A multi-instrument fuzzy logic boundary-layer top detection algorithm

Smith, Elizabeth N.; Carlin, Jacob T.

doi:https://doi.org/10.5194/egusphere-2023-2050

Preprints

https://doi.org/10.5194/egusphere-2023-2050

Preprints

11 Dec 2023

| 11 Dec 2023

A multi-instrument fuzzy logic boundary-layer top detection algorithm

Elizabeth N. Smith and Jacob T. Carlin

Abstract. Understanding the boundary-layer height and its dynamics is crucial for a wide array of applications spanning various fields. Accurate identification of the boundary-layer top contributes to improved air quality predictions, pollutant transport assessments, and enhanced numerical weather prediction through parameterization and assimilation techniques. Despite its significance, defining and observing the boundary-layer top remains challenging. Existing methods of estimating the boundary-layer height encompass radiosonde-based methods, radar-based retrievals, and more. As emerging boundary-layer observation platforms emerge, it is useful to reevaluate the efficacy of existing boundary-layer top detection methods and explore new ones.

This study introduces a fuzzy-logic algorithm that leverages the synergy of multiple remote-sensing boundary-layer profiling instruments. By harnessing the distinct advantages of each sensing platform, the proposed method enables accurate boundary-layer height estimation both during daytime and nocturnal conditions. The algorithm is benchmarked against radiosonde-derived boundary-layer top estimates obtained from balloon launches across diverse locations in Wisconsin, Oklahoma, and Louisiana during summer and fall. The findings reveal notable similarities between the results produced by the proposed fuzzy-logic algorithm and traditional radiosonde-based approaches. However, this study delves into the nuanced differences in their behavior, providing insightful analyses about the underlying causes of the observed discrepancies. The fuzzy-logic boundary-layer top detection algorithm, called BLISS-FL, is released publicly fostering collaboration and advancement within the research community.

Received: 07 Sep 2023 – Discussion started: 11 Dec 2023

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this preprint. The responsibility to include appropriate place names lies with the authors.

Download & links

Preprint (PDF, 10816 KB)

Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
Preprint (10816 KB)

Download & links

The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.

Journal article(s) based on this preprint

11 Jul 2024

A multi-instrument fuzzy logic boundary-layer-top detection algorithm

Elizabeth N. Smith and Jacob T. Carlin

Atmos. Meas. Tech., 17, 4087–4107, https://doi.org/10.5194/amt-17-4087-2024,https://doi.org/10.5194/amt-17-4087-2024, 2024

Short summary

Elizabeth N. Smith and Jacob T. Carlin

Interactive discussion

Status: closed

CC1:
'Radar question - Comment on egusphere-2023-2050', Vagner Castro, 14 Dec 2023

Dear Author,
Your study mentions using radar data. Would you mind sharing some specific details about the system, like the type and the operating frequency?
Thanks,
Vagner

Citation: https://doi.org/10.5194/egusphere-2023-2050-CC1
- CC2: 'Reply on CC1', Jacob Carlin, 15 Dec 2023
  
  Thank you for your comment. The current study does not make direct use of any radar data. However, as discussed in section 3.3, part of the CLAMPS dataset used for validation was collected during a study comparing CLAMPS BL depth estimates using the proposed algorithm and radar Bragg-scatter-based estimates. In that case, we used data from the polarimetric WSR—88D radars that comprise the NEXRAD network in the United States, which operate in the S band between 2.7 and 3.0 GHz (e.g., https://nap.nationalacademies.org/read/10394/chapter/11). The results of that comparison are forthcoming in a follow-up study.
  
  Citation: https://doi.org/10.5194/egusphere-2023-2050-CC2
RC1:
'Comment on egusphere-2023-2050', Anonymous Referee #1, 05 Jan 2024

This paper presents the results of using a fuzzy logic algorithm with a combination of doppler lidar, radiance inferometer and microwave radiometer, with validation done by comparisons with radiosonde data. I feel the paper needs a major revision as the presentation is confusing and the results are not very clear. I realize that there are lots of different cases to present, and many options in the radiosonde comparisons, but in the end, it's not obvious as to how these retrievals might ultimately be used. But I think a rewrite could make it much easier to extract this information. My specific comments are:

