Assessment of AIRS Version 7 Temperature Profiles and Low-Level Inversions with GRUAN Radiosonde Observations in the Arctic

Round 1

Reviewer 1 Report

“Assessment of AIRS version 7 temperature profiles and low-level inversions with GRUAN radiosonde observations in the Arctic” by L. Zhang et al provides a comparison of AIRS v7 against the older v6 product with respect to radiosonde data from three separate high-latitude locations. Results suggest that the newer AIRS v7 is generally improved with a reduced mean temperature bias across the atmospheric profile and warmer troposphere, but larger cold season bias in the boundary layer relative to v6. 

 

The paper is interesting, timely, and well-written. Most of my comments are relatively minor in nature with suggestions involving some light editing. To assist the reader, two suggestions include: 1) adding a table to the discussion or conclusions section where AIRS v6 and v7 results are summarized by their performance (e.g., a check mark under which version performs better in the overall atmospheric profile, stratosphere, upper troposphere, boundary layer, etc) would be a useful guide to future AIRS users, and 2) labeling all paneled figures and referencing them in the text accordingly (e.g., Figure 6a, Figure 6b, etc). My remarks are listed below by line number (L) are presented in the submitted manuscript.

 

Minor remarks:

 

L12: remove “high quality”

 

L67: please clarify what is meant by “in scenes”? 

 

L150-151: would suggest deleting this sentence as it does not add anything to the site descriptions and is confusing given the stratospheric vortex can shift position or split and lie over any of the GRUAN stations.

 

Figure 3: It would be helpful to clarify in the caption the time period of these comparisons (since 2002)?

 

Figure 5: Did you test for statistical significance in the correlation calculations and if so, at what coefficient threshold was significance obtained? While binning to monthly scale yields a small sample size, it would be a good idea to mention these details if possible in the caption and text. 

 

L329-337: I find this paragraph hard to follow. Namely how the LLI percentage (mis)matches are derived, i.e., “65% (88%) are over snow” – which is referencing what figure and sites specifically? Adding specifics would be helpful.

 

Figure 7: Is the bottom right panel (LLI Depth) supposed to have “fit” statistics similar to the adjacent LLI Intensity plot? Labeling these and all paneled figures in the paper would make it easier to follow the text.

 

L405: What is meant by “vertical bias oscillations”?

Author Response

 

Author Response File: Author Response.docx

Reviewer 2 Report

This paper compares v.6 and v.7 AIRS temperature retrievals to radiosonde profiles from three high-latitude stations.  The authors are well versed in the Arctic climate.  The paper is well motivated and well organized.  The introduction is good.  AIRS contributes to the climate record and therefore the coherence between versions is important.  The main subject is the Low-Level Inversion (LLI) and the skill of the two versions to observe and characterize Arctic LLIs.  

 

Some constructive comments to consider for the final version:

 

1. Line 165 says “plenty”, which is vague and nonspecific.  This should be replaced by a quantitative statement (perhaps found in the numbers in lines 216-221).
2. The application of the AIRS weighting functions to the radiosonde profiles is appropriate to compare with AIRS retrievals.  I think that the discussion in section 2.3 (lines 183-200) could be improved to make the process clearer.  Conceptually, I understand what the authors are doing but I would not be able to reproduce their results from this description.  Some figures could be useful.
3. It is unfortunate that there are only three stations.  This limits spatial sampling and since the satellite is in a synchronous orbit, it also affects temporal sampling.     It would be interesting to know if there are time-of-day sampling biases for each station and how this might affect the bias statistics.
4. Figure 3 shows that there is a large difference in biases between stations in the lower troposphere.  It would be good to have some notion of the significance of the means in each box (i.e., related to number of samples).
5. The control for LLIs should be independent of the AIRS weighting functions.  The LLI state should be determined from the unweighted radiosonde profiles.  It’s OK to use the weightings for comparing versions.  From line 314, I understand that the weighting function has been applied.
6. The term “kernelled radiosonde” is not a good terminology.  Try to find another.
7. This paper is good motivation to improve surface treatments in the AIRS product algorithm.  

Author Response

 

Author Response File: Author Response.docx