the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 12 Jul 2022
High accuracy calculation and data quality evaluation of ship emissions based on the sniffer method
Abstract. More attention has been paid to the air pollution caused by ship emissions; hence the establishment of accurate emission inventories is an important means to assess the impact on the environment and human beings. The emission factor is an important parameter in the process of compiling the ship emission inventory, yet there is some uncertainty in its estimation based on the sniffer method. In this study, taking the calculation of SO2 emission factors as an example and aiming at the selection of gas measurement values using the sniffer method, the concept of standard deviation of peak density was proposed to determine the optimal integral interval length of the measured values of SO2 and CO2. Then, the improved Manhattan distance was used to characterize the position of the peak points in the SO2 and CO2 average series. Using the dynamic time warping algorithm, the corresponding relationship of the peak points in the average series of the measured gases was determined, and the global optimal peak points were selected from it. To evaluate the credibility of calculated emission factors, 16 evaluation indexes that reflect the characteristics of the measured data were selected. The confidence interval of 95 % of each evaluation index was calculated using self-development sampling of the measured data, and the evaluation result of the evaluation index for the quality of the measured data was obtained. Combined with the data quality label, the indexes with high correct rate were screened. Finally, the evaluation scores were determined according to these selected indexes. We collected a total of 148 sets of "SO2+CO2" measurement data between 2019 and 2021 using the unmanned aerial vehicle sniffing monitoring system in the Waigaoqiao Port area of Shanghai, China for verification using the method proposed in this study. The results show that for this data set, 12 s is the most suitable integral length, with which the algorithm can automatically calculate the emission factor. The screening results of the global optimal peak points of 129 groups of data are consistent with those of artificial screening, with a correct rate of 87.16 %. The accuracy of the combined evaluation of sample entropy (SO2), information entropy (SO2), skewness (CO2) and quartile spacing (SO2) is 71 %. Previous calculation of the emission factor of ships mainly focused on different conditions such as time, region, fuel, engine, ship type, and navigation status. Our in-depth study proposes a high accuracy ship emission factors calculation method and an evaluation of the quality of the measurement data that reduces uncertainty in the current sniffer technique monitoring ship emission research.
- Preprint
(2104 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on acp-2022-452', Anonymous Referee #1, 30 Nov 2022
Review of the manuscript „High accuracy calculation and data quality evaluation of ship emissions based on the sniffer method“ by Zhu and Zhou
Shipping emissions are a relevant source of pollution in particular in the clean marine boundary layer. A better characterisation of ship emissions is a relevant topic for atmospheric chemistry studies and for the control of environmental regulations, and improving the accuracy of emission estimates is necessary for a better understanding of the relevance of ship emissions.
In this manuscript, the authors propose a complex algorithm to match air-borne in-situ observations of CO2 and SO2 in ship emission plumes with the aim to determine SO2 emission ratios. Briefly, the algorithm consists of 1) the smoothing of the time series, 2) the assignment of matching peaks in the CO2 and SO2 time series 3) the identification and removal of invalid matches and 4) a quality estimate for the remaining matches. The algorithm is described and applied to a small sample of real measurements of variable quality, and good performance of the algorithm is found in comparison to a subjective assignment.
Unfortunately, I cannot recommend this manuscript for publication. The reason for this assessment is that for each of the steps performed, I disagree with the assumptions made and the approaches taken. In many places, ad hoc parameters are introduced and conclusions are drawn without good justification. In addition, the description is hard to follow and often unclear, both because of issues with the use of English and the way the text is written.
If I understood the manuscript well, the authors try to identify the matching maxima in the time series of SO2 and CO2 for an individual plume measurement in order to determine the SO2 / CO2 ratio from this value. However, I do not see the rationale for using this specific value. The ratio should be constant throughout the plume and therefore could be computed from any pair of matching measurements. If this is not possible for signal to noise reasons, the ratio of the integrals of the two quantities over the plume transect should be taken as this will be less sensitive to noise and to small time lags between the measurements. I therefore think that the whole idea of finding the maxima in the time series is unnecessary and actually not asking the right question.
For the smoothing, the authors try to establish an objective criterion by maximising the “peak density standard deviation”. However, while this criterion is quantitative, it is not clear to me why this value should lead to the optimum smoothing.
