Detection of Upper and Lower Planetary-Boundary Layer Curves and Estimation of Their Heights from Ceilometer Observations under All-Weather Conditions: Case of Athens, Greece
Round 1
Reviewer 1 Report
- Overall assessment: This manuscript developed a new algorithm to retrieve boundary layer height (BLH) by using ceilometer data under all-weather conditions in Athens Greece. The work is important and can promote the development of the scientific community. However, the discussions need some improvements as commented below. Therefore, I recommend a major revision for current version.
- I don’t think ‘matlab’ can be a keyword for an academic paper. The authors developed a new algorithm (idea) for PBLH detection. This is important and meaningful, but matlab is not. Therefore, I suggest the authors remove the word ‘matlab’ throughout this paper.
- Line 247-249: ‘where q is the mixing ratio (kg kg-1 or g kg-1); qw is the mixing ratio of water vapour and ql the mixing ratio of liquid water; q is defined as the ratio of the mass of water vapour (or liquid water), mv (kg or g), to the mass of dry air, md (kg).’…The symbol q, mv, and md cannot find in formula (2).
- Section 2.4 and 2.5: The authors declare that their algorithm is superior to other traditional methods. But detailed comparisons of different methods are missed in the two sections.
- Figure 5. The figure is not clear.
- Figure 6, Figure 7. The two figures are very ugly. The authors missed the x/y-label and colorbar-label.
- Line435:’the form: (a)PBLHCL31=0.98 PBLH+44 33, R2=0.83…’ what’s the meaning of ‘44 33’?
- Some figures can be merged to reduce the figure numbers. Such as, figure 11(a,b) and figure 12(a,b)
- Most of the equations in this paper are unclear. I suggest the authors revised these equations.
Author Response
Please see the attachment.
Reviewer 2 Report
Review of the manuscript "Detection of Upper and Lower Planetary-Boundary Layer Curves and Estimation of their Heights from Ceilometer Observations under All-weather Conditions: Case of Athens, Greece", by Kambezidis et al.
The atmospheric boundary layer is an important factor in the earth surface - atmosphere interaction and its study over urban areas is of increasing interest, especially in relation to air pollution. For that reason, such work is undoubtedly of interest to the scientific community. Unfortunately, the manuscript gives the impression of being prepared hastily without sufficient attention to detail and superficial analysis. Therefore, I believe that it deserves to be published only after serious corrections.
Major comments:
The authors use the terms UPBL and LPBL, which are not popular in the study of the atmospheric boundary layer and need in-depth justification. According to the manuscript, the UPBL is composed of CBL and RL, but in the conclusion UPBL was replaced by ML, which I find incorrect (I admit the interchangeability of CBL with ML, but the replacement of RL with ML is disputable).
Several instruments (radiosondes, passive and active remote sensing instruments) for PBLH determination are listed (lines 45-49), but more details on their individual advantages and disadvantages as well as instruments synergy would be useful to the readers and [1-4] can be cited here.
The authors refer to ALC water vapor sensitivity (lines 71-72) however additional ALC drawback in MLH tracking - layer selection during evening transition (CBL to NBL) need to be mentioned here and e.g. [5] should be cited.
The authors refer to various MLH retrieval algorithms (lines 203-204) but graph based methods should be also cited here e.g. [6]
It will be positively evaluated by potential readers if MLH retrieval methods listed on line 240 are described in more detail.
The authors claim (lines 269-274) that the Holzworth method is used to calculate MLH at night (02:00 LST), but as far as I know the Holzworth method is relevant for CBL height (MLH) estimation. According to the method, MLH is estimated by extending a dry adiabat from surface temperature to its intersection with the radiosonde temperature profile. Such crossing point is impossible in your case (fig.2). Please explain.
The authors should explain lines 277-280, where you say “Lower layers are detected at 815 m agl (decrease in q) and at 907 m agl (increase in WS)”, however according line 57 "…SBL is capped by surface inversion.." which is about 100m in that case, additionally wind has local maximum (according line 119 it is evidence of LLJ on top of NBL) at about 200-250m, so I cannot understand why 815/907m are chosen. Please explain.
On lines 287-288 (related to radiosonde profiles at 14:00LST – fig.3) is written “lPBL height is found at 225 m agl (240 m asl) as the first altitude where a decrease in WS occurs”, however according lines 115-117 "ceilometer also “sees” the lower part of the ABL that includes the NBL with an STI on top, and low-level jets (LLJs) during nighttime", so I cannot agree with NBL existence in the early afternoon. Please explain.
