the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 26 Feb 2025
Extreme heat and mortality in the State of Rio de Janeiro in the 2023/24 season: attribution to climate change and ENSO
Abstract. During the 2023/24 season, the State of Rio de Janeiro experienced unprecedented maximum temperatures, resulting in a substantial increase in human mortality. This study aims to analyze the contribution of global warming to changes in the distribution of annual maximum temperatures and their subsequent impact on mortality rates. Our analysis of extreme temperatures reveals that a non-stationary model, in which the location parameter shifts linearly as a function of El Niño-Southern Oscillation (ENSO) and/or climate change, provides a significantly better fit to the data than its stationary counterpart. The northern region of the State exhibited the strongest response to climate change, while ENSO effects were most pronounced in the eastern region. Events as likely as the 2023 heatwave were estimated about 2 °C colder in pre-industrial times. Under a 2 °C global warming scenario, the probability of experiencing 2023-like daytime temperatures increases by at least a factor of three. These findings highlight climate change as the primary driver of extreme temperature intensification, with ENSO acting as a secondary but significant factor in the eastern region. As global warming approaches 2 °C, Rio de Janeiro is projected to experience heatwaves of that magnitude every four years approximately. Climate change has contributed to one in three heat-related deaths recorded during the peak of the event. Without adaptation measures, global warming would further increase the death toll during extreme events of the same frequency as the 2023/24 heat wave.
- Preprint
(2202 KB) - Metadata XML
-
Supplement
(1183 KB) - BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on egusphere-2025-792', Anonymous Referee #1, 19 Apr 2025
Why did authors use the 15% threshold for showing missing data points "less than 15% of missing data in the period 1971-2024 are identified with a black dot and the name?" The second warmest day in 83738 Resende appear to be around 1976 (Figure 2) however it does not match with Table 1 with its reporting in 2023. Similarly, peaks in temperature appears much before as shown in Figure 2 than its reporting in Table 1 (83718 Cordeiro). The ENSO is itself impacted by climate change, then how do authors decouple the separate impact of climate change and ENSO. In addition, without a scientific evidence it is vague to say or quantify how they both impact extreme heat. Mortality is heavily driven by extreme heat in combination with higher humidity which is not at all explored. The figure qualities are also inadequate and not suitable for scientific standard publications.
- AC1: 'Reply on RC1', Soledad Maribel Collazo Inglesini, 23 May 2025
-
RC2: 'Comment on egusphere-2025-792', Anonymous Referee #2, 25 Apr 2025
The manuscript entitled 'Extreme heat and mortality in the State of Rio de Janeiro in the 2023/24 season: attribution to climate change and ENSO' by Soledad Collazo and collaborators is an attribution analysis of the 2023 maximum temperatures in the State of Rio de Janeiro and the associated mortality, based on statistical methods. I think the topic is relevant, the application of this methodology is novel for the region and could help to provide a basis for future attribution studies. However, I think that clarifications in the methodology and in the presentation of the results are needed to improve the quality and impact of the study.
Major commentsMy main concern is with the methodology proposed to take account of the ENSO signal. I have some observations which make me think that the values obtained may be highly dependent on the methodological choices made and that they do not necessarily give an idea of the relationship of the TXx with ENSO.
- According to what you say on p.5. l.132. you consider for each year only the monthly anomaly of this index for the month month determining the TXx value of each year. By considering the index monthly and not seasonally, there are intra-seasonal variations that can influence the results and that do not necessarily account for the phenomenon you are trying to describe, which manifests itself on
predominantly seasonal scales. - Another observation in the same line is that, considering the month in which TXx is obtained and the seasonal cycle of SSTs in the tropical Pacific, the month of the year in which TXx is given can have an important influence.
- What is the period over which the anomalies are calculated, and can there be a bias in recent years due to the observed trend?
- In the link provided I only see data from 1982. Was this the only period used for the adjustment?
- Can the results be improved by considering an ENSO index that takes into account the atmospheric part of ENSO (e.g. SOI or MEI)?
In summary, I think it would be important for you to assess whether you are considering the effect of ENSO in the most appropriate way and that the concerns that I raised above could not significantly change the results. Some discussion of the limitations of methodological choices should also be included in the text.
Table S1 shows that there are large differences between the populations of the different cities for which each station is representative. In this sense, Would it be logical to calculate the mean daily TX for the five weather stations in order to establish a correlation between temperature and mortality? Would it be more appropriate to consider a method of analysing the Tx that incorporates a weighting based on these population differences?
I recognise the effort to obtain results related to the impact of anthropogenic warming on health in such an extreme event and the limitations of data availability, but I wonder how robust the results are if only a 20-year period is considered in order to hypothesise what would happen at other levels of global warming.
Minor comments- Title: As the study finally focuses on the November 2023 event, it seems more correct to me to remove the reference to 2024 from the title and other relevant parts.
- p.1 l. 17-18 change to “as a function of global warming and/or El Niño-Southern Oscillation (ENSO)”
- p.1 l.20-24. I think the words you use in these lines like 'heatwaves' or '2023-like daytime temperatures' are not quite accurate to what you actually assessed in the study. So I recommend you rewrite it.
- p.1 l.25. I suggest also adding ‘mitigation measures’
- p.1 l.30 Please clarify which climatology.
- p.1 l.31 Does November not belong to spring?
- p2. l35-37. The impact of El Niño events on your region of interest is also (and probably more) related to Rossby wave trains.
- p.2 l.43. Please check what is the recommended way to cite a specific chapter of he IPCC AR6.
- p.4 l.99. Table S1 shows that Itaperuna has 15.00% missing data. Is it really below the selection criteria if we have more digits? Moreover, the missing data for this station is relatively high, both for the whole period and for the period of interest, to what extent could this feature affect the results, especially considering that you are working with extremes? I think it would be desirable to have a discussion or a note of caution regarding this.
- p.4 l.117. Please explain further how you apply the block maxima approach..
- p.5. l-123 Please specify the parameters used for the LOESS smoothing.
- p.6 l.174. Please include the references of those studies that use this approach for South America.
- p.7 l.195-199. Can you give some indication of how these numerical values are interpreted?
- p.14 l.333. Where exactly did you get the '25 years' value, is it the mean around the 3 stations? In order to visualise what you said more easily, I think it might help to clearly indicate (with a different colour or higher line width) the horizontal line corresponding to this value in Figure 4.
- p.14 l.355. I think that ‘(Fig. 5c)' is not the correct reference to what you are saying.
- p.17 l.396 "In terms of gender, the usual mortality pattern was observed". Which is the usual mortality pattern?
- p.18 l.414-420. How do you calculate the CI for daily AF? Is it based on the uncertainty of the methodology you used or the uncertainty associated with the estimation of Tx for future conditions? I think it is important to take both uncertainties into account.
- p.19 l.446. ‘...causing the global mean near-surface warming trend in the urban core…”
- p.26 l.666-668 Kew et al (2023) reference is incomplete.
- Faranda & Alberti (2024) reference is missing
- Please unify the spelling of 'heat wave'/'heatwave’
- Figure 1: Add a source for the elevation.
- Figure 2: In the caption, be more specific about what you are plotting.
- As you have only five points, I would suggest replacing Fig. 3, Fig. 5, Fig. S2 and Fig. S4 with tables. This will make it clear which is the exact value of the corresponding variable and also you cloud include there other information such as the CI. Should you consider it worthwhile, you may wish to include a concise description to provide an indication of the location (e.g. 'West', 'South', etc.).
- I suggest that the information in Figure S2 be included as a table in the main text, as it contains key information for the reproduction of the study.
- Figure 4 (and other similars): Could you better explain how you made the shifting of the observations?
- Maybe it is a problem of the generation of the file corresponding to the manuscript, but the figures have some problems of definition for their correct visualisation.
Citation: https://doi.org/10.5194/egusphere-2025-792-RC2 - AC2: 'Reply on RC2', Soledad Maribel Collazo Inglesini, 23 May 2025
- According to what you say on p.5. l.132. you consider for each year only the monthly anomaly of this index for the month month determining the TXx value of each year. By considering the index monthly and not seasonally, there are intra-seasonal variations that can influence the results and that do not necessarily account for the phenomenon you are trying to describe, which manifests itself on
Status: closed
-
RC1: 'Comment on egusphere-2025-792', Anonymous Referee #1, 19 Apr 2025
Why did authors use the 15% threshold for showing missing data points "less than 15% of missing data in the period 1971-2024 are identified with a black dot and the name?" The second warmest day in 83738 Resende appear to be around 1976 (Figure 2) however it does not match with Table 1 with its reporting in 2023. Similarly, peaks in temperature appears much before as shown in Figure 2 than its reporting in Table 1 (83718 Cordeiro). The ENSO is itself impacted by climate change, then how do authors decouple the separate impact of climate change and ENSO. In addition, without a scientific evidence it is vague to say or quantify how they both impact extreme heat. Mortality is heavily driven by extreme heat in combination with higher humidity which is not at all explored. The figure qualities are also inadequate and not suitable for scientific standard publications.
- AC1: 'Reply on RC1', Soledad Maribel Collazo Inglesini, 23 May 2025
-
RC2: 'Comment on egusphere-2025-792', Anonymous Referee #2, 25 Apr 2025
The manuscript entitled 'Extreme heat and mortality in the State of Rio de Janeiro in the 2023/24 season: attribution to climate change and ENSO' by Soledad Collazo and collaborators is an attribution analysis of the 2023 maximum temperatures in the State of Rio de Janeiro and the associated mortality, based on statistical methods. I think the topic is relevant, the application of this methodology is novel for the region and could help to provide a basis for future attribution studies. However, I think that clarifications in the methodology and in the presentation of the results are needed to improve the quality and impact of the study.
Major commentsMy main concern is with the methodology proposed to take account of the ENSO signal. I have some observations which make me think that the values obtained may be highly dependent on the methodological choices made and that they do not necessarily give an idea of the relationship of the TXx with ENSO.
- According to what you say on p.5. l.132. you consider for each year only the monthly anomaly of this index for the month month determining the TXx value of each year. By considering the index monthly and not seasonally, there are intra-seasonal variations that can influence the results and that do not necessarily account for the phenomenon you are trying to describe, which manifests itself on
predominantly seasonal scales. - Another observation in the same line is that, considering the month in which TXx is obtained and the seasonal cycle of SSTs in the tropical Pacific, the month of the year in which TXx is given can have an important influence.
- What is the period over which the anomalies are calculated, and can there be a bias in recent years due to the observed trend?
- In the link provided I only see data from 1982. Was this the only period used for the adjustment?
- Can the results be improved by considering an ENSO index that takes into account the atmospheric part of ENSO (e.g. SOI or MEI)?
In summary, I think it would be important for you to assess whether you are considering the effect of ENSO in the most appropriate way and that the concerns that I raised above could not significantly change the results. Some discussion of the limitations of methodological choices should also be included in the text.
Table S1 shows that there are large differences between the populations of the different cities for which each station is representative. In this sense, Would it be logical to calculate the mean daily TX for the five weather stations in order to establish a correlation between temperature and mortality? Would it be more appropriate to consider a method of analysing the Tx that incorporates a weighting based on these population differences?
I recognise the effort to obtain results related to the impact of anthropogenic warming on health in such an extreme event and the limitations of data availability, but I wonder how robust the results are if only a 20-year period is considered in order to hypothesise what would happen at other levels of global warming.
Minor comments- Title: As the study finally focuses on the November 2023 event, it seems more correct to me to remove the reference to 2024 from the title and other relevant parts.
- p.1 l. 17-18 change to “as a function of global warming and/or El Niño-Southern Oscillation (ENSO)”
- p.1 l.20-24. I think the words you use in these lines like 'heatwaves' or '2023-like daytime temperatures' are not quite accurate to what you actually assessed in the study. So I recommend you rewrite it.
- p.1 l.25. I suggest also adding ‘mitigation measures’
- p.1 l.30 Please clarify which climatology.
- p.1 l.31 Does November not belong to spring?
- p2. l35-37. The impact of El Niño events on your region of interest is also (and probably more) related to Rossby wave trains.
- p.2 l.43. Please check what is the recommended way to cite a specific chapter of he IPCC AR6.
- p.4 l.99. Table S1 shows that Itaperuna has 15.00% missing data. Is it really below the selection criteria if we have more digits? Moreover, the missing data for this station is relatively high, both for the whole period and for the period of interest, to what extent could this feature affect the results, especially considering that you are working with extremes? I think it would be desirable to have a discussion or a note of caution regarding this.
- p.4 l.117. Please explain further how you apply the block maxima approach..
- p.5. l-123 Please specify the parameters used for the LOESS smoothing.
- p.6 l.174. Please include the references of those studies that use this approach for South America.
- p.7 l.195-199. Can you give some indication of how these numerical values are interpreted?
- p.14 l.333. Where exactly did you get the '25 years' value, is it the mean around the 3 stations? In order to visualise what you said more easily, I think it might help to clearly indicate (with a different colour or higher line width) the horizontal line corresponding to this value in Figure 4.
- p.14 l.355. I think that ‘(Fig. 5c)' is not the correct reference to what you are saying.
- p.17 l.396 "In terms of gender, the usual mortality pattern was observed". Which is the usual mortality pattern?
- p.18 l.414-420. How do you calculate the CI for daily AF? Is it based on the uncertainty of the methodology you used or the uncertainty associated with the estimation of Tx for future conditions? I think it is important to take both uncertainties into account.
- p.19 l.446. ‘...causing the global mean near-surface warming trend in the urban core…”
- p.26 l.666-668 Kew et al (2023) reference is incomplete.
- Faranda & Alberti (2024) reference is missing
- Please unify the spelling of 'heat wave'/'heatwave’
- Figure 1: Add a source for the elevation.
- Figure 2: In the caption, be more specific about what you are plotting.
- As you have only five points, I would suggest replacing Fig. 3, Fig. 5, Fig. S2 and Fig. S4 with tables. This will make it clear which is the exact value of the corresponding variable and also you cloud include there other information such as the CI. Should you consider it worthwhile, you may wish to include a concise description to provide an indication of the location (e.g. 'West', 'South', etc.).
- I suggest that the information in Figure S2 be included as a table in the main text, as it contains key information for the reproduction of the study.
- Figure 4 (and other similars): Could you better explain how you made the shifting of the observations?
- Maybe it is a problem of the generation of the file corresponding to the manuscript, but the figures have some problems of definition for their correct visualisation.
Citation: https://doi.org/10.5194/egusphere-2025-792-RC2 - AC2: 'Reply on RC2', Soledad Maribel Collazo Inglesini, 23 May 2025
- According to what you say on p.5. l.132. you consider for each year only the monthly anomaly of this index for the month month determining the TXx value of each year. By considering the index monthly and not seasonally, there are intra-seasonal variations that can influence the results and that do not necessarily account for the phenomenon you are trying to describe, which manifests itself on
Data sets
Temperature in the State of Rio de Janeiro Brazilian National Institute of Meteorology (INMET) https://bdmep.inmet.gov.br/#
Global mean temperature anomalies Met Office https://www.metoffice.gov.uk/hadobs/hadcrut5/
ENSO NOAA https://www.cpc.ncep.noaa.gov/data/indices/sstoi.indices
Mortaliry in the State of Rio de Janeiro Secretaria de Estado de Saude of Rio de Janeiro http://sistemas.saude.rj.gov.br/tabnetbd/dhx.exe?sim/tf_sim_do_geral.def
Model code and software
Non-stationary-GEV in R Soledad Collazo https://doi.org/10.5281/zenodo.13913445
Heat-related mortality analysis L. d. C. M. Ferreira et al. https://doi.org/10.1038/s41598-019-50235-8
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 232 | 89 | 15 | 336 | 21 | 11 | 21 |
- HTML: 232
- PDF: 89
- XML: 15
- Total: 336
- Supplement: 21
- BibTeX: 11
- EndNote: 21
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1