the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 14 Nov 2022
Data Quality Enhancement for Atmospheric Chemistry Field Experiments via Sequential Monte Carlo Filters
Abstract. In this study we explore the applications and limitations of Sequential Monte Carlo filters (SMC) to atmospheric chemistry field experiments. The proposed algorithm is simple, fast, versatile and returns a complete probability distribution. It combines information from measurements with known system dynamics to decrease the uncertainty of measured variables. The method shows high potential to increase data coverage, precision and even possibilities to infer unmeasured variables. We extend the original SMC algorithm with an activity variable that gates the proposed reactions. This extension makes the algorithm more robust when dynamical processes not considered in the calculation dominate and the information provided via measurements is limited. The activity variable also provides a quantitative measure of the dominant processes. Free parameters of the algorithm and their effect on the SMC result are analyzed. The algorithm reacts very sensitively to the estimated speed of stochastic variation. We provide a scheme to choose this value appropriately. In a simulation study O3, NO, NO2 and jNO2 are tested for interpolation and de-noising using measurement data of a field campaign. Generally, the SMC method performs well under most conditions, with some dependence on the particular variable being analyzed.
-
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
(2673 KB)
-
Supplement
(6783 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2673 KB) - Metadata XML
-
Supplement
(6783 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2022-1080', Anonymous Referee #2, 05 Dec 2022
This paper applies SMC to the optimization of atmospheric measurement data, finds that SMC can better improve the quality of measurement data, and also discusses the limitations of the algorithm application. The analysis is relatively complete and has good application value.
Questions for this article include:
1.Why is the nighttime PSS value in Fig4 so abnormal? Is it the problem of box model simulation?
2.Why is the ensemble mean value of SMC so different from the PPS calculation at 12 o'clock in Fig6?
Citation: https://doi.org/10.5194/egusphere-2022-1080-RC1 - AC1: 'Reply on RC1', Lenard Röder, 19 Jan 2023
-
RC2: 'Comment on egusphere-2022-1080', Anonymous Referee #1, 27 Dec 2022
-
AC2: 'Reply on RC2', Lenard Röder, 19 Jan 2023
Dear Anonymous Referee #1,
we thank you very much for your time and effort spent to provide your comments.
- My major concern is the robustness of SMC algorithm when applied to a more complicated system. The study only uses one experiment with 5 dimensions (O3, NO, NO2, JNO2, and gate variable). How does the performance of this algorithm change as the dimension of the system increases?
We originally thought about including increased dimension sizes. Unfortunately, a detailed analysis as shown in the article for the 5-dimensional case was not possible for an increased dimension size for any of the datasets available. All considered data sets featured limitations in data coverage and quality that limit the significance of this analysis when the number of dimensions is increased.
The system NO, NO2, O3, j_NO2 is particularly simple and is not very sensitive to other trace gases in such a tropospheric environment. However, increasing the dimensionality to include variables like HOx, VOCs or peroxides quickly introduces many reactions. Applying SMC to such systems may yield great chemical insights, similar to constrained box-model studies. However, this would shift the scope of the study from a measurement technique perspective (AMT) to application and chemical analysis (ACP), as no “ground truth” benchmark can be provided.
Our conclusion is that the SMC needs many studies of different chemical systems in the future to fully rate its potential. This study provides a first step by testing limitations on a relatively limited dataset.
We added the following sentences to the conclusion regarding the system dimension (L 354ff):An open question is the stability of the algorithm when applied to a more complicated system with higher dimension. Repeating the technical procedure of this study using a higher-dimensional system is restricted by data coverage and data quality in existing datasets. We suggest many applications of this method for different chemical systems are necessary in the future to fully rate the potential of the SMC in the analysis of atmospheric chemistry field experimental data.
- Introducing activity variable as a regularization term prevent the system from mode collapse. However, it is unclear how p(η) is chosen even though it is discussed in section 4.4.
The term pη corresponds to a value between 0 and 1 used for a binomial experiment. The value corresponds to the probability that the state is switched from active to passive or vice versa. As mentioned in Algorithm 1 or Table S1, the value is chosen to be 2.5% during the study. This value seemed a reasonable choice after applying the considerations discussed in section 4.4.
- Line 285-290: Please specify that the results of χ² is shown in Figure S3.
Thank you for this hint, the text might be unclear without this information. We added a sentence (L285).
