Review Reports - Health and Economic Benefits of Ozone Reduction: Case Study in Santiago and Valparaíso

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Focus on health and economic benefits of ozone reduction, the relative risks of respiratory and cardiovascular mortality and morbidity were evaluated by integrating health, ozone, and meteorological data from 2009 to 2020. It is an interesting and meaningful work. Some comments were listed below:

The title is too long.
Abstract: It is suggested that the significance and necessity of this study be supplemented in the first sentence, and the results obtained from this study be refined.
Introduction: Regarding the research status of health issues caused by ozone pollution, the existing elaborations are insufficient; there is also a lack of discussion on which problems remain unclear and which technologies are in urgent need of breakthroughs at present. In addition, the introduction of the two regions should be integrated into Section 2.1, and there is no need to elaborate too much in the background part. Besides, it is also necessary to explain the representativeness of the research on those two regions.
Methods: “The population was stratified into three age groups to assess differential vulnerabilities.” The grouping of people seems too simple. Could the age group of 5 to 64 be further subdivided?
Method: Section 2.2 on data preprocessing and imputation, it is recommended to validate the imputed results after completion, for example, by employing cross-validation methods to assess the effectiveness and reliability of the imputation. In the results analysis, it is important to discuss the limitations of the imputation method and potential sources of bias to enhance the transparency and interpretability of the study. Additionally, consider the impact of long-term trends and seasonal factors (e.g., weekday effects) on the results for a more comprehensive understanding of the data.
Methods: “Despite the onset of the COVID-19 pandemic in 2020, sensitivity analyses excluding the 2020 data showed no significant impact on the results, justifying the inclusion of the full 2009–2020 period.” It is difficult to accept that the COVID-19 pandemic did not have a significant impact on this research. After all, the COVID-19 pandemic has brought about significant changes in health. Please elaborate in detail on how the sensitivity analysis is carried out? How can the above conclusion be drawn?
𝐸𝑞.1: This function is the core of this study. It is suggested to supplement the supporting theory for the construction of this function. For example, is the setting of degrees of freedom an optimized choice? On what basis is the optimization or selection made?

Also, why do different influencing factors exhibit a linear additive relationship? For instance, meteorological factors themselves largely affect the formation of ozone pollution; pollutants (e.g., PM2.5) also exert an impact on the formation of ozone pollution. How should the influence of such cross-interaction effects be considered?

Results and discussions: The comparison of pollution conditions in the two regions still needs to be strengthened. Under what kind of emission conditions and meteorological characteristics do the two regions have, what are the characteristics of ozone pollution formation?
Section 3.3: The relationship between the reduction of health issues and ozone control can only be discussed after demonstrating the scientific nature of the methods.
Section 3.4: Whether the restriction on economic development caused by the implementation of ozone pollution control measures has been taken into consideration?
Figure 4: the right vertical axis labels are incomplete.

Author Response

Reviewer 1

The title is too long.

We appreciate Reviewer 1's positive feedback regarding the focus on the health and economic benefits of ozone reduction, as well as the integration of health, ozone, and meteorological data from 2009 to 2020, which underscores the study's relevance. To address the concern about the title being too long, we have revised the original title to Quantifying Health and Economic Benefits of Ozone Reduction in Chile's Metropolitan and Valparaíso Regions: BenMAP-Based Estimates of Avoided Mortality, Morbidity, and Cost Savings Under Primary and WHO Air Quality Standards. The updated title, "Health and Economic Benefits of Ozone Reduction: A Case Study in Santiago and Valparaíso," now comprises 13 words. This revision maintains the core focus on health and economic benefits while specifying the study regions concisely, aligning with the Special Issue's emphasis on urban-environment-health interactions and improving readability for a broad readership.

Abstract: It is suggested that the significance and necessity of this study be supplemented in the first sentence, and the results obtained from this study be refined.

The abstract has been revised to emphasize the significance and necessity of the study in the first sentence, highlighting the urgent public health challenge posed by ozone exposure in Chile's polluted urban areas. The opening now reads: "This study addresses the urgent need to assess health risks from short-term ozone exposure in 13 polluted communes across Chile's Santiago Metropolitan and Valparaíso regions." Additionally, the results have been refined for clarity and accuracy, updating the estimated avoided cases to 837,498 in Santiago and 17,992 in Valparaíso, and the economic savings to $ 7,439,930,640 and $ 1,044,568,800 under the WHO guideline, respectively, aligning with the latest data. These adjustments enhance the abstract's precision and relevance.

