Unveiling the Power of Stochastic Methods: Advancements in Air Pollution Sensitivity Analysis of the Digital Twin

Round 1

Reviewer 1 Report

The authors present a sensitivity study of input variables (anthropogenic emissions and chemical reaction rate constants) based on the UNI-DEN and DIGITAL AIR models applied over four European cities. They used a historical period (1989-2004) to assess these inputs' influence.

The manuscript needs to be notably improved for its potential publication.

The current organization of the manuscript is:

1.      Introduction

2.      Sensitivity analysis

a.      Definitions (9)

3.      Mathematical model UNI-DEM

4.      Preliminary calculations with UNI-DEM

5.      Methods and algorithms

a.      More definitions (2)

6.      Calculation for evaluating sensitivity analysis

7.      Conclusions

The organization must be upgraded, providing a proper background in the Introduction, which currently indicates too general information about environmental concerns and climate change. The authors should prioritize the air pollution issues and climate change.

The manuscript prioritizes, even excessively, the mathematical component. This component can be summarized. Conversely, other features about the influence of climate change on air pollution need to be included.

Due to the number of variables, authors should consider including a list of abbreviations.

Provide a unique section describing the Methods components and maybe all the definitions.

Provide a unique section with the results.

Provide a Discussion section (missed in the current version).

Provide a table or a figure about the components (emissions, reaction rate constants, UNI-DEM , DIGITAL-AIR, and others?) to clarify their relationships in this study.

References are unsorted in the text. The first reference is [40] and the [52 to 55]. The authors need to reorganize the enumeration.

The study is based on the variation of anthropogenic emissions and chemical reaction rate constants for modeling air quality levels from 1989 to 2004. The assumed variations seem to be based on future scenarios by the IPCC over the study region. However, this is not clearly supported.

Include a figure, maybe the first, of the region of study and the computational domain.

It seems that UNI-DEM was used for air quality modeling in the study region from 1989 to 2004. Authors should include as background references the use of the model, the involved air quality variables, and the corresponding modeling performance.  

The authors highlighted the modeling of the levels of O3. Its levels indeed are of concern in summer in some European regions. However, what about PM10?. There are events of PM10 pollution in southern Europe due to the arrival of PM from the north of Africa. With higher mean temperature values, this component could be more critical in the future.

What about NO2 levels, which currently are a concern in some European cities?

In this sense, the authors can incorporate a table summarizing the air quality levels in the four selected cities.

What about the influence of the increase in temperature on NMVOC (precursors of O3 formation) emissions from vegetation?. NMVOC emission is a crucial component deserving attention in assessing the effect of climate change on air quality.

The authors used average mean values of air pollutants. Based on the nature of the contaminant, it is better to use the periods suggested by the World Health Organization. For example, the current guideline by the WHO for O3 (daily basis) is limited to 100 µg/m3, its concentration for eight consecutive hours.

Author Response

The organization must be upgraded, providing a proper background in the Introduction, which currently indicates too general information about environmental concerns and climate change. The authors should prioritize the air pollution issues and climate change.

Many thanks to the reviewer for the very useful suggestion, some new paragraphs concerning these issues are added in the Introduction. The paper has been totally reorganized prioritizing air pollution issues and climate change as the reviewer suggestion.

The manuscript prioritizes, even excessively, the mathematical component. This component can be summarized. Conversely, other features about the influence of climate change on air pollution need to be included.

Many thanks to the reviewer. We try to extend the paper including some paragraphs for the influence of climate change on air pollution, we include a new paragraph for the influence of natural (biogenic) emissions.

Due to the number of variables, authors should consider including a list of abbreviations.

Many thanks to the reviewer for the very useful remark, a new list with all the abbreviations used in the text has been added.

Provide a unique section describing the Methods components and maybe all the definitions.

All definitions have been given in the Sensitivity Analysis section.

Provide a unique section with the results.

Many thanks to the reviewer for the useful suggestion, a new section named Numerical results has been added.

Provide a Discussion section (missed in the current version).

Many thanks to the reviewer for the useful suggestion, a new section named Discussion and applicability has been added.

Provide a table or a figure about the components (emissions, reaction rate constants, UNI-DEM, DIGITAL-AIR, and others?) to clarify their relationships in this study.

We give a new figure with the methodology for performing multidimensional sensitivity analysis to clarify their relationship in this study.

We also add a Figure with the components of UNI-DEM to clarify their relationship in the study.

References are unsorted in the text. The first reference is [40] and the [52 to 55]. The authors need to reorganize the enumeration.

Many thanks to the reviewer, we totally reorganize the enumeration.

The study is based on the variation of anthropogenic emissions and chemical reaction rate constants for modeling air quality levels from 1989 to 2004. The assumed variations seem to be based on future scenarios by the IPCC over the study region. However, this is not clearly supported.

In the current version of the manuscript we have added a Figure with the expected increases of the temperature in first horizontal level of the space domain of UNI-DEM according to the IPCC reports to support the future scenarios over the study region.