1. Both the and abstract should state more clearly what observations are being used for the retrieval and validation (see the first sentence above).
2. Line 110: do you mean the first generation step in B18, or is this changed in the current algorithm?
3. Line 126: Why are buoyant processes important during the night? This seems counter intuitive as buoyancy should grow during the daytime.
4. Line 131: So you mean that the B18 algorithm was not successful in estimating overnight BL height?
5. Section 3: Do radiosonde observations measure thermodynamic BLH rather than the ML height? If so, is it really a good comparison?
6. Line 402: It is fine not to recommend a preferred radiosonde PBLH algorithm, but it would be really helpful to see a plot like
7. Figure 3 is really helpful to see the evolution of the PBLH estimate during the course of a day. It would be even better if you could plot the radiosonde estimates of PBLH on top of this. There is lots of discussion of how the algorithm does at different times of the day, but much of this could be clarified with a comparison like this.
8. The summary section (5) doesn't really have any concrete conclusions. As you say, it's hard to make a fair comparison since a 100 m difference means something very different depending on the time of day. But plotting the height estimates in comparison to radiosondes as a function of time would really help with this.
9. Figure 7: It's not clear what you mean by dark and light colors.
10. Table 3 appears to be the primary result of this work. Would it help to include a percent difference along with the absolute difference?

Citation: https://doi.org/10.5194/egusphere-2023-2050-RC1
- AC1: 'Reply to RC1', Elizabeth Smith, 07 Feb 2024
  
  The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2023/egusphere-2023-2050/egusphere-2023-2050-AC1-supplement.pdf
  
  Citation: https://doi.org/10.5194/egusphere-2023-2050-AC1
RC2:
'Comment on egusphere-2023-2050', Anonymous Referee #2, 06 Jan 2024

Please find the attached for comments.

Citation: https://doi.org/10.5194/egusphere-2023-2050-RC2
- AC2: 'Reply to RC2', Elizabeth Smith, 07 Feb 2024
  
  The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2023/egusphere-2023-2050/egusphere-2023-2050-AC2-supplement.pdf
  
  Citation: https://doi.org/10.5194/egusphere-2023-2050-AC2

Interactive discussion

Status: closed

CC1:
'Radar question - Comment on egusphere-2023-2050', Vagner Castro, 14 Dec 2023

Dear Author,
Your study mentions using radar data. Would you mind sharing some specific details about the system, like the type and the operating frequency?
Thanks,
Vagner

Citation: https://doi.org/10.5194/egusphere-2023-2050-CC1
- CC2: 'Reply on CC1', Jacob Carlin, 15 Dec 2023
  
  Thank you for your comment. The current study does not make direct use of any radar data. However, as discussed in section 3.3, part of the CLAMPS dataset used for validation was collected during a study comparing CLAMPS BL depth estimates using the proposed algorithm and radar Bragg-scatter-based estimates. In that case, we used data from the polarimetric WSR—88D radars that comprise the NEXRAD network in the United States, which operate in the S band between 2.7 and 3.0 GHz (e.g., https://nap.nationalacademies.org/read/10394/chapter/11). The results of that comparison are forthcoming in a follow-up study.
  
  Citation: https://doi.org/10.5194/egusphere-2023-2050-CC2
RC1:
'Comment on egusphere-2023-2050', Anonymous Referee #1, 05 Jan 2024

This paper presents the results of using a fuzzy logic algorithm with a combination of doppler lidar, radiance inferometer and microwave radiometer, with validation done by comparisons with radiosonde data. I feel the paper needs a major revision as the presentation is confusing and the results are not very clear. I realize that there are lots of different cases to present, and many options in the radiosonde comparisons, but in the end, it's not obvious as to how these retrievals might ultimately be used. But I think a rewrite could make it much easier to extract this information. My specific comments are:

1. Both the and abstract should state more clearly what observations are being used for the retrieval and validation (see the first sentence above).
2. Line 110: do you mean the first generation step in B18, or is this changed in the current algorithm?
3. Line 126: Why are buoyant processes important during the night? This seems counter intuitive as buoyancy should grow during the daytime.
4. Line 131: So you mean that the B18 algorithm was not successful in estimating overnight BL height?
5. Section 3: Do radiosonde observations measure thermodynamic BLH rather than the ML height? If so, is it really a good comparison?
6. Line 402: It is fine not to recommend a preferred radiosonde PBLH algorithm, but it would be really helpful to see a plot like
7. Figure 3 is really helpful to see the evolution of the PBLH estimate during the course of a day. It would be even better if you could plot the radiosonde estimates of PBLH on top of this. There is lots of discussion of how the algorithm does at different times of the day, but much of this could be clarified with a comparison like this.
8. The summary section (5) doesn't really have any concrete conclusions. As you say, it's hard to make a fair comparison since a 100 m difference means something very different depending on the time of day. But plotting the height estimates in comparison to radiosondes as a function of time would really help with this.
9. Figure 7: It's not clear what you mean by dark and light colors.
10. Table 3 appears to be the primary result of this work. Would it help to include a percent difference along with the absolute difference?

