the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
An estimate of excess mortality resulting from air pollution caused by wildfires in the eastern and central Mediterranean basin in 2021
Abstract. Wildfires result in human fatalities not only due to the direct exposure to flames, but also indirectly through smoke inhalation. The Mediterranean basin with its hot and dry summers is a hotspot for such devastating events. The situation has further been aggravated in recent years by climate change as well as a growing and aging population in the region. To assess the health impacts due to short-term exposure to air pollution created by the 2021 summer wildfires in the eastern and central Mediterranean basin, we used a regional-scale chemistry transport model to simulate concentrations of major air pollutants such as fine particulate matter with an aerodynamic diameter less than 2.5 µm (PM2.5), SO2, NO2, and O3 – in a fire and a no-fire scenario. Elevated short-term exposure of the population to air pollutants are associated with excess all-cause mortality using relative risks (RRs) for individual pollutants from previously published meta-analyses. Our estimates indicate that the short-term exposure to wildfire-caused changes in O3 accounted for 289 (95 % CI: 214–364) excess deaths in total over the entire region of investigation during the wildfire season between mid-July to early October 2021. This is followed by 87 (95 % CI: 56–118) excess deaths due to elevated PM2.5 exposure, rendering the health effect of increased O3 from wildfires larger than the effect of increased PM2.5. We attributed this largely to the spatially more widespread impact of wildfires on O3. Our study concludes with a discussion on uncertainties associated with the health impact assessment based on different air pollutants.
This preprint has been withdrawn.
-
Withdrawal notice
This preprint has been withdrawn.
-
Preprint
(6272 KB)
Interactive discussion
Status: closed
-
RC1: 'Comment on nhess-2023-111', Anonymous Referee #1, 15 Aug 2023
Summary:
The authors used WRF-Chem model to simulate air pollution concentrations, including PM2.5, O3, NO2, and SO2, and quantified the health impacts attributable to wildfire-related O3 and PM2.5 over the eastern and central Mediterranean basis during the summer season of 2021. They found that fire-caused O3 accounted for more excess deaths than those due to fire-specific PM2.5. I recommend the paper be reconsidered after major revisions.
Major Comments:
- The authors used the WRF-Chem model to simulate concentrations of PM2.5, SO2, NO2, and O3. Consequently, my anticipation was to discover the health impact results pertaining to these four air pollutants. However, the authors only focused on the health impacts due to fire-caused PM2.5 and O3. While Section 3.3 indicate that “wildfires have not resulted in a discernible increase of population exposure to NO2 and SO2 within the simulation period and region”. I believe it remains valuable to estimate the excess mortality attributable to fire-related NO2 and SO2. Alternatively, the authors could opt to concentrate solely on PM2.5 and O3 throughout the manuscript.
- The study’s innovation is not articulated with clarity. Given the presence of preexisting studies conducting fire-related health impact assessments, it is imperative for the authors to elucidate the points of differentiation between their study and the prior works. This elucidation will enable a better comprehension of the unique aspects and novelty intrinsic to their approach.
- The author made lots of assumptions in Section 2.4.2: 1) “To interpolate the annual deaths to the summer months in 2021, we used the multi-annual average of monthly deaths as proxy” 2) For countries where monthly deaths are not available, the monthly mean deaths averaged over all other countries in the region were used as proxy” 3) “Total monthly deaths were equally distributed to each day of the month.” It’s not clear why these assumptions hold. The author provides references or analysis to support these assumptions.
Minor Comments:
- Abstract: Line 5- “we used a regional-scale chemistry transport model to simulate …” I suggest the authors mention WRF-Chem here to inform readers about the model they used for simulation.
- Abstract Line 6 – “such as PM2.5, SO2, NO2, …” , replace such as with including. Remove the hyphen before “in a fire and a no-fire scenario”.
- Abstract: The authors mentioned the number of excess mortalities due to O3 and PM2.5, but didn’t provide any information regarding SO2 and NO2. I recommend adding one or two sentences to describe the health outcomes due to SO2 and NO2.
- Introduction – Last paragraph: “An online-coupled atmospheric chemistry transport model was employed to ….” I suggest specifying the use of the WRF-Chem model in this context.