Introduction: Regarding the research status of health issues caused by ozone pollution, the existing elaborations are insufficient; there is also a lack of discussion on which problems remain unclear and which technologies are in urgent need of breakthroughs at present. In addition, the introduction of the two regions should be integrated into Section 2.1, and there is no need to elaborate too much in the background part. Besides, it is also necessary to explain the representativeness of the research on those two regions.

The introduction has been reformulated to address the identified shortcomings by enhancing the depth of the research status on health issues caused by ozone pollution. Additional content has been incorporated to elaborate on unresolved problems, such as the synergistic effects with PM2.5 and socioeconomic disparities in susceptibility, as well as the urgent need for breakthroughs in real-time precursor monitoring and advanced photochemical modeling technologies. The revised version now includes 24 studies providing a more comprehensive literature base to support the study's context. Furthermore, detailed descriptions of the Santiago Metropolitan and Valparaíso regions, previously elaborated in the introduction, have been integrated into Section 2.1 to streamline the background and focus the introductory section on scientific gaps and objectives. Section 2.1 now includes specific details on the regions' topographic and meteorological characteristics, enhancing methodological clarity. The representativeness of these regions has been justified by highlighting their diverse urban, industrial, and coastal settings as archetypes for Latin American megacities, reflecting a range of pollution sources and dynamics relevant to broader environmental health research.

Methods: "The population was stratified into three age groups to assess differential vulnerabilities." The grouping of people seems too simple. Could the age group of 5 to 64 be further subdivided?

We divided the same age group as the methodology applied in Pino-Cortés, E., Díaz-Robles, L.A., Campos, V. et al. Effect of socioeconomic status on the relationship between short-term exposure to PM2.5 and cardiorespiratory mortality and morbidity in a megacity: the case of Santiago de Chile. Air Qual Atmos Health 13, 509–517 (2020). https://doi.org/10.1007/s11869-020-00818-6

We considered this assumption to compare our results with the same age group.

Method: Section 2.2 on data preprocessing and imputation, it is recommended to validate the imputed results after completion, for example, by employing cross-validation methods to assess the effectiveness and reliability of the imputation. In the results analysis, it is important to discuss the limitations of the imputation method and potential sources of bias to enhance the transparency and interpretability of the study. Additionally, consider the impact of long-term trends and seasonal factors (e.g., weekday effects) on the results for a more comprehensive understanding of the data.

Sections 2.2 and 3.1 on data preprocessing and imputation have been revised to enhance methodological rigor. In Section 2.2, the imputation process using the MissForest algorithm now incorporates 10-fold cross-validation, targeting an R² threshold of at least 0.90 to ensure reliability and address potential biases. However, limitations such as the assumption of random missingness and possible systematic gaps (e.g., monitor downtime) are explicitly noted. Long-term trends and seasonal factors, including weekday effects and summer ozone peaks, were incorporated into the regression models, with sensitivity analyses conducted to assess their impact. In Section 3.1, the results discussion validates the imputation's effectiveness, reporting an average R² of 0.92 and a 15% reduction in model residuals due to these adjustments, improving the study's transparency and interpretability in line with the reviewer's recommendations.

Methods: "Despite the onset of the COVID-19 pandemic in 2020, sensitivity analyses excluding the 2020 data showed no significant impact on the results, justifying the inclusion of the full 2009–2020 period." It is difficult to accept that the COVID-19 pandemic did not have a significant impact on this research. After all, the COVID-19 pandemic has brought about significant changes in health. Please elaborate in detail on how the sensitivity analysis is carried out? How can the above conclusion be drawn?

The assertion that the COVID-19 pandemic in 2020 had no significant impact on the results is supported by the extensive 12-year dataset (2009–2020), where the 9-month period from March to December 2020 constitutes less than 6% of the total observations. The sensitivity analysis was conducted by excluding 2020 data and recalculating relative risks (RRs) using the Poisson regression model, adjusted for trends, meteorological factors, and day-of-week effects, across all communes for key health outcomes (mortality, hospital admissions, emergency visits). It revealed average RR variations of less than 2.5%, with confidence intervals showing substantial overlap (p > 0.05), indicating statistical stability. The minimal impact is attributable to the dilution of 2020 effects within the long-term trend, where preliminary SINCA data suggest reduced ozone precursors (e.g., NOx, VOCs) due to lockdowns, likely offsetting any increased health vulnerability from healthcare strain. However, we acknowledge potential underreporting of non-COVID health outcomes in 2020 as a limitation, given the focus on pandemic-related data. The robustness of the 12-year baseline thus justifies the inclusion of the full period, though future studies with extended post-2020 data could further clarify pandemic-specific effects.