As the backscatter signal is the only data the authors use in routines A,B,C, they need to explain how it is related to temperature inversions (lines 332-333)?
It is not clear whether all estimates or only accepted ones have been used in the calculation of the correlation coefficient (lines 444-445) and R2(fig.8). I also recommend to include additional accuracy metrics such as mean error and RMSE
The authors’ argument (lines 550-554) for high uPBL in winter due to sunny weather and high maximum temperatures seems slightly unconvincing as the summer maximum temperatures in Athens are at least 10 degrees higher (https://en.climate-data.org/europe/greece/athens/athens-7/).
According to lines 41-42: "Above the NBL the residual layer (RL) exists, which is the left-over of the previous day's ML" also according to lines 113-115 "... a ceilometer" sees "the upper part of the PBL that consists of a CI on top of the RL during night time and the CBL (or ML) during daytime" therefore uPBL is composed of ML and RL (the latter is ML from the day before). So, I wonder why uPBL should show clear diurnal variations (fig.15).
The authors explain observed maximum of uPBL at7:00-8:00 LST in winter (lines 576-578, see fig.16a) by sunny days, but as I already wrote winter maximum of uPBL due to fair-weather and sunny days sounds implausible, especially early in the morning, because in Athens the sun rises around 7:30 in winter (https://www.timeanddate.com/sun/greece/athens?month=1&year=2009).
Minor comments:
Lines 52-54 (“In fair weather, the PBL height has a well-defined structure and a diurnal cycle [12]; during daytime (between sunrise and sunset) heating”). To be more precise, surface heating, which leads to the ML formation, occurs when the net radiation is positive, which begins shortly after sunrise and ends before sunset.
Line 70 (“… can detect 3 cloud-base heights for low, mid, and high clouds”) It should be changed to "up to 3 layers", as if the lowest clouds are optically thick, they will obscure (due to complete signal attenuation) the upper clouds.
Line 76 (“Several ceilometers are now in 24-h-basis operation…”) According EUMETNET (https://www.eumetnet.eu/activities/observations-programme/current-activities/e-profile/alc-network/) "...280 stations from 22 countries are currently available…"
I do (and potential readers will) wonder why authors have decided to reduce available data to about 20% (6day per month) (lines 187-193), the argument "... enormous workload” in file converting is unconvincing - such tasks can easily be automated on computer. Are there any problems in the rest of the days?
Eq 11 I slightly confusing. What does d2BSC/d(z-1)2 mean, variable substitution (z to z-1), or second derivative of BSC at height = z – 1 ?
I wonder why the sum (90%+60%) is more than 100% (lines 507-509)?
Line charts (fig.12) should be avoided when data are missing, as it can be misinterpreted because the gaps are filled by interpolation.
According line 528 (“…PBLH values in February are artificial (derived from linear interpolation between the values of January/March)…”), but according fig. 10 there are measurements in February, why they are missing here ?
Technical comments:
I find the verbose description of file formats transformation and Excel file structure (lines 169-180) unnecessary and recommend shorten it.
Equation 3-8, 11-14 are blurred.
Line 254 ?? should be θ (theta)
Line 262 (“…unstable atmospheric conditions (θ < 0 …”). The potential temperature is always positive (see the definition – eq.1), the atmosphere is unstable if dθ/dz < 0
Line 266 (“…stable atmospheric conditions (θ > 0…”) see the previous comment.
Routines E (line 345) and F(line 362) have almost identical names
In my opinion lines 380-381 are redundant.
Although one can assume what the “…overall accuracy...” (line 438) is (ratio of accepted estimates to all estimates), the latter should be clearly defined.
Do you mean uPBL, not PBLH (line 476)
Duplicated paragraph (lines 495-502), see also lines 511-518
Lines 542-543 are not related to fig.13.
