly applied in EC flux calculations 15 can lead to co-spectral distortions and a loss of flux" - my interpretation of the Finnigan et al. (2003) paper is that they specifically address this topic in tall (forest) vegetation whereas they explicitly state "This failure to close the energy balance is less common close to the surface over short roughness but is still sometimes seen, especially in complex topography." I think it cannot implicitly assumed that what they write about tall vegetation can be transferred to short-statured vegetation, thus I would add to the statement on lines 14/15 that this applies to tall vegetation, but not automatically also to tundra vegetation (I know that this was added due to Lars Kutzbach's review; but it would be good to clarify such important differences between short-statured and tall vegetation). Or has anyone shown that the Finnigan et al. (2003) findings can unanimously be applied to tundra vegetation? In my view it has to do with the fact that over forests we measure in the roughness sublayer, whereas in tundra we measure well above the roughness sublayer.
4/14, 5/24: avoid nested parentheses with citation
5/1: the 10^-1.5 Hz number looks odd. Write "around 0.032 Hz", please. This is actually the "energy-containing range" in micrometerological terminology. Maybe use this terminology as well (if it is what your tool does).
7/7: "which does, however, not mean that 25 s are in general enough to determine a 30 min flux" - why "in general"? Under which cases are 25 s sufficient? Would be interesting to know! If you however agree to remove "in general" then the issue could be solved in this way as well. You basically want to say that an average of multiple 25-second segments would be enough to estimate the flux, but not a single 25-second segment. I am still not 100% confident about this, to be honest.
Signed review: Werner Eugster