A Comparison of Wintertime Atmospheric Boundary Layer Heights Determined by Tethered Balloon Soundings and Lidar at the Site of SACOL

Round 1

Reviewer 1 Report

The authors heavily reviewed the paper, and the quality of the paper seems to be raised to a good level, deserving publication.

Author Response

Thanks to the reviewer for improving the quality of the manuscript. 

Reviewer 2 Report

The authors have adressed the issues I raised and the mansuscript has significantly improved. In my opinion it can be published as is.

In line 401, I have some text leftovers: "...ChinaResubmitted from remotesensing-1091902Resubmitted from remotesensing-1091902."

Please remove them!

Author Response

Thanks to the reviewer for improving the quality of the manuscript. The text leftovers "...ChinaResubmitted from remotesensing-1091902Resubmitted from remotesensing-1091902." have already been removed from the manuscript.

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

The paper deals with the problem of Atmospheric Boundary Layer determination by Rayleigh Lidar, compared to temperature and wind observations, obtained by a tethered balloon. The high number of balloon operations provided a very valuable dataset for comparison. The paper is clear, well-written, and the results are interesting, although some methodological issues at the base of the work should be discussed and solved.

General comments:

The (zenith pointing) aerosol lidar profiles quickly decrease because of both spherical dilution and atmospheric decrease of cross section, due to the decrease of molecular density versus height. The attenuation can be neglected in a first approximation, in particular for IR systems, but also at 532 nm only introduces a smooth transfer function, that can be easily estimated for molecular extinction. The aerosol contribution is more complex to estimate, but for aerosol layers not too opaque (as clouds, for example) can be neglected for boundary layer height estimations. However, methods based on the gradient (opposed to methods based on turbulence, which require higher spatial and temporal resolutions) can lead to systematic errors if the signal is not regularized to be comparable to the Aerosol Backscattering Coefficient (ABC), which is the best product achievable by Rayleigh lidar. This represents the volume cross section of the aerosol hit by the laser beam, and through aerosol models can be converted to mass and other microphysical properties of the aerosol particles. However, retrieving aerosol backscatter coefficient is not straightforward, since it requires accurate calibration of the instrument, operation still performed by a skilled operator, and hence time-consuming and not possible for continuous monitoring. For this reason, the best automated product should be used as a substitute of the ABC.

The Range-Corrected Signal (RCS) is employed by the authors as a proxy of the ABC, and is a typical product automatically obtained from lidar raw data, as explained by the authors. However, it is not the best kind of product, since it still decreases quickly and may mask contributions from higher layers in the atmosphere. In particular, if the gradient method (in any declination: Haar wavelet, fit, numerical derivative) is employed, the contribution of elevated layers may lead to underestimation of the ABL height.  In Angelini and Gobbi (2014) it was demonstrated that the ln(RCS) is the best candidate for this task, since its behavior is quite similar to the ABC.

1. I suggest the authors to try the analysis using the logarithm of the Range Corrected Signal. This can change some determinations, especially under convective conditions, where elevated layers are present, and reduce the underestimation.

As for the ABL determination from balloon profiles, my main concern is about the method of determination from the potential temperature (line 132). The maximum of the gradient of the temperature profile is used as a proxy of the mixing height during convective conditions. Since the standard methods (Holzworth 1964 and Stull) do not employ the gradient, but check for the height where the potential temperature equals that observed at ground (with eventual corrections of overshooting). The method used by the authors resembles more the Heffter method, still not completely. The authors should demonstrate that their method is proven as valid and/or explain why they used this criterion, in particular since they stated that temperature-derived determinations perform better than Richardson bulk method (line 158).

2. My suggestion is either to use the standard methods or cite some references to support their choice. In general, I suggest to take more care in this step, since the conclusions strongly depends on the results of the comparison.

Night-time stable Boundary layer is more complex to determine from temperature profiles, and also wind-driven turbulence should be considered in addition to the top of the inversion layer (line  159). However, also lidar data may suffer from blind zone, so it is not easy to determine whether the night data are reliable or not.

3. I am surprised that in a completely continental site nocturnal ABL heights are about 200 m in average. I suggest to take into account the wind speed, especially for night-time measurements.