Citation: https://doi.org/10.5194/egusphere-2023-2050-RC1
- AC1: 'Reply to RC1', Elizabeth Smith, 07 Feb 2024
  
  The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2023/egusphere-2023-2050/egusphere-2023-2050-AC1-supplement.pdf
  
  Citation: https://doi.org/10.5194/egusphere-2023-2050-AC1
RC2:
'Comment on egusphere-2023-2050', Anonymous Referee #2, 06 Jan 2024

Please find the attached for comments.

Citation: https://doi.org/10.5194/egusphere-2023-2050-RC2
- AC2: 'Reply to RC2', Elizabeth Smith, 07 Feb 2024
  
  The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2023/egusphere-2023-2050/egusphere-2023-2050-AC2-supplement.pdf
  
  Citation: https://doi.org/10.5194/egusphere-2023-2050-AC2

Peer review completion

AR: Author's response | RR: Referee report | ED: Editor decision | EF: Editorial file upload

AR by Elizabeth Smith on behalf of the Authors (08 Feb 2024) Author's response Author's tracked changes Manuscript

ED: Referee Nomination & Report Request started (14 Feb 2024) by Laura Bianco

RR by Anonymous Referee #2 (04 Mar 2024)

RR by Anonymous Referee #1 (05 Mar 2024)

ED: Reconsider after major revisions (13 Mar 2024) by Laura Bianco

AR by Elizabeth Smith on behalf of the Authors (10 Apr 2024) Author's response Author's tracked changes Manuscript

ED: Referee Nomination & Report Request started (12 Apr 2024) by Laura Bianco

RR by Anonymous Referee #2 (24 Apr 2024)

Suggestions for revision or reasons for rejection

The authors’ reply clarifies some of my comments. However, there still some confusing statements/definitions/discussions for me as listed below.
Comments:
1. While I concur that leveraging the synergy of multi-remote sensing boundary layer profiling measurements could enhance PBLH estimations by providing more constraining information, I find the authors' claim that 'the proposed method enables accurate boundary-layer height estimation both during daytime and nocturnal conditions' unconvincing based on the comparisons in Figures 8-11. The PBLHs estimated by remote-sensing (Fuzzy-logic) are consistently lower than those derived from radiosondes for PBLHs > 500 m or during afternoon convective boundary layer conditions (Figure 10). Why or what causes the underestimations from Fussy-logic PBLHs?
2. Based on Figure 2, it appears that the second-generation step yields even lower PBLHs than the first-generation step. This suggests that the second-generation step further underestimates PBLHs under conditions of a convective boundary layer. Is that correct?
3. The comparisons between Fuzzy-logic and radiosonde PBLHs do not seem as robust as those shown in previous studies that exclusively used Doppler lidar measurements, such as those in Tucker et al. (2009) and Krishnamurthy et al. (2021). Could you provide the correlation coefficients between Fuzzy-logic PBLHs and Radiosonde PBLHs as depicted in Figures 8, 9, and 11?
References:
Tucker, S. C., C. J. Senff, A. M. Weickmann, W. A. Brewer, R. M. Banta, S. P. Sandberg, D. C. Law, and R. M. Hardesty, 2009: Doppler Lidar Estimation of Mixing Height Using Turbulence, Shear, and Aerosol Profiles. J. Atmos. Oceanic Technol., 26, 673–688, https://doi.org/10.1175/2008JTECHA1157.1.
Krishnamurthy, R., Newsom, R. K., Berg, L. K., Xiao, H., Ma, P.-L., and Turner, D. D.: On the estimation of boundary layer heights: a machine learning approach, Atmos. Meas. Tech., 14, 4403–4424, https://doi.org/10.5194/amt-14-4403-2021, 2021.
4. The authors have noted that in the first-generation step, the temperature inversion height was used as an input. Could you clarify how the temperature inversion height is derived? What degree of temperature difference is considered to constitute an inversion? The term 'temperature inversion height' is introduced in line 139 without an explanation of how it is obtained. In the response, the authors claim, 'We have never stated that inversion height is derived from the first-generation step.' If so, could you specify when the 'inversion height' is derived?
5. Typically, a figure caption should provide a brief explanation of what the figure represents, such as the observations or variables depicted in Figure 2. This would provide readers with basic information, and I don't believe it would be as 'extensive' as the authors suggested in their response. They could relocate the detailed discussions or illustrations of the figure to the main body of the text.
6. Despite several rounds of discussion, the precise definitions of 'buoyancy-driven processes' and 'mechanically-driven processes' remain unclear.
7. The authors replied that ‘if dz = the spacing between any two levels in a profile, and we know that dz varies throughout the profile, 300 is the average of all dz values in the profile. The minimum dz that occurs in the lowest levels of the atmosphere is on the order of 10s of meters. The maximum dz that occurs higher up in the atmosphere is on the order of 100s of meters.’ If a group of numbers have a minimum of 10 and maximum of 100, how can the average to be 300?
8. As for the comparison of high- vs. low-resolution sounding data, the authors state, ‘Analyzing the methods separately highlights that using high-resolution sounding data tends to lead to lower median BL height, but the ranges and interquartile range values do not change much between coarse and high resolution datasets. There is no indication from this analysis that using high resolution data necessarily yields more accurate BL height values’. It appears that the authors are suggesting that because ‘the coarse and high resolution datasets have similar PBLH ranges and interquartile range values’, therefore, ‘There is no indication from this analysis that using high resolution data necessarily yields more accurate BL height values’. I don't believe this is a solid conclusion. For instance, under an extreme condition where the coarse sounding data produce identical PBLHs, resulting in ranges and interquartile range values of zero, it certainly doesn't imply that coarse sounding data provides accurate PBLH estimations.