A similar plot showing the resulting values of χ² is shown in the supplement, Figure S3. The value of χ² of the photolysis frequency decreases at the beginning, […]
- Line 326: Figure A8 should be Figure S8
Thank you for your note, we further changed all mentions of “appendix” to “supplement” accordingly. (Lines 328 and 332)
Citation: https://doi.org/10.5194/egusphere-2022-1080-AC2
-
AC2: 'Reply on RC2', Lenard Röder, 19 Jan 2023
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2022-1080', Anonymous Referee #2, 05 Dec 2022
This paper applies SMC to the optimization of atmospheric measurement data, finds that SMC can better improve the quality of measurement data, and also discusses the limitations of the algorithm application. The analysis is relatively complete and has good application value.
Questions for this article include:
1.Why is the nighttime PSS value in Fig4 so abnormal? Is it the problem of box model simulation?
2.Why is the ensemble mean value of SMC so different from the PPS calculation at 12 o'clock in Fig6?
Citation: https://doi.org/10.5194/egusphere-2022-1080-RC1 - AC1: 'Reply on RC1', Lenard Röder, 19 Jan 2023
-
RC2: 'Comment on egusphere-2022-1080', Anonymous Referee #1, 27 Dec 2022
-
AC2: 'Reply on RC2', Lenard Röder, 19 Jan 2023
Dear Anonymous Referee #1,
we thank you very much for your time and effort spent to provide your comments.
- My major concern is the robustness of SMC algorithm when applied to a more complicated system. The study only uses one experiment with 5 dimensions (O3, NO, NO2, JNO2, and gate variable). How does the performance of this algorithm change as the dimension of the system increases?
We originally thought about including increased dimension sizes. Unfortunately, a detailed analysis as shown in the article for the 5-dimensional case was not possible for an increased dimension size for any of the datasets available. All considered data sets featured limitations in data coverage and quality that limit the significance of this analysis when the number of dimensions is increased.
The system NO, NO2, O3, j_NO2 is particularly simple and is not very sensitive to other trace gases in such a tropospheric environment. However, increasing the dimensionality to include variables like HOx, VOCs or peroxides quickly introduces many reactions. Applying SMC to such systems may yield great chemical insights, similar to constrained box-model studies. However, this would shift the scope of the study from a measurement technique perspective (AMT) to application and chemical analysis (ACP), as no “ground truth” benchmark can be provided.
Our conclusion is that the SMC needs many studies of different chemical systems in the future to fully rate its potential. This study provides a first step by testing limitations on a relatively limited dataset.
We added the following sentences to the conclusion regarding the system dimension (L 354ff):An open question is the stability of the algorithm when applied to a more complicated system with higher dimension. Repeating the technical procedure of this study using a higher-dimensional system is restricted by data coverage and data quality in existing datasets. We suggest many applications of this method for different chemical systems are necessary in the future to fully rate the potential of the SMC in the analysis of atmospheric chemistry field experimental data.
- Introducing activity variable as a regularization term prevent the system from mode collapse. However, it is unclear how p(η) is chosen even though it is discussed in section 4.4.
The term pη corresponds to a value between 0 and 1 used for a binomial experiment. The value corresponds to the probability that the state is switched from active to passive or vice versa. As mentioned in Algorithm 1 or Table S1, the value is chosen to be 2.5% during the study. This value seemed a reasonable choice after applying the considerations discussed in section 4.4.
- Line 285-290: Please specify that the results of χ² is shown in Figure S3.
Thank you for this hint, the text might be unclear without this information. We added a sentence (L285).
A similar plot showing the resulting values of χ² is shown in the supplement, Figure S3. The value of χ² of the photolysis frequency decreases at the beginning, […]
- Line 326: Figure A8 should be Figure S8
Thank you for your note, we further changed all mentions of “appendix” to “supplement” accordingly. (Lines 328 and 332)
Citation: https://doi.org/10.5194/egusphere-2022-1080-AC2
-
AC2: 'Reply on RC2', Lenard Röder, 19 Jan 2023
Peer review completion
Journal article(s) based on this preprint
Data sets
Data from TO2021 campaign Crowley, J. N., Dewald, P., Nussbaumer, C. M., Ringsdorf, A., Edtbauer, A., Schuladen, J., Fischer, H., Williams, J., Röder, L., and Hamryszczak, Z. https://keeper.mpdl.mpg.de/d/f12c1d71d4734a89a6ef/
Model code and software
Sequential Monte Carlo Filters for Chemical Box-Models Lenard L. Röder https://github.com/lenroed/smc-boxmodel
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 185 | 85 | 11 | 281 | 33 | 5 | 6 |
- HTML: 185
- PDF: 85
- XML: 11
- Total: 281
- Supplement: 33
- BibTeX: 5
- EndNote: 6
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Patrick Dewald
Clara M. Nussbaumer
Jan Schuladen
John N. Crowley
Jos Lelieveld
Horst Fischer
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2673 KB) - Metadata XML
-
Supplement
(6783 KB) - BibTeX
- EndNote
- Final revised paper