??.1: This function is the core of this study. It is suggested to supplement the supporting theory for the construction of this function. For example, is the setting of degrees of freedom an optimized choice? On what basis is the optimization or selection made?

The core function (Eq. 1) is a time-series Poisson regression model based on the National Morbidity, Mortality, and Air Pollution Study (NMMAPS) protocol, a widely recognized standard in environmental epidemiology. The degrees of freedom (df) for spline functions were set at 7 per year for temporal trends, 6 for temperature, and 3 for relative humidity, wind speed, and confounding pollutants (PM10, PM2.5, SO2, NO2, CO), consistent with the 12-year dataset size and variability, as supported by Peng and Dominici (2008). This selection aligns with established practices for balancing model fit, validated by the stability of RR estimates in preliminary analyses. The linear additive relationship assumes that confounding factors (meteorology, pollutants) are independently adjusted to isolate ozone's effect, a standard approach in short-term exposure studies when interaction terms show no significant effect (p > 0.10 in initial tests). These pollutants are included as confounders to control co-exposure effects, not as primary exposures. However, we recognize that meteorological factors (e.g., temperature) and pollutants (e.g., PM2.5) influence ozone formation, and future analyses could explore interaction terms or multi-pollutant models to address cross-interaction effects, though current data limitations and the focus on ozone-specific RRs justify the additive model.

Results and discussions: The comparison of pollution conditions in the two regions still needs to be strengthened. Under what kind of emission conditions and meteorological characteristics do the two regions have, what are the characteristics of ozone pollution formation?

The comparison of pollution conditions between the Santiago Metropolitan and Valparaíso regions has been strengthened in Section 3. A new introductory paragraph contrasts the regions' emission conditions and meteorological characteristics: Santiago's basin topography traps vehicular and industrial emissions, with temperature inversions and low winds enhancing ozone formation, while Valparaíso's coastal setting features sea breeze dilution of ozone (2–3 times lower than Santiago) offset by industrial emissions (e.g., thermoelectric plants) in specific communes. The characteristics of ozone pollution formation are detailed, with Santiago's urban heat island driving persistent high ozone and Valparaíso's variability reflecting point-source pollution and maritime influence. These adjustments clarify the spatial and environmental factors shaping the observed RRs, addressing the reviewer's recommendation.

Section 3.3: The relationship between the reduction of health issues and ozone control can only be discussed after demonstrating the scientific nature of the methods.

The scientific nature of the methods underpinning the relationship between ozone control and health issue reduction is well-established. The relative risks (RRs) in Sections 3.1 and 3.2 were derived using a time-series Poisson regression model based on the NMMAPS protocol [22], a globally recognized standard, as detailed in Section 2 (Statistical Analysis, Eq. 1). These RRs were then integrated into the BenMAP-CE model [43], a validated tool by the U.S. EPA for estimating avoided health outcomes, ensuring a robust scientific foundation. The discussion in Section 3.3 on avoided cases (e.g., 1,465–7,800 in the Metropolitan Region, 200–1,200 in Valparaíso) is thus directly supported by these methods, linking ozone reduction scenarios to quantifiable health benefits. No further demonstration of methodological validity is required, as the established protocols and tools provide sufficient rigor.

Section 3.4: Whether the restriction on economic development caused by the implementation of ozone pollution control measures has been taken into consideration?

The economic benefits presented in Section 3.4 reflect the monetized savings from avoided health cases (e.g., 2,465,478,010–7,625,923,945 USD in the Metropolitan Region, 532,504,187–1,070,683,061 USD in Valparaíso) using the BenMAP-CE model, which quantifies health benefits based on relative risks and established cost metrics (e.g., VSL of 2,460,000 USD). This analysis focuses on the societal and healthcare cost reductions resulting from ozone control, as derived from the NMMAPS-based Poisson regression (Eq. 1) and validated scenarios (Primary Standard, WHO Recommendation). Restrictions on economic development due to the implementation of control measures (e.g., industrial regulations, traffic restrictions) were not explicitly considered, as BenMAP-CE is designed to estimate health-related benefits rather than implementation costs or macroeconomic impacts. These aspects, which would require a separate cost-benefit analysis involving data on policy costs and economic productivity, are beyond the current study's scope but could be explored in future research to provide a comprehensive economic assessment.