- Introduction – Last paragraph: “with detailed discussion on the uncertainties associated with the estimates” It would be beneficial to provide more clarity on which sources of uncertainty the authors focused on in their study. Did they primarily address uncertainties stemming from RR, air pollution predictions, or perhaps other sources of uncertainty?
- Section 2.1 Study area and period: Label the countries mentioned in the manuscript in Figure 1. The same rule applied to other figures (i.e., Figures 2&3) In addition, for Figure 1, consider utilizing a slightly darker color to delineate the boundaries of the regions or countries of interest.
- Line 109-110: Needs further elaboration on the nature of fire and no-fire scenarios. Does the fire scenario solely encompasses fire emissions or include both fire and other emission sources?
- Throughout the manuscript, the author frequently mentioned they conducted short-term exposure health impacts assessment. The authors need to specify the temporal resolution. Is it hourly or daily, or weekly?
- Section 2.4.1 “Population weighted exposure was computed for both fire and non-fire scenarios, with their difference representing the additional health burden attributable to wildfires”. Their difference is population-weighted fire-contributed air pollution. This difference does not yet reflect any associated health burden.
- Section 3.3 “By contrast, wildfires have not resulted in a discernible increase of population exposure…” Suggest to correct it and make it quantitatively.
- Section 4 Discussion: “We have accounted for uncertainties…. However, we have to acknowledge several sources of uncertainty that have not been taken into account..”. Providing a few examples of these unconsidered uncertainties would enrich the discussion.
Citation: https://doi.org/10.5194/nhess-2023-111-RC1 -
RC2: 'Comment on nhess-2023-111', Anonymous Referee #2, 29 Aug 2023
The study evaluates the impacts of wildfires on air pollutants over the Mediterranean basin in 2021. They found that wildfires caused ozone accounted for more excess deaths than that caused by wildfire PM2.5. While the conclusion is interesting, I have some concerns about the methods and the results.
Major Comments:
- It is interesting that wildfires cause more ozone pollution than PM2.5 pollution, which is opposite to what we find in other fire episodes which PM2.5 pollution is generally dominant. It’s unclear to me what physical and chemical mechanisms have led to the larger impacts of ozone. It seems that the model overpredicts ozone by 45% and underpredicts PM2.5 by 39% (Table B2), so it is very likely the model may also overestimate the impacts of ozone but underestimate PM2.5. It’s unclear to me what led to such a big bias in the model. Since the calculation of the health impacts is dependent on the baseline PM2.5 and ozone level, the model bias may well be the reason why the authors find counter-intuitive results. I don’t think the conclusions are robust without further improvement of the model performance.
- The evaluation of the model performance is mostly focused on general performance. Since the topic of this manuscript is to simulate the impacts of wildfires, it’d be more interesting to evaluate the model performance within fire plumes. For example, when the stations are affected by fire smoke, what is the model's performance versus smoke-free conditions? I’d also suggest the authors evaluate model simulations of fire impacts using remote sensing observations of CO, which is a good indicator of biomass burning smoke.
- I’m concerned about whether this manuscript would make a good fit for NHESS. The authors show that the impacts of fires over their study region are almost negligible, meaning that these are not extreme events. And there is no clear distinction between agriculture and wildland fires, so it’s unclear how it’s related to “natural” phenomenon.
- The authors focus their analysis on PM2.5, SO2, NO2 and O3, but SO2 is generally not emitted from fires. The authors should instead look at the chemical species that can be emitted from fires, such as VOCs, CO.
- O3 is not emitted directly, and its formation in fire plumes is related to emissions of NOx and VOCs. Smoke aerosols may also negatively influence ozone formation either chemically or radiatively. There are no discussions on how ozone is formed from fires, and why widespread impacts are found despite ozone precursor NO2 is limited to source regions.
- This study focuses on wildfires, but it seems that agriculture fires are also included in FINN. The authors didn’t explain how they separate wildfires from agriculture fires.
Minor Comments:
- The descriptions of the fire emissions and chemistry in WRF-Chem should be expanded to include emission factors, plume rise schemes, and evaluation of FINN from previous studies. It’s been shown that the emissions of OC+BC in FINN v1.5 are much lower than other biomass burning emission inventories (Liu et al., 2019). The authors should discuss whether uncertainties of the biomass burning inventories affect their conclusions.