Include a figure, maybe the first, of the region of study and the computational domain.

Many thanks to the review for the very useful comment, a Figure with the region of the study has been added.

It seems that UNI-DEM was used for air quality modeling in the study region from 1989 to 2004. Authors should include as background references the use of the model, the involved air quality variables, and the corresponding modeling performance.  

Many thanks to the reviewer, some more background references are included, regarding the modeling performance.

The authors highlighted the modeling of the levels of O3. Its levels indeed are of concern in summer in some European regions. However, what about PM10?. There are events of PM10 pollution in southern Europe due to the arrival of PM from the north of Africa. With higher mean temperature values, this component could be more critical in the future.

Many thanks to the reviewer for the very useful comment. Unfortunately, PM10 is not included in the current version of UNI-DEM. We realize that this is a serious drawback and in the future we will work hardly on optimizing the UNI-DEM model, however this needs additional years of study.

What about NO2 levels, which currently are a concern in some European cities?

We have added a new paragraph in the Discussion section to your suggestions regarding nitrogen oxides. All four groups of pollutants have an influence on the considered important species, with nitrogen oxides and anthropogenic hydrocarbons exhibiting a slight but not negligible effect. More about NO2 levels can be found in the following paper:

10. Zlatev and D. Syrakov: “A Fine Resolution Modelling Study of Pollution Levels in Bulgaria. Part 1: and Emissions”, International Journal of Environmental Pollution, Vol. 22 (2004), pp. 186-202, http://dx.doi.org/10.1504/IJEP.2004.005508.

In this sense, the authors can incorporate a table summarizing the air quality levels in the four selected cities.

Some references with a table summarizing this air quality levels in the four selected cities has been cited. Also in the Discussion section we give some comments about these air quality levels as suggested by the reviewer, which help us to improve significantly the applicability of the results.

What about the influence of the increase in temperature on NMVOC (precursors of O3 formation) emissions from vegetation? NMVOC emission is a crucial component deserving attention in assessing the effect of climate change on air quality.

Many thanks to the reviewer for this suggestion. The ozone does not necessarily participate in all these reactions. Important precursors of ozone participate in our reactions, but unfortunately, NMVOC emission has not been analyzed. We realize this is important and further investigations will be necessary.

The authors used average mean values of air pollutants. Based on the nature of the contaminant, it is better to use the periods suggested by the World Health Organization. For example, the current guideline by the WHO for O3 (daily basis) is limited to 100 µg/m3, its concentration for eight consecutive hours.

We add a new paragraph and a Figure which provides visual representations illustrating the distribution of "bad days" throughout Europe. To qualify as a "bad day," we examine the maximum value, denoted as cmax, of the 8-hour average ozone concentrations at a given location in Europe on any given day and if cmax exceeds 60 ppb at least once during that day, it is categorized as a "bad day."

Author Response File: Author Response.docx

Reviewer 2 Report

1.       The temperature rise will lead to increase in air pollution is not clearly established. The authors are requested to give more references and elaborate how increase in temperature will raise pollution.

2.       The authors are using the Digital Twin application for the work, but didn’t cite the direct application publications, such as: https://doi.org/10.3389/frsc.2021.786563

3.       Can you state the objective of the work clearly?

4.       Can you describe UNI-DEM biases? What is the accuracy of the model? Have you tested its accuracy for the present scenario?

5.       So, the model has been run for a single domain? If yes, what about the data resolution adjustment from input data to the aforesaid resolution of the model? Don’t you think a two- or three-way nesting will give you better results?

6.       How many vertical levels model had? How many of them are in the boundary layer? What are the parametrization schemes used by the model? All this information should be provided in the revised manuscript.

7.       I am concerned about the model validation results. Have you verified them?

8.       Can you explain your conclusions in plain language where general people understanding air pollution can understand these points?

The language is ok. Needs minor corrections by the authors.

Author Response

1. The temperature rise will lead to increase in air pollution is not clearly established. The authors are requested to give more references and elaborate how increase in temperature will raise pollution.

Many thanks to the reviewer for the very useful comment. It should be mentioned that the levels of the natural (biogenic) emissions are becoming higher when the temperature is increased. This additional fact is also leading to increased levels of some potentially dangerous pollutants, including the levels of the ozone concentrations. We give some references and write a new paragraph about the natural emissions. We give a new Figure with expected increases of the temperature in first horizontal level of the space domain of UNI-DEM according to the IPCC reports.

2. The authors are using the Digital Twin application for the work, but didn’t cite the direct application publications, such as: https://doi.org/10.3389/frsc.2021.786563

Many thanks to the reviewer for this very important paper in the field, we cite it in the Introduction in the revised version of the manuscript.

3. Can you state the objective of the work clearly?

Many thanks to the reviewer for this important note, the objective of the study is on the preparation and further investigation of the digital twin through sensitivity analysis with improved stochastic approaches for investigating high level air pollutants. For the objective of the work we add some references regarding performing multidimensional sensitivity analysis for large-scale air pollution models.