Figure 4: the right vertical axis labels are incomplete.

Figure 4 has been updated.

Reviewer 2 Report

Comments and Suggestions for Authors

Page 2 line 3, correctly place the CH4 subscript.

Page 3 correctly place the subscripts of the compounds and particulate matter.

Page 6, near the end, correctly place the 2.5 index of particulate matter.

Figure 2 y 3 does not show any blue spaces; instead, there are blank spaces. I recommend supplementing theses figures with the nomenclature explained in the text.

In section 3.3 tables are mentioned that do not appear in the document, it is suggested to reorganize this information.

Page 10, correctly place the subscripts that correspond to the PM.

Author Response

Reviewer 2

Page 2 line 3, correctly place the CH4 subscript.

Done

Page 3 correctly place the subscripts of the compounds and particulate matter.

Done

Page 6, near the end, correctly place the 2.5 index of particulate matter.

Done

Figure 2 y 3 does not show any blue spaces; instead, there are blank spaces. I recommend supplementing theses figures with the nomenclature explained in the text.

Figures 2 and 3 have been updated.

In section 3.3 tables are mentioned that do not appear in the document, it is suggested to reorganize this information.

Corrected.

Page 10, correctly place the subscripts that correspond to the PM.

Done.

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript is focused on an interesting topic related to health and economic benefits of the reduction scenarios of ozone concentration. However, the revision process has proved to be very difficult and major concerns are related to figures/tables shown to present the results.

In my opinion, they are not adequate to support results and discussions and compromise the overall readability of the manuscript: in fact, the way the authors choose to show RR values and confidence intervals in figures 2 and 3 is not standard and it is almost impossible to get a clear idea of the actual importance of the results and their statistically significance. In addition, most RR values reported in section 3 Results and discussions are not present in figures 2 and 3, that is quite unexpected.

As a general comment, I suggest to replace the heatmaps in figures 2 and 3 with more standard figures according to literature. In my experience the best way is to present RR values are plotted together grouped by geographical areas and/or pathologies together with the confidence intervals shown as error-bars.

In addition, it is not clear whether mortality, hospital admissions and emergency visits are added together, because they seem to be considered separately in the text, but it does not emerge from the figures where only a single RR is present for each pathology. Anyway, this is not a common approach and it is not advisable. Finally, comments are made in relation to age groups (in particular "G2"), but there is no reference to them in the figures/tables.

The 1-4 stages of the modeling process are presented in Methods, but their relevance is not quite clear.

Also the tables in the supplementary materials must be improved and re-organized since they are almost unreadable in order to follow the comments in the manuscript.

Author Response

Reviewer 3

We sincerely appreciate the reviewer's insightful comments regarding the presentation of results in Figures 2 and 3, as well as the overall readability of the manuscript. We have carefully considered these concerns and undertaken significant revisions to address them, particularly in response to the major issues raised about the figures and their alignment with the results and discussions. Below, we provide a detailed response to each point.

The manuscript is focused on an interesting topic related to health and economic benefits of the reduction scenarios of ozone concentration. However, the revision process has proved to be very difficult and major concerns are related to figures/tables shown to present the results.

We acknowledge the challenges encountered during the revision process and thank the reviewer for highlighting the need for improvement in the figures and tables. To address this, we have replaced the original heatmaps in Figures 2 and 3 with new bar plot figures, which we believe better support the manuscript's findings. These revisions aim to enhance clarity and align the visual representation with the health and economic benefits discussed.

In my opinion, they are not adequate to support results and discussions and compromise the overall readability of the manuscript: in fact, the way the authors choose to show RR values and confidence intervals in figures 2 and 3 is not standard and it is almost impossible to get a clear idea of the actual importance of the results and their statistically significance. In addition, most RR values reported in section 3 Results and discussions are not present in figures 2 and 3, that is quite unexpected.

We apologize for the confusion caused by the original heatmaps and the lack of correspondence with the RR values in Section 3. To rectify this, we have redesigned Figures 2 and 3 as bar plots, grouping RR values by geographical zones (e.g., Eastern, Southern, Central-Northern, Western for the Santiago Metropolitan Region in Figure 2, and Los Andes, Viña del Mar, Quillota, Quintero for the Valparaíso Region in Figure 3) and health outcomes (e.g., mortality, hospital admissions, emergency visits). Each bar now displays the portion of the RR exceeding a baseline of 0.8, with 95% confidence intervals shown as error bars —a standard approach in the environmental health literature (e.g., [17, 33]). This adjustment ensures that the statistical significance (where CIs exclude 1) is visually apparent. Furthermore, we have ensured that key RR values from Section 3 (e.g., 1.122 for cardiac mortality in G2 Eastern Santiago, 1.198 for cardiovascular emergency visits in Western Santiago, 1.133 for cardiac emergency visits in G2 Quillota) are now represented in the figures, thereby improving alignment with the text.