The next step is the matching of the peaks in the two time series. The authors use a “Dynamic time warping algorithm” with the Manhatten distance in a normalised time-concentration plane to find matching pairs. However, it is not at all clear why a) this distance is a good metric for finding the peaks and b) why taking the absolute values in the Manhatten distance makes more sense than computing the Eularian distance. More importantly, the need for the matching arises from differences in the time response of the SO2 and CO2 measurements, which should be constant over the short time periods needed to measure a plume. Therefore, the complex algorithm which assigns peaks in the time series allowing variable time lags results in unphysical assignments. It would have been much simpler and a better representation of the physical effects behind the shift in the time series to calculate cross-correlations of the two time series for different realistic time lags to determine the optimal time shift to be applied to the time series.
As the algorithm proposed in the manuscript is not well constrained (see the last point), it often produces unphysical results. In order to identify and remove them, the authors propose a k-means clustering of the “Normalised Concentration Differences” with two clusters. Why such an approach should be able to separate valid and invalid results is not clear. In my opinion, a simple threshold removing pairs with “too large” differences would lead to similar results and be equally subjective.
The final step in the algorithm is a quality assessment of the derived pairs. Here, the authors apply 16 different index definitions on data sets created by “10000 self-development sampling” and then use a complex scheme to extract a high-level quality index from this set of results. I must admit that I did not follow this indexing in detail, as it appears completely arbitrary and pointless to me.
In summary, I think the manuscript tries to solve the wrong problem with a complex and, in many respects, arbitrary algorithm. Instead, the authors should follow a simple and physics-based approach by comparing the SO2 and CO2 values integrated over the plume transect after allowing for a time shift correcting for the different response times of the two instruments used.
Citation: https://doi.org/10.5194/acp-2022-452-RC1 - AC1: 'Reply on RC1', Fan Zhou, 20 Apr 2023
- AC2: 'Reply on RC1', Fan Zhou, 20 Apr 2023
- AC4: 'Reply on RC1', Fan Zhou, 20 Apr 2023
-
RC2: 'Comment on acp-2022-452', Anonymous Referee #2, 25 Jan 2023
Current literature has substantial emission factors for shipping emissions. But these factors are not constant due to developing machine technology and by the shift of fuel content. For example, sulphur dioxide emissions cannot be represented well by the current emission factors. Therefore, the necessity of improving emission factors is obvious. For that reason, I find the study topic useful for the scientific community. In this study, the authors utilised the sniffer technique. The study also aims to target to reduce some uncertainties due to the known drawbacks of the technique. The strengths and limitations of the method are defined.
In my point of view, the methodology has too many assumptions and is quite complex. Therefore, the findings should be verified by a conventional measurement method. Unless it is not easy to decide whether the findings are accurate or not. After incorporating such a cross-check result, the paper can be considered for publication in ACP.
Citation: https://doi.org/10.5194/acp-2022-452-RC2 -
AC3: 'Reply on RC2', Fan Zhou, 20 Apr 2023
The comment was uploaded in the form of a supplement: https://acp.copernicus.org/preprints/acp-2022-452/acp-2022-452-AC3-supplement.pdf
-
AC3: 'Reply on RC2', Fan Zhou, 20 Apr 2023
Status: closed
-
RC1: 'Comment on acp-2022-452', Anonymous Referee #1, 30 Nov 2022
Review of the manuscript „High accuracy calculation and data quality evaluation of ship emissions based on the sniffer method“ by Zhu and Zhou
Shipping emissions are a relevant source of pollution in particular in the clean marine boundary layer. A better characterisation of ship emissions is a relevant topic for atmospheric chemistry studies and for the control of environmental regulations, and improving the accuracy of emission estimates is necessary for a better understanding of the relevance of ship emissions.
In this manuscript, the authors propose a complex algorithm to match air-borne in-situ observations of CO2 and SO2 in ship emission plumes with the aim to determine SO2 emission ratios. Briefly, the algorithm consists of 1) the smoothing of the time series, 2) the assignment of matching peaks in the CO2 and SO2 time series 3) the identification and removal of invalid matches and 4) a quality estimate for the remaining matches. The algorithm is described and applied to a small sample of real measurements of variable quality, and good performance of the algorithm is found in comparison to a subjective assignment.
Unfortunately, I cannot recommend this manuscript for publication. The reason for this assessment is that for each of the steps performed, I disagree with the assumptions made and the approaches taken. In many places, ad hoc parameters are introduced and conclusions are drawn without good justification. In addition, the description is hard to follow and often unclear, both because of issues with the use of English and the way the text is written.
If I understood the manuscript well, the authors try to identify the matching maxima in the time series of SO2 and CO2 for an individual plume measurement in order to determine the SO2 / CO2 ratio from this value. However, I do not see the rationale for using this specific value. The ratio should be constant throughout the plume and therefore could be computed from any pair of matching measurements. If this is not possible for signal to noise reasons, the ratio of the integrals of the two quantities over the plume transect should be taken as this will be less sensitive to noise and to small time lags between the measurements. I therefore think that the whole idea of finding the maxima in the time series is unnecessary and actually not asking the right question.
For the smoothing, the authors try to establish an objective criterion by maximising the “peak density standard deviation”. However, while this criterion is quantitative, it is not clear to me why this value should lead to the optimum smoothing.
The next step is the matching of the peaks in the two time series. The authors use a “Dynamic time warping algorithm” with the Manhatten distance in a normalised time-concentration plane to find matching pairs. However, it is not at all clear why a) this distance is a good metric for finding the peaks and b) why taking the absolute values in the Manhatten distance makes more sense than computing the Eularian distance. More importantly, the need for the matching arises from differences in the time response of the SO2 and CO2 measurements, which should be constant over the short time periods needed to measure a plume. Therefore, the complex algorithm which assigns peaks in the time series allowing variable time lags results in unphysical assignments. It would have been much simpler and a better representation of the physical effects behind the shift in the time series to calculate cross-correlations of the two time series for different realistic time lags to determine the optimal time shift to be applied to the time series.
As the algorithm proposed in the manuscript is not well constrained (see the last point), it often produces unphysical results. In order to identify and remove them, the authors propose a k-means clustering of the “Normalised Concentration Differences” with two clusters. Why such an approach should be able to separate valid and invalid results is not clear. In my opinion, a simple threshold removing pairs with “too large” differences would lead to similar results and be equally subjective.
The final step in the algorithm is a quality assessment of the derived pairs. Here, the authors apply 16 different index definitions on data sets created by “10000 self-development sampling” and then use a complex scheme to extract a high-level quality index from this set of results. I must admit that I did not follow this indexing in detail, as it appears completely arbitrary and pointless to me.
In summary, I think the manuscript tries to solve the wrong problem with a complex and, in many respects, arbitrary algorithm. Instead, the authors should follow a simple and physics-based approach by comparing the SO2 and CO2 values integrated over the plume transect after allowing for a time shift correcting for the different response times of the two instruments used.
Citation: https://doi.org/10.5194/acp-2022-452-RC1 - AC1: 'Reply on RC1', Fan Zhou, 20 Apr 2023
- AC2: 'Reply on RC1', Fan Zhou, 20 Apr 2023
- AC4: 'Reply on RC1', Fan Zhou, 20 Apr 2023
-
RC2: 'Comment on acp-2022-452', Anonymous Referee #2, 25 Jan 2023
Current literature has substantial emission factors for shipping emissions. But these factors are not constant due to developing machine technology and by the shift of fuel content. For example, sulphur dioxide emissions cannot be represented well by the current emission factors. Therefore, the necessity of improving emission factors is obvious. For that reason, I find the study topic useful for the scientific community. In this study, the authors utilised the sniffer technique. The study also aims to target to reduce some uncertainties due to the known drawbacks of the technique. The strengths and limitations of the method are defined.
In my point of view, the methodology has too many assumptions and is quite complex. Therefore, the findings should be verified by a conventional measurement method. Unless it is not easy to decide whether the findings are accurate or not. After incorporating such a cross-check result, the paper can be considered for publication in ACP.
Citation: https://doi.org/10.5194/acp-2022-452-RC2 -
AC3: 'Reply on RC2', Fan Zhou, 20 Apr 2023
The comment was uploaded in the form of a supplement: https://acp.copernicus.org/preprints/acp-2022-452/acp-2022-452-AC3-supplement.pdf
-
AC3: 'Reply on RC2', Fan Zhou, 20 Apr 2023
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 559 | 165 | 28 | 752 | 16 | 17 |
- HTML: 559
- PDF: 165
- XML: 28
- Total: 752
- BibTeX: 16
- EndNote: 17
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1