Line 581 “… solar radiation values are absent ...”. Your model does not rely on solar radiation, you probably meant lPBL
Reference:
- Liu, B.; Guo, J.; Gong, W.; Shi, Y.; Jin, S. Boundary Layer Height as Estimated from Radar Wind Profilers in Four Cities in China: Relative Contributions from Aerosols and Surface Features. Remote Sens. 2020, 12, 1657. https://doi.org/10.3390/rs12101657
- Yang, Y.; Fan, S.; Wang, L.; Gao, Z.; Zhang, Y.; Zou, H.; Miao, S.; Li, Y.; Huang, M.; Yim, S.H.L.; Lolli, S. Diurnal Evolution of the Wintertime Boundary Layer in Urban Beijing, China: Insights from Doppler Lidar and a 325-m Meteorological Tower. Remote Sens. 2020, 12, 3935. https://doi.org/10.3390/rs12233935
- Wang, D.; Stachlewska, I.S.; Song, X.; Heese, B.; Nemuc, A. Variability of the Boundary Layer Over an Urban Continental Site Based on 10 Years of Active Remote Sensing Observations in Warsaw. Remote Sens. 2020, 12, 340. https://doi.org/10.3390/rs12020340
- Saeed, U., , Rocadenbosch, F., Crewell, "Performance test of the synergetic use of simulated lidar and microwave radiometer observations for mixing-layer height detection", Proc. SPIE 9640, Remote Sensing of Clouds and the Atmosphere XX, 964008 (16 October 2015); https://doi.org/10.1117/12.2195091
- Danchovski, V. Summertime Urban Mixing Layer Height over Sofia, Bulgaria. Atmosphere 2019, 10, 36. https://doi.org/10.3390/atmos10010036
- Bruine, M. D., Apituley, A., Donovan, D. P., Klein Baltink, H., & Haij, M. J. D. (2017). Pathfinder: applying graph theory to consistent tracking of daytime mixed layer height with backscatter lidar. Atmospheric Measurement Techniques 2017, 10(5), 1893-1909. https://doi.org/10.5194/amt-10-1893-2017
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 3 Report
Review of the manuscript "Detection of Upper and Lower Planetary-Boundary Layer Curves and Estimation of their Heights from Ceilometer Observations under All-weather Conditions: Case of Athens, Greece" by H. Kambezidis et al. submitted to Remote Sensing
The manuscript aims at long-term studies and characterization of the planetary boundary layer (PBL) over Athens based on the available ceilometer monitoring data. The derivation of the PBL height is based on a new algorithm developed by the authors. I find the research performed interesting and relevant for publication in Remote sensing, however, its presentation should be considerably improved. In my opinion, the most important shortcomings of the paper are that its goals are not clearly stated and that the conclusions should be to the point and drawn from the work presented in this paper. The methodology and results related sections should be more concise and better organized. I recommend a major revision of the manuscript.
Comments on the paper structure and sections
Introduction and title
After presenting the state of the art, the authors should clearly set the aim of the paper, which is not the case now. After reading the manuscript I would say its mail goals are “long term studies of atmospheric boundary layer properties and dynamics in Athens based on ceilometer data” and “development and validation of an improved algorithm for the extraction of PBL height from ceilometer data”. The manuscript title should reflect these goals as well. I also suggest removing the sentence explaining the organization of the paper as this is of a standard research paper.
Materials and methods
As the authors use well documented standard instrumentation to obtain their data, I believe that Table 1 can safely be omitted. The authors should however add here the exact time interval of the study and duty time of the instrument (%) during this period.
I understand that the authors handle data acquisition and processing semi-manually and the procedure should be presented, however, technical details about file extensions and Excel spreadsheets are neither needed nor relevant here. The authors should make this section more procedure oriented and more general, as other researchers using different software tools benefit nothing from this. I guess the result of semi-manual data processing and the use of “office” tools which are not intended for manipulation of large quantities of data is also the reason why 80% of the data is discarded and only 6 days per month are used for further analysis. This is a serious shortcoming of the work performed.
The same is true for the new PBL retrieval method, which should be presented in general terms (although the authors implemented it in Matlab, others may use different platforms).
The authors use local radiosonde data from Athens, provided by HNMS. This data is however downloaded from a US site and not retrieved locally. Why? This is in particular a problem as the radiosonde data (presented in Figure 3) has a very poor spatial resolution, which may be due to the storage of sparsified data in the U. of Wyoming repositories (the sonde provides much better data). I recommend using local high resolution radiosonde data.
Results
In addition to the validation results of the new algorithm (the authors should present the uncertainties of the PBL retrieval method itself), the authors do present the long term monitoring results, which are interesting, but the presentation needs improving, in particular the (many) trend plots. I would recommend only those, that directly support the conclusions drawn on the analyses performed (which are presented in the Conclusions section). In particular, large error bars in all these plots need to be explained (I assume they originate from PBL variability).
Conclusions
The current “Discussion and conclusions” section is (to a large part) an overview of the paper, I suggest the conclusions to be rewritten so that it is evident what the actual conclusions are and what is the added value of the research performed. This needs to be fixed.
