Summary:

This study addresses an important topic related to discrepancies in measurement and modeled surface air temperature heights.  The study provides evidence supporting that the height differences between surface air temperature measurements should be accounted for when evaluating model performances and assimilating data. These contributions will be valuable to publish and account for in future research and operational modeling; however, there are major revisions required prior to this paper being suitable for publication that are addressed below.

Overarching comments/concerns:

The manuscript requires thorough editing by a native English speaker prior to its resubmission. There are common wording errors, citations outside of parenthesis, grammar issues, and awkwardly worded sentences that need to be resolved prior to publication. Examples are provided in the first 5 specific comments below, but this comment applies throughout the manuscript.

Analyses for Section 3.1 are only conducted at 2 sites. This seems lacking and would require a justification of this limitation. Why are the other stations (e.g., from Figure 6) not included in this initial analysis? Even if both 2m and 5m temperature observations are only available in a few locations, it seems this analysis can and should be broadened by: (i) comparing modelled 2m and 5m temperature across a broader spatially continuous domain, and (ii) comparing modelled data with more ground observations, and group results by station height to evaluate the potential discrepancies between simulated T2 and T5, and provide deeper insights on how validating modelled T2 with observed T5, or DA practices, can induce issues. (i) Could be further used to evaluate how discrepancies between modelled T2 and T5 vary with geographic, climate, and vegetation conditions.

It was not clear why results were presented in the order they were presented, and it is not particularly easy to follow. A clear explanation for the paper’s logical flow to start the results section, e.g., focusing on addressing specific science questions, would be very useful.

Specific comments:

L18: “Becken (2010)” citation should be inside parenthesis.

L25: Also, at local & global scales

L28: “were” rather than “are”, and citations in parenthesis.

L34: “high” altitude regions.

L33-36: Awkwardly worded sentences, suggest revising.

L95-96: This is a crucial statement for the paper’s scope and therefore requires citation(s).

Paragraph starting in L53: This paragraph seems to focus on cold biases, but biases reported as positive values. If the bias is a cold bias, then it should be reported as a negative number (i.e., model – obs).

The introduction could also benefit from including the motivation of the snow-albedo feedback. That is, surface air temperature biases can propagate to snowpack biases (e.g., in snow cover) which can have albedo feedbacks due to the high albedo of snow that in turn feedback to and increase the original temperature biases.

Figure 1 should be presented more clearly, (e.g., with (a), (b), (c), etc) labeling to show the flow of the figure.

There are many definitions and abbreviations used throughout the paper. There should be a table in Methods which clearly defines these.

Figure 4: It may be more useful to have OPER and OBS lines on separate panels, and show shading for respective lines to represent temporal variability. Can you provide an explanation for the differences between the measurement heights, particularly why max daily T2 is larger than max daily T5, but T2 is lower than T5 in most other time steps (at CDP); whereas at CLB, T5 is higher at all time steps relative to T2.  Importantly, because only 2 sites are analyzed, and the sites show differences in patterns, how can results be generalizable? Finally, would these discrepancies in diurnal cycles look different for periods of snow cover vs. no snow cover (e.g., winter vs. summer)?

Throughout the paper I recommend using different wording than “guess” which is confusing (e.g., in Figure 6). Guess is also not clearly defined making the results related to this wording difficult to follow.

L363:366: I am not sure if this makes sense, because the guess at 2m is also much lower than the diagnostic analysis and forecast at 2m as well.

Figure 8: it does not seem to make sense that the symbols should be connected with dashed lines. These results may be better presented in a table format than a figure.

Figure 9: pseudo-biases are not clearly defined and therefore it is difficult to make sense of this figure.

Overall, much of the discussion section seems more like additional results sub sections, rather than a true discussion of the authors’ perspectives on the results and insights for future research.

The Conclusions section should be shortened to more concisely highlight the key takeaways and implications. Much of the discussion that is currently in the Conclusions section may be better placed in the Discussion section.

Please make data used for this study publicly available to support reproducibility.

See attached PDF.