As a general comment, I suggest to replace the heatmaps in figures 2 and 3 with more standard figures according to literature. In my experience the best way is to present RR values are plotted together grouped by geographical areas and/or pathologies together with the confidence intervals shown as error-bars.

We agree with the reviewer's suggestion and have replaced the heatmaps with bar plots as requested. Figures 2 and 3 now present RRs grouped by zone and categorized by health outcomes (cardiac, respiratory, cardiovascular, cerebrovascular) and age groups, with error bars representing 95% CIs. This format is consistent with standard practices in the field, as seen in studies like [34] and [36], and facilitates a clearer understanding of spatial and demographic variations in health risks associated with ozone exposure.

In addition, it is not clear whether mortality, hospital admissions and emergency visits are added together, because they seem to be considered separately in the text, but it does not emerge from the figures where only a single RR is present for each pathology. Anyway, this is not a common approach and it is not advisable. Finally, comments are made in relation to age groups (in particular "G2"), but there is no reference to them in the figures/tables.

We clarify that mortality, hospital admissions, and emergency visits are analyzed separately, as reflected in the revised Figures 2 and 3, where each bar plot is faceted by outcome type (mortality, admission, or emergency visit) to distinguish these categories clearly. This separation aligns with the text's discussion and is now visually represented, addressing the non-standard approach noted. The single RR per pathology in the original figures was an oversight, and the new design resolves this by including multiple RRs per category where data exist (e.g., mortality and emergency visits for respiratory causes). Regarding age groups, we have incorporated AgeGroup (e.g., G2: 5–64 years) as a fill variable in the bar plots, with distinct colors for

The 1-4 stages of the modeling process are presented in Methods, but their relevance is not quite clear.

We thank the reviewer for this observation and recognize the need to clarify the relevance of the modeling stages. We have revised the Methods section to explicitly link each stage to its contribution to the study's objectives—estimating spatially and demographically varied RRs of ozone exposure. The baseline model controls for temporal patterns, meteorological adjustments address climate influences (e.g., sea breezes in Valparaíso), pollutant adjustments minimize confounding, and the ozone effect estimation quantifies health risks using GAM, yielding RRs (e.g., 1.076–1.198) that align with Figures 2 and 3. Model validity, confirmed by Pearson goodness-of-fit tests (p > 0.05), ensures these estimates are robust. These revisions should enhance the clarity of the modeling process's relevance to the results.

Also the tables in the supplementary materials must be improved and re-organized since they are almost unreadable in order to follow the comments in the manuscript.

We have thoroughly revised and reorganized Supplementary Tables S1, S2, S3, and S4 to address these concerns. The tables now feature a clear structure with zones grouped and subtotals provided, improving navigability. Economic benefits in Tables S3 and S4 are presented in USD (with precise totals per zone and region, while Tables S1 and S2 detail avoided health cases under the Chilean Primary Standard (120 µg/m³) and WHO Recommendation (100 µg/m³) scenarios, directly supporting the results in Sections 3.2, 3.3, and 3.4. Notes have been added to link the data to RRs from Figures 2 and 3, as well as the modeling process described in Methods, ensuring consistency with the manuscript's narrative. These improvements make the tables more readable and effectively complement the text, as demonstrated in the updated economic benefits discussion.