- It is unclear to me how the authors define ‘short-term exposure’. Are you looking at acute health effects? It looks like they’re looking at multi-month mean, which is more likely to be a chronic or intermediate-term exposure analysis.
References:
Liu, T. et al. Diagnosing spatial biases and uncertainties in global fire emissions inventories_ Indonesia as regional case study. Remote Sensing of Environment 237, 111557 (2019).
Citation: https://doi.org/10.5194/nhess-2023-111-RC2
Interactive discussion
Status: closed
-
RC1: 'Comment on nhess-2023-111', Anonymous Referee #1, 15 Aug 2023
Summary:
The authors used WRF-Chem model to simulate air pollution concentrations, including PM2.5, O3, NO2, and SO2, and quantified the health impacts attributable to wildfire-related O3 and PM2.5 over the eastern and central Mediterranean basis during the summer season of 2021. They found that fire-caused O3 accounted for more excess deaths than those due to fire-specific PM2.5. I recommend the paper be reconsidered after major revisions.
Major Comments:
- The authors used the WRF-Chem model to simulate concentrations of PM2.5, SO2, NO2, and O3. Consequently, my anticipation was to discover the health impact results pertaining to these four air pollutants. However, the authors only focused on the health impacts due to fire-caused PM2.5 and O3. While Section 3.3 indicate that “wildfires have not resulted in a discernible increase of population exposure to NO2 and SO2 within the simulation period and region”. I believe it remains valuable to estimate the excess mortality attributable to fire-related NO2 and SO2. Alternatively, the authors could opt to concentrate solely on PM2.5 and O3 throughout the manuscript.
- The study’s innovation is not articulated with clarity. Given the presence of preexisting studies conducting fire-related health impact assessments, it is imperative for the authors to elucidate the points of differentiation between their study and the prior works. This elucidation will enable a better comprehension of the unique aspects and novelty intrinsic to their approach.
- The author made lots of assumptions in Section 2.4.2: 1) “To interpolate the annual deaths to the summer months in 2021, we used the multi-annual average of monthly deaths as proxy” 2) For countries where monthly deaths are not available, the monthly mean deaths averaged over all other countries in the region were used as proxy” 3) “Total monthly deaths were equally distributed to each day of the month.” It’s not clear why these assumptions hold. The author provides references or analysis to support these assumptions.
Minor Comments:
- Abstract: Line 5- “we used a regional-scale chemistry transport model to simulate …” I suggest the authors mention WRF-Chem here to inform readers about the model they used for simulation.
- Abstract Line 6 – “such as PM2.5, SO2, NO2, …” , replace such as with including. Remove the hyphen before “in a fire and a no-fire scenario”.
- Abstract: The authors mentioned the number of excess mortalities due to O3 and PM2.5, but didn’t provide any information regarding SO2 and NO2. I recommend adding one or two sentences to describe the health outcomes due to SO2 and NO2.
- Introduction – Last paragraph: “An online-coupled atmospheric chemistry transport model was employed to ….” I suggest specifying the use of the WRF-Chem model in this context.
- Introduction – Last paragraph: “with detailed discussion on the uncertainties associated with the estimates” It would be beneficial to provide more clarity on which sources of uncertainty the authors focused on in their study. Did they primarily address uncertainties stemming from RR, air pollution predictions, or perhaps other sources of uncertainty?
- Section 2.1 Study area and period: Label the countries mentioned in the manuscript in Figure 1. The same rule applied to other figures (i.e., Figures 2&3) In addition, for Figure 1, consider utilizing a slightly darker color to delineate the boundaries of the regions or countries of interest.
- Line 109-110: Needs further elaboration on the nature of fire and no-fire scenarios. Does the fire scenario solely encompasses fire emissions or include both fire and other emission sources?
- Throughout the manuscript, the author frequently mentioned they conducted short-term exposure health impacts assessment. The authors need to specify the temporal resolution. Is it hourly or daily, or weekly?
- Section 2.4.1 “Population weighted exposure was computed for both fire and non-fire scenarios, with their difference representing the additional health burden attributable to wildfires”. Their difference is population-weighted fire-contributed air pollution. This difference does not yet reflect any associated health burden.