 Minor comments:

It is not clear why only the two case studies of December, 12 and 24 are used for a complete diurnal analysis. Please, explain the role of these two days among all measurements.

Line 111: why only observation during ascent are used?

Line 127: Gaussian filter is not probably the best choice: Anisotropic diffusion filters (e.g. the Perona-Malik) perform better, preserving gradients unlike the Gaussian.

Line 139 ignore => ignored

Line 273 ‘makes’ has different font

Reviewer 2 Report

General comments:

The authors extensively use abbreviations, which makes it really hard to follow the thoughts. Please try to avoid that as much as you can.

Furthermore, the authors mix the naming of specific quantities, e.g.  sometimes normalized range-corrected backscatter signal  (NRB) is used, sometimes only backscatter etc..

Please unify and use only one wording for one quantity.

Furthermore, colors in the plots are changing, e.g. sometimes backscatter is grey, sometimes black. Please homogenize for the same quantities and try to make the Figures self-explaining.

Furthermore, the methodology section need to be extended, as some of the used methodologies (see specific comment) are not introduced, like e.g. the characterization of the different boundary layer types.

In general, lidar-based HBL estimates have been validated by many other studies  the authors mention by themselves. However, the manuscript is an appropriate extension of these other works. Some conclusions are not supported by data and cannot be generalized due to the limited time period and the specific localization of the campaign.

Specific comments:

Line 35: “dramatic” is vague

Line 49: I really miss the citation of DOI: http://dx.doi.org/10.7149/OPA.47.2.123 applying the wavelet covariance method for the ABL in Colombia.

Line 60: Link to [37-42] is somehow broken

Line 65: high resolution in terms of?

Line 111: ~30 m min-1: “-1” has to be in superscript to my opinion.

Line 112: Why only ascent data were chosen?

Line 133: Why you did not take θ as symbol for potential temperature?

Line 140: Justify why u_s and v_s were set to zero. What uncertainty does arise from this assumption. I think the assumption is not valid, in case you have strong surface winds.

Figure 2: Showing comparison of Rb and Pt in a,b) and Rb and T in c,d) is confusing. Please unify showing T and Pt for both profiles. And clearly indicate the two different times in the plot.

Line 149: Any ideas why it was lower than inversion layer height? In general, to my knowledge, the Rb is not applicable during thermodynamically stable conditions. Please state on this problem and give reference.

General comment on Section 2.2: Please discuss if it is possible to say that the assumptions of u_s=v_s=0 leads to underestimation of BLH in shallow cases. To me it seems, with probable larger v_s and u_s at night, the dominator in Eq. (4) gets smaller and hence Rb too small.

Line 182: Please give insights on why “a” was set to 60 and 300 m, respectively. It appears a bit out of the blue.

Line 204: Specify how you determined RMLs and RCILs and so on. This was never explained in the manuscript. The results are interesting but without explanation how these different categories of boundary layers are defined/classified is cannot be shown. Thus, please add a respective section in the methodology part.  

Fig. 5 and so on. Please explain abbreviations in Caption.

Fig. 6 Please indicate time period in caption.

Line 251: Not only the aerosol but also the state of humidification of the aerosol. Larger humidification leads to stronger backscatter signal. According to Figure S1, specific humidity was large during night time. Please discuss.

Line 328 and 329: I do not get the bias estimates. Please specify the content of the brackets. How did you determine it? And, according to the low number of observations and the moderate scatter, can you really speak of a bias?

Also, in Fig. 11 and Fig. 12 you show different estimates of a for your WCT. However, you have not discussed that why and for what.

Comments on Conclusion: Line 417 to line 421. For me there was no evidence shown or at least it was not clearly shown that wind led to the underestimation. The reason is the aerosol layering and not the wind. Similar finding were already concluded earlier, e.g. in https://acp.copernicus.org/articles/8/7281/2008/ and other publications. Thus, please do discuss this at least.

Figure 12 and 11: I think you mean TBS-BLH instead of TS-BLH in your legend. Please provide also profiles of the potential temperature. Gradient in Fig. 11c) looks largest at 470m to me.

Line 425 to 426: Please rephrase last sentence, i.e. stop after: “..using simultaneous sounding data. And the omit the “because..” part.