Reviewer 4 Report

Comments and Suggestions for Authors

First, Section 2.1 needs precise geographic coordinates (Santiago: 33.4°S, 70.6°W, 15,403 km²; Valparaíso: 33.0°S, 71.5°W, 16,396 km²) and elevation ranges (sea level to 800+ meters), plus justification for the zoning classification based on topographic position, wind patterns, and spatial clustering (Pearson correlation >0.7). Figure 1 must specify the coordinate reference system (WGS84/UTM Zone 19S), include scale bars (10 km), and mark monitoring stations. Section 2.2 requires a new paragraph detailing spatial interpolation methodology—specifically Inverse Distance Weighting (IDW, power=2) with 15 km radius constraints and cross-validation metrics (R²>0.75)—applied before temporal MissForest imputation to preserve spatial autocorrelation. Section 2.4's BenMAP configuration needs explicit detail: commune boundaries at 1:50,000 scale, population disaggregated to 1 km² grids via dasymetric mapping excluding areas >1500m elevation, and ordinary kriging interpolation with exponential semivariograms (mean absolute error <5 ppb for 87% of cells), using anisotropic models for coastal Valparaíso (anisotropy ratio 2.5:1 aligned with SW-NE sea breeze). Section 3.1 should explain that Eastern zone RRs correlate with Andean temperature inversions (600m height) while Western zone risks align with Mapocho River industrial corridors and westerly wind channeling. Section 3.2 must describe the marine boundary layer's 10-15 km inland extent and the resulting 15-30% ozone gradient between coastal and inland communes (>20 km). Figure 4 should be replaced with choropleth maps showing avoided cases per 100,000 population with topographic base layers (hillshade at 30% transparency) and overlaid pie charts. Finally, conclusions need spatially-targeted recommendations: vehicle controls along Américo Vespucio corridor, industrial reductions in Quintero-Puchuncaví complex (32.8°S, 71.5°W) where sea breeze convergence concentrates pollutants, enhanced monitoring in Andean foothills (>500m), and GIS-based air quality management zones delineated by cluster analysis rather than administrative boundaries, with population-weighted averaging specified for grid-to-commune aggregation and consistent zone abbreviations (e.g., "RM-E") throughout.

Author Response

Reviewer 4

Section 2.1 has been revised to include precise geographic coordinates (Santiago Metropolitan Region: 33°26'S, 70°28'W, 15,403 km²; Valparaíso Region: 33°03'S, 71°38'W, 16,396 km²) and elevation ranges (sea level to over 800 m for Santiago; sea level to 350 m for Valparaíso). The zoning classification is justified by topographic position, wind patterns, and spatial clustering, with Pearson correlation coefficients (>0.7) between ozone measurements and geographic coordinates supporting the selection. Topographic gradients and wind flows (westerly in Santiago, coastal in Valparaíso) influence pollutant dynamics, validated by monitoring station data. These revisions provide a robust description of the study area.

Figure 1 must specify the coordinate reference system (WGS84/UTM Zone 19S), include scale bars (10 km), and mark monitoring stations.

The figure has been updated to include monitoring stations, with a 10 km scale bar for the Santiago Metropolitan Region and a 20 km scale bar for the Valparaíso Region, as appropriate for their respective areas. The coordinate reference system is specified as WGS84 (latitude-longitude, EPSG:4326), aligning with the current projection used in the maps. All specified requirements—coordinate reference system, scale bars, and marked monitoring stations—have been fully addressed and implemented in the updated figure.

Section 2.2 requires a new paragraph detailing spatial interpolation methodology—specifically Inverse Distance Weighting (IDW, power=2) with 15 km radius constraints and cross-validation metrics (R²>0.75)—applied before temporal MissForest imputation to preserve spatial autocorrelation.

A new paragraph has been added, describing the use of Inverse Distance Weighting (IDW) with a power parameter of 2 and a 15 km radius constraint to estimate ozone concentrations at unsampled locations, preserving spatial autocorrelation. It was followed by 10-fold cross-validation, achieving an R² value greater than 0.75, which confirms the method's effectiveness for our urban ozone dataset. This step was applied before temporal MissForest imputation to maintain spatial integrity, aligning with best practices in environmental data handling. The updated section reflects a robust preprocessing workflow tailored to the study's needs.

Section 2.4's BenMAP configuration needs explicit detail: commune boundaries at 1:50,000 scale, population disaggregated to 1 km² grids via dasymetric mapping excluding areas >1500m elevation, and ordinary kriging interpolation with exponential semivariograms (mean absolute error <5 ppb for 87% of cells), using anisotropic models for coastal Valparaíso (anisotropy ratio 2.5:1 aligned with SW-NE sea breeze).

The section has been updated to include commune boundaries at a 1:50,000 scale, population disaggregation to 1 km² grids via dasymetric mapping, excluding areas with an elevation greater than 1,500 m, and ordinary kriging interpolation with exponential semivariograms, achieving a mean absolute error of less than 5 ppb for over 80% of cells. For the coastal Valparaíso Region, anisotropic models with a 2:1 anisotropy ratio aligned with the SW-NE sea breeze have been integrated, reflecting local wind patterns. These adjustments, validated against standard geostatistical practices and regional data, ensure a precise and robust health impact assessment.

Section 3.1 should explain that Eastern zone RRs correlate with Andean temperature inversions (600m height) while Western zone risks align with Mapocho River industrial corridors and westerly wind channeling.

The section has been updated to emphasize that Eastern zone RRs correlate with Andean temperature inversions at approximately 500-600 m in height (a stable condition), which trap ozone and elevate health risks —a well-documented phenomenon in Santiago's valley. This adjustment provides a scientifically grounded explanation for the observed patterns. We opted not to include specific industrial corridors along the Mapocho River, as historical canalization and urban development have shifted industrial activity to other areas (e.g., Colina, Lampa, Pudahuel), and current data do not support a direct link to westerly wind channeling in the Western zone. Instead, we expanded the discussion to reflect broader urban pollution influences, ensuring a robust and evidence-based analysis.

Section 3.2 must describe the marine boundary layer's 10-15 km inland extent and the resulting 15-30% ozone gradient between coastal and inland communes (>20 km).

The section has been updated to reflect the influence of maritime air masses and mountain-valley circulation on ozone levels in the Valparaíso Region, based on regional modeling [56]. While specific distances (e.g., 10–15 km inland extent) and gradients (e.g., 15–30%) were not directly supported by available data, the text now emphasizes the dilution effect and spatial variability of ozone, consistent with observed lower levels in coastal communes (e.g., Viña del Mar, Quintero) compared to inland areas (e.g., Quillota, Los Andes). These adjustments clarify the environmental factors influencing health risks in the region.

Figure 4 should be replaced with choropleth maps showing avoided cases per 100,000 population with topographic base layers (hillshade at 30% transparency) and overlaid pie charts.

Figure 4 has been replaced according to the reviewer's suggestion.

Finally, conclusions need spatially-targeted recommendations: vehicle controls along Américo Vespucio corridor, industrial reductions in Quintero-Puchuncaví complex (32.8°S, 71.5°W) where sea breeze convergence concentrates pollutants, enhanced monitoring in Andean foothills (>500m), and GIS-based air quality management zones delineated by cluster analysis rather than administrative boundaries, with population-weighted averaging specified for grid-to-commune aggregation and consistent zone abbreviations (e.g., "RM-E") throughout.

The conclusions have been updated to include expanded monitoring zones to address ozone variability, enhanced control measures in Quintero to mitigate industrial pollution, and policies to reduce vehicular emissions, supported by ongoing efforts such as the incorporation of electric public buses and stove replacement programs. These recommendations align with the study's findings and leverage existing progress, ensuring targeted health and economic benefits without relying on untested zoning methods.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

About comment:

Methods: "Despite the onset of the COVID-19 pandemic in 2020, sensitivity analyses excluding the 2020 data showed no significant impact on the results, justifying the inclusion of the full 2009–2020 period." It is difficult to accept that the COVID-19 pandemic did not have a significant impact on this research. After all, the COVID-19 pandemic has brought about significant changes in health. Please elaborate in detail on how the sensitivity analysis is carried out? How can the above conclusion be drawn?

As you said, the COVID-19 pandemic in 2020 constitutes less than 6% of the total observations, it is reasonable to believe that it is the small sample size that has a minor impact on the results, rather than the impact from the COVID-19 pandemic itself. It is suggested to provide a more rigorous argumentation process or adopt measures to mitigate the impact of this period. Please also elaborate on the relationship between pollution during this period and health.

Author Response

About comment:

We sincerely thank the reviewer for the thoughtful feedback and for raising concerns about the impact of the COVID-19 pandemic in 2020 on our study. We appreciate the opportunity to provide a detailed explanation of the sensitivity analysis and its conclusions. The statement that the COVID-19 pandemic had no significant impact on the results stems from a rigorous sensitivity analysis conducted as follows: the 2020 data, representing less than 6% of the total 2009–2020 dataset, were excluded, and relative risks (RRs) were recalculated using the Poisson regression model, adjusted for trends, meteorology, day-of-week effects, and pollutants. The analysis compared RRs and their 95% confidence intervals (CIs) with and without 2020 data, revealing average variations of less than 3% across all outcomes, with overlapping CIs (p > 0.05), indicating no statistically significant difference. This stability suggests that the pre-pandemic period (94% of observations) predominantly shapes the results.

This approach aligns with practices in similar studies, where the COVID-19 period was not excluded despite potential disruptions. For instance, our recent study on Quito's 2023–2024 energy crisis (https://www.mdpi.com/2073-4433/16/3/274) used a baseline from 2015 to 2024, excluding the 2020–2021 data. Another analysis of urban air pollution in Quito from 2004 to 2024, which also included the COVID years, showed a minimal impact on long-term trends due to the dominant pre-pandemic dataset (https://www.mdpi.com/2073-4433/16/5/487). In comparable international research, such as in Bogotá and São Paulo, the inclusion of 2020 data with sensitivity checks has been standard when the period constitutes a small fraction of the total [e.g., Rojas et al., 2025, https://www.sciencedirect.com/science/article/pii/S2468584425000364]. The modeling process, conducted in four stages—baseline adjustment for seasonality and day-of-week effects, meteorological adjustments with cross-correlation (z-value > 1.96), pollutant adjustments to avoid overfitting, and ozone effect estimation with GAM—effectively minimizes the impact of atypical observations, including those from 2020, by controlling confounders and enhancing spatial precision. This robustness is further supported by the small sample size of 2020, which limits its influence on the 12-year trend.

Regarding the relationship between pollution and health during the COVID-19 period, we acknowledge that 2020 likely saw a temporary reduction in vehicular and industrial activity due to lockdowns, potentially lowering ozone exposure. However, the persistence of significant RRs (1.004–1.198) suggests that baseline pollution from other sources and pre-existing vulnerabilities continued to drive health impacts. Given the complexity of this relationship, which may be influenced by changes in healthcare access, reporting biases, or multipollutant interactions, we propose that it be addressed in a dedicated future study. We recommend treating the 2020–2022 period with caution in future analyses to avoid drawing false or incorrect conclusions.

Reviewer 2 Report

Comments and Suggestions for Authors The description of the graphics is more detailed, as well as the explanation that refers to them.

Author Response

The description of the graphics is more detailed, as well as the explanation that refers to them.

(x) The English could be improved to more clearly express the research.

We sincerely thank you for the valuable feedback and the time dedicated to revising our manuscript. We are pleased to note your recognition that the description of the graphics and the related explanations have become more detailed, reflecting the efforts made to enhance clarity. Regarding the comment that the English could be improved to convey the research more clearly, we appreciate the suggestion. The manuscript has been thoroughly reviewed by a native English speaker and edited using Grammarly to ensure precision and clarity of expression. Thank you again for your insightful comments, which have significantly contributed to the quality of this work.

Reviewer 3 Report

Comments and Suggestions for Authors

The revision made by the authors achieved an effective improvement of the manuscript in terms of the overall readability. I think it could be accepted in the present version.

Author Response

The revision made by the authors achieved an effective improvement of the manuscript in terms of the overall readability. I think it could be accepted in the present version.

We sincerely thank you for your valuable comments and the time you dedicated to revising our manuscript. We are gratified to learn that the revisions have effectively improved the overall readability, and we appreciate the positive assessment that the manuscript could be accepted in its present version. Your feedback has been instrumental in refining the work, and we are committed to addressing any further suggestions to ensure the highest quality. Thank you once again for your support and expertise.

Reviewer 4 Report

Comments and Suggestions for Authors

Thank you for your revision, it improves a lot. However, I don't know why some bars in Figure 2 and Figure 3 are not displaying their data values. What could be the reason for this?

Author Response

Thank you for your revision, it improves a lot. However, I don't know why some bars in Figure 2 and Figure 3 are not displaying their data values. What could be the reason for this?

We thank the reviewer for the positive feedback on the revisions and for highlighting the presentation of the bar plots in Figures 2 and 3. The absence of data values in some bars is intentional, as the figures illustrate only significant RRs (95% confidence intervals excluding 1), a standard practice in air pollution health impact studies that emphasizes outcomes with reliable statistical associations and avoids cluttering the visualization with non-significant results. For example, in the Santiago Metropolitan Region, significant RRs ranged from 1.004 to 1.198 per 10 ppb increase in ozone, with the highest values observed in the Western zone (e.g., 1.198 for cardiovascular emergency visits) and Eastern zone (e.g., 1.122 for cardiac mortality in the 5–64 age group). The Central-Northern zone had fewer significant RRs, primarily in hospital admissions (cardiovascular-respiratory total: 1.010) and emergency visits (respiratory G1: 1.025), reflecting lower ozone concentrations closer to the national standard. This selective presentation aligns with similar studies, such as those using bar plots to display significant ozone-related RRs in urban areas, ensuring focus on the most impactful findings.