- Section 3.3 “By contrast, wildfires have not resulted in a discernible increase of population exposure…” Suggest to correct it and make it quantitatively.
- Section 4 Discussion: “We have accounted for uncertainties…. However, we have to acknowledge several sources of uncertainty that have not been taken into account..”. Providing a few examples of these unconsidered uncertainties would enrich the discussion.
Citation: https://doi.org/10.5194/nhess-2023-111-RC1 -
RC2: 'Comment on nhess-2023-111', Anonymous Referee #2, 29 Aug 2023
The study evaluates the impacts of wildfires on air pollutants over the Mediterranean basin in 2021. They found that wildfires caused ozone accounted for more excess deaths than that caused by wildfire PM2.5. While the conclusion is interesting, I have some concerns about the methods and the results.
Major Comments:
- It is interesting that wildfires cause more ozone pollution than PM2.5 pollution, which is opposite to what we find in other fire episodes which PM2.5 pollution is generally dominant. It’s unclear to me what physical and chemical mechanisms have led to the larger impacts of ozone. It seems that the model overpredicts ozone by 45% and underpredicts PM2.5 by 39% (Table B2), so it is very likely the model may also overestimate the impacts of ozone but underestimate PM2.5. It’s unclear to me what led to such a big bias in the model. Since the calculation of the health impacts is dependent on the baseline PM2.5 and ozone level, the model bias may well be the reason why the authors find counter-intuitive results. I don’t think the conclusions are robust without further improvement of the model performance.
- The evaluation of the model performance is mostly focused on general performance. Since the topic of this manuscript is to simulate the impacts of wildfires, it’d be more interesting to evaluate the model performance within fire plumes. For example, when the stations are affected by fire smoke, what is the model's performance versus smoke-free conditions? I’d also suggest the authors evaluate model simulations of fire impacts using remote sensing observations of CO, which is a good indicator of biomass burning smoke.
- I’m concerned about whether this manuscript would make a good fit for NHESS. The authors show that the impacts of fires over their study region are almost negligible, meaning that these are not extreme events. And there is no clear distinction between agriculture and wildland fires, so it’s unclear how it’s related to “natural” phenomenon.
- The authors focus their analysis on PM2.5, SO2, NO2 and O3, but SO2 is generally not emitted from fires. The authors should instead look at the chemical species that can be emitted from fires, such as VOCs, CO.
- O3 is not emitted directly, and its formation in fire plumes is related to emissions of NOx and VOCs. Smoke aerosols may also negatively influence ozone formation either chemically or radiatively. There are no discussions on how ozone is formed from fires, and why widespread impacts are found despite ozone precursor NO2 is limited to source regions.
- This study focuses on wildfires, but it seems that agriculture fires are also included in FINN. The authors didn’t explain how they separate wildfires from agriculture fires.
Minor Comments:
- The descriptions of the fire emissions and chemistry in WRF-Chem should be expanded to include emission factors, plume rise schemes, and evaluation of FINN from previous studies. It’s been shown that the emissions of OC+BC in FINN v1.5 are much lower than other biomass burning emission inventories (Liu et al., 2019). The authors should discuss whether uncertainties of the biomass burning inventories affect their conclusions.
- It is unclear to me how the authors define ‘short-term exposure’. Are you looking at acute health effects? It looks like they’re looking at multi-month mean, which is more likely to be a chronic or intermediate-term exposure analysis.
References:
Liu, T. et al. Diagnosing spatial biases and uncertainties in global fire emissions inventories_ Indonesia as regional case study. Remote Sensing of Environment 237, 111557 (2019).
Citation: https://doi.org/10.5194/nhess-2023-111-RC2
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
410 | 148 | 37 | 595 | 32 | 32 |
- HTML: 410
- PDF: 148
- XML: 37
- Total: 595
- BibTeX: 32
- EndNote: 32
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1
Cited
2 citations as recorded by crossref.
- Assessment of forest fire emissions in Uttarakhand State, India, using Open Geospatial data and Google Earth Engine L. Goparaju et al. 10.1007/s11356-023-29311-0
- Short-Term Exposure to Fine Particulate Matter and Ozone: Source Impacts and Attributable Mortalities S. Liu et al. 10.1021/acs.est.4c00339