Referee Report: PDF

Hide

ED: Publish subject to minor revisions (review by editor) (02 May 2024) by Laura Bianco

AR by Elizabeth Smith on behalf of the Authors (06 May 2024) Author's response Author's tracked changes Manuscript

ED: Publish as is (14 May 2024) by Laura Bianco

AR by Elizabeth Smith on behalf of the Authors (24 May 2024) Manuscript

Journal article(s) based on this preprint

11 Jul 2024

A multi-instrument fuzzy logic boundary-layer-top detection algorithm

Elizabeth N. Smith and Jacob T. Carlin

Atmos. Meas. Tech., 17, 4087–4107, https://doi.org/10.5194/amt-17-4087-2024,https://doi.org/10.5194/amt-17-4087-2024, 2024

Short summary

Elizabeth N. Smith and Jacob T. Carlin

Viewed

Total article views: 538 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
372	117	49	538	21	18

HTML: 372
PDF: 117
XML: 49
Total: 538
BibTeX: 21
EndNote: 18

Views and downloads (calculated since 11 Dec 2023)

Month	HTML	PDF	XML	Total
Dec 2023	129	38	13	180
Jan 2024	53	13	2	68
Feb 2024	42	23	7	72
Mar 2024	26	14	1	41
Apr 2024	31	8	4	43
May 2024	24	10	1	35
Jun 2024	57	9	21	87
Jul 2024	10	2	0	12

Cumulative views and downloads (calculated since 11 Dec 2023)

Month	HTML	PDF	XML	Total
Dec 2023	129	38	13	180
Jan 2024	53	13	2	68
Feb 2024	42	23	7	72
Mar 2024	26	14	1	41
Apr 2024	31	8	4	43
May 2024	24	10	1	35
Jun 2024	57	9	21	87
Jul 2024	10	2	0	12

Viewed (geographical distribution)

Total article views: 529 (including HTML, PDF, and XML) Thereof 529 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 11 Jul 2024

Download

The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.

Preprint (10816 KB)
Metadata XML

Short summary

Boundary-layer height observations remain sparse in time and space. In this study we create a new fuzzy-logic method for synergistically combining boundary-layer height estimates from a suite of instruments. These estimates generally compare well to those from radiosondes, plus the approach offers near-continuous estimates through the entire diurnal cycle. Suspected reasons for discrepancies are discussed. The code for the newly presented fuzzy logic method is provided for the community to use.


Total:	0
HTML:	0
PDF:	0
XML:	0