4. Can you describe UNI-DEM biases? What is the accuracy of the model? Have you tested its accuracy for the present scenario?

The accuracy of the model has been tested in previous work of prof. Zahari Zlatev, we add his important works regarding the accuracy of the model in the References. In these works, all important tests, including the well-known “rotation tests” are performed and analyzed.

5. 5.So, the model has been run for a single domain? If yes, what about the data resolution adjustment from input data to the aforesaid resolution of the model? Don’t you think a two- or three-way nesting will give you better results?

      The data resolution adjustment has already been made by Prof. Zahari Zlatev in his previous works. Many thanks about this concerns, two or three ways nestling possible will optimize the model, however we expect it will not affect the stochastic approaches for multidimensional sensitivity analysis, it will be an object of our future study.

6. 6.How many vertical levels model had? How many of them are in the boundary layer? What are the parametrization schemes used by the model? All this information should be provided in the revised manuscript.

A new version of the model was used in this study. The same space domain is discretized on a 480 × 480 grid (10 × 10 km resolution). This refinement increases the computations enormously but the comparisons between the results obtained on the rough and the fine grids show that these efforts are worth it, especially when the 3-D versions with 10 non-equidistant layers in the vertical direction are used.  It was not possible to obtain input data on this high-resolution grid, neither emission nor meteorological data. The available emission data on the 50 km grid are evenly distributed on 25 small grid-squares obtained during the transition to 10 km resolution. Simple linear interpolation (both in space and in time) is used to prepare meteorological data for the fine grid. One should take into account that for the climate research, when multi-year scenarios are run such a level of discretization is satisfactory.

7. 7.I am concerned about the model validation results. Have you verified them?

We add some comments about the verification of the results regarding the sensitivity indices of the mathematical model in the Discussion section. Model validation approaches may be found in Z, Zlatev, I. Dimov, Computational and Numerical Challenges in Environmental Modelling, (2006), Amsterdam-Boston-Heidelberg-London-New York-Oxford-Paris-San Diego-San Francisco-Singapore-Sydney-Tokyo, 373 p., in Elsevier ISBN-13: 978-0-444-52209-2 (monograph),  The validation is performed for the actual version of UNI-DEM and for the actual parametrization schemes.

8. Can you explain your conclusions in plain language where general people understanding air pollution can understand these points?

      Many thanks to the reviewer for this useful comment which help us to improve the Conclusion. We add some new paragraph in the Conclusion to explain our conclusion in plain language.

Author Response File: Author Response.docx

Reviewer 3 Report

This article presents a thorough examination of air pollutant distribution and the factors contributing to high concentrations, emphasizing the importance of understanding the detrimental effects of elevated levels. The authors propose the use of DIGITAL AIR, a Digital Twin tool that encompasses relevant atmospheric processes, and highlight the need for precise sensitivity analysis to enhance its effectiveness. All illustrations, presented data and equations are presented adequately and with the purpose of conveying the article's needs. I fully recommend  the article for publication.

Author Response

This article presents a thorough examination of air pollutant distribution and the factors contributing to high concentrations, emphasizing the importance of understanding the detrimental effects of elevated levels. The authors propose the use of DIGITAL AIR, a Digital Twin tool that encompasses relevant atmospheric processes, and highlight the need for precise sensitivity analysis to enhance its effectiveness. All illustrations, presented data and equations are presented adequately and with the purpose of conveying the article's needs. I fully recommend  the article for publication.

Many thanks to the reviewer for evaluate our manuscript so highly and recommend it for publication.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

The authors have substantially addressed the comments and suggestions made during the first review. Authors should highlight, maybe in the last section, the study's limitations (e.g., PM10 is not included in the current version of UNI-DEM) and the future related activities. After that, I believe this manuscript can be accepted for publishing.

Author Response

The authors have substantially addressed the comments and suggestions made during the first review. Authors should highlight, maybe in the last section, the study's limitations (e.g., PM10 is not included in the current version of UNI-DEM) and the future related activities. After that, I believe this manuscript can be accepted for publishing.

We agree with the remark. We have added the following remark to the Conclusion:

“The current version of UNI-DEM, although a powerful mathematical model for air pollution analysis, has certain limitations that should be acknowledged. One notable limitation is that it does not account for PM10 (consists of small particles suspended in the air, such as dust, pollen, soot, and other solid or liquid pollutants) in its calculations. These particles are small enough to be inhaled into the respiratory system, posing potential health risks. Monitoring and controlling PM10 levels are crucial for assessing air quality and understanding its impact on human health and the environment. However, future iterations of UNI-DEM are expected to address this limitation by incorporating PM10 data and considering its impact on air pollution dynamics. By taking into account this important particulate matter, the model will provide a more comprehensive and accurate representation of air quality, contributing to a deeper understanding of the factors influencing air pollution levels.”

Author Response File: Author Response.docx