Comments on the paper formatting
Figures
The figures in a manuscript should be used as references to confirm or emphasize a certain statement, and not the other way around, where the main manuscript text is explaining them. Including a number of different profiles with huge error bars is not a good idea (Figure 16) as it is hard to read such plots. What is the source of these error bars? Retrieval errors? Variability of atmospheric conditions? As it is now, the plots do not tell us much.
Citations
The authors are using a mix of two citation styles in the manuscript, namely the style where references are cited according to author names and AIP style, which is generally used for physics, where the numbered references are in brackets. A standard style should be used and I recommend the later (numbered) one. There is no need to reiterate authors’ names in the text.
Some general issues and typos
Line 15: plays important role → plays an important role
Line 21 and other: asl → define the abbreviation → above sea level (a.s.l.)
Line 31: no need to introduce ABL, it only further complicates things
Line 46: remote-sounding → remote sensing (only SODAR uses sound)
Line 78 and other: the many hyperlinks in the manuscript should be moved to Refences and cited. The last access date should be stated, as is customary for URL.
Line 115: do not use the ABL abbeviation, use PBL instead
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
I don't have any other comments or suggestions
Author Response
We thank Reviewer 1 for his/her constructive comments with which the manuscript has improved drastically.
Reviewer 2 Report
It seems that the authors have made great efforts to improve the quality of their manuscript and I appreciate that. My comments and remarks have been answered, so I think the work is now publishable, but please correct the following two lines:
Line 215 “graphical approach” should be replaced by “graph-based approach”
Line 675 “graphical method” should be replaced by “graph-based method”
Author Response
We thank Reviewer 2 for his/her constructive comments, which improved tha quality of the paper drastically. In the revised version of the manuscript (round 2) the two minor comments have been taken into account and corrected.
Reviewer 3 Report
Review of the revised manuscript "Detection of Upper and Lower Planetary-Boundary Layer Curves and Estimation of their Heights from Ceilometer Observations under All-weather Conditions: Case of Athens, Greece" by H. Kambezidis et al. submitted to Remote Sensing
The revised manuscript and the cover letter satisfactorily addresses the concerns I raised and implements my suggestions, therefore I recommend the publication of this work.
Despite the arguments given by the authors in their cover letter regarding their project funding related time constraints and difficulties with more advanced data manipulation, I still maintain that the authors’ work would greatly benefit from a more automated and less Windows reliant manual approach, which should definitely be implemented in the future. I would also recommend reaching some agreement with HNMS (which is practically the authors’ next door neighbor) for high resolution radiosonde profiles.
Related to this, the authors show the correlation between PBLH from CL31 measurements and PBLH from radiosounding in Fig. 8. I do understand the correlation is as high as 0.91, but the best fit line y=kx+n is in both cases slower than y=x (k is less than 1). Why? Is there any systematic reason why the PBLH is underestimated from CL31 data (or overestimated from RDS)? Were measurement errors taken into account in this plot and in particular in the fitting (there is uncertainty both in the height of each data point as well as the retrieved PBLH value)? Please explain this in the figure caption and the manuscript.
General issues
New text lines 383-384:
… of the PBL as upper NOUN MISSING implies the combination of the RL and CBL curves in a 24-h period; in the same way, the characterisation of the PBL as lower NOUN MISSING includes the SL and NBL curves in the period of a whole day. The uPBL and lPBL notations are, therefore, introduced for the first time in the literature to depict the above-mentioned situations as shown in an artistic manner in Fig. 5 below.
Suggestion
The characterization of the upper boundary of the PBL implies the combination of the RL and CBL curves in a 24-h period; in the same way, the characterisation of the lower PBL boundary includes the SL and NBL curves in the period of a whole day. The uPBL and lPBL notations are, therefore, introduced for the first time in the literature to depict the above-mentioned situation, presented schematically in Fig. 5.
Author Response
We thank Reviewer 3 for his/her constructive and extensive comments, which resulted in an improved quality of the paper. As far as the new comments are concerned, we understand the concern of the Reviewer in relation to high-resolution RDS from HNMS. That was our concern from first place, but in the limited time of the project not many moves were possible. We anticipate that in a future paper this will certainly be taken into account.
Now as far as the second comment is concerned, we have added an explanation in lines 450-458 (green text with red characters).
Also, the third issue has been confronted (corrected, lines within the description of Routine F; green text with red characters).
