the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 08 Mar 2023
Projection of snowfall extremes in the French Alps as a function of elevation and global warming level
Erwan Le Roux
Guillaume Evin
Nicolas Eckert
Juliette Blanchet
Samuel Morin
Abstract. Following the projected increase in extreme precipitation, an increase in extreme snowfall may be expected in cold regions, e.g. for high latitudes or at high elevations. By contrast, in low/medium elevation areas, the probability to experience rainfall instead of snowfall is generally projected to increase due to warming conditions. In mountainous areas, despite the likely existence of these contrasted trends according to elevation, changes in extreme snowfall with warming remain poorly quantified. This paper assesses projected changes in heavy and extreme snowfall, i.e. in mean annual maxima and 100-year return levels, in the French Alps as a function of elevation and global warming level using a recent methodology based on non-stationary extreme value models. This methodology is applied to an ensemble of 20 adjusted GCM-RCM pairs from the EURO-CORDEX experiment under the scenario RCP8.5. available for each of the 23 massifs of the French Alps from 1951 to 2100, and every 300 m of elevations. Results are provided as relative or absolute changes computed w.r.t. current climate conditions, at the massif scale and averaged over all available massifs. Overall, mean annual maxima are projected to decrease below 3000 m and increase above 3600 m, while 100-year return levels are projected to decrease below 2400 m and increase above 3300 m. At elevations in between, values are on average projected to increase until +3 °C of global warming, and then decrease. At +4 °C, average relative changes in mean annual maxima and 100-year return levels respectively vary from −26 % (−7 kg m−2) and −15 % (−11 kg m−2) at 900 m, to +3 % (+3 kg m−2) and +8 % (+13 kg m−2) at 3600 m. Finally, for each global warming level, we compute the elevation threshold that separates contrasted trends, i.e. where the average relative change equals zero. This elevation threshold is projected to rise between +1.5 °C and +4 °C: from 3000 m to 3350 m for mean annual maxima, and from 2600 m to 3000 m for 100-year return levels. These results have implications for the management of risks related to extreme snowfall.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(2770 KB)
-
Supplement
(56 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2770 KB) - Metadata XML
-
Supplement
(56 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Erwan Le Roux et al.
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-335', Anonymous Referee #1, 29 Mar 2023
Comments to the Authors
This work presents an analysis of the impact of high-emission scenarios on future extreme snowfall in the French Alps. The results are analyzed with respect to elevation and global warming levels, using a recent methodology based on non-stationary extreme value models. The authors have observed a general increase in extreme snowfall at elevations above 3600 m, but a decrease at lower elevations. This study is interesting and useful in helping to understand how snowfall patterns may respond to climate change. Overall, the methods employed in this research are robust, and the results are well-founded. However, the authors have noted data uncertainties at high elevations, which are difficult to overcome, given that the S2M data-assimilation scheme has already assimilated the best snow observations in the French Alps.
The manuscript is well-conceived and organized, and the findings are worthy of publication in The Cryosphere after undergoing an English proofreading. I am willing to accept the manuscript for publication as soon as you address the following suggestions.
Major comments:
I would appreciate a more in-depth discussion of your results. I recommend discussing your values with the results found by other authors using different snow scenarios, projections, and indicators. You should also discuss the seasonality of snowfall, the physical reasons for the changes you detected, the implications of your results, and why it is important to analyze extreme snowfall. A thorough review of the results found by other authors in the Alps, Pyrenees, Apennines, and nearby mountain ranges could complete the study and make it more relevant.
Comments:
Abstract:
L8: Please specify the duration (months) of the season you analyzed.
L8 – L9: I suggest briefly showing the projected changes in temperature and precipitation under RCP8.5.
L9: Please determine the minimum and maximum elevation range you analyzed.
L9: Please avoid abbreviations such as w.r.t. in the abstract.
L10: Please specify what massifs are (i.e., mountain zones with similar meteorological conditions) when you mention "available massifs."
L12: Please briefly show the spatial differences found (if any) in your study.
L14: Please avoid mentioning absolute values here.
L15: Please include the statement "Finally, for each global warming level..." between L5 and L8, where you briefly explain the methods.
L16: "The elevation threshold is projected to rise between +1.5◦C and +4ºC: from 3000 m to 3350 m" to address the question. There are no changes if warming is <= 1.5ºC ? and for > 4ºC?, please rephrase.
Introduction
L35: I recommend describing the main conclusions of these works.
L49: A better justification is needed to highlight the innovative aspects of this study, distinguishing it from previous works, since a similar degree approach methodology has been previously applied (such as Verfaille et al. (2017) in The Cryosphere).
Considering that the introduction, Figure 2, and Table 1 are similar to LeRoux et al. (2021), I suggest providing a more comprehensive review of past and future snow trends in mid-latitude mountain areas, including complementing extreme snowfall trends with other snowpack projections. The questions that this study aims to answer should be introduced in the last sentences of the introduction.
Data and methods
L57. A detailed explanation is needed to clarify why the S2M works by 300 m and massifs.
The ADAMONT methods should be described more comprehensively, including the weather types included and the rationale for selecting quantile-mapping.
Results
The article's content needs further editing in several places. For instance, the description of the data and methods should avoid using the term "observations" since the data analyzed are not observed data (L103). Sections 4.1 and 4.2 should describe the maximum and minimum changes for both indicators, as well as any seasonal and spatial differences, and provide an explanation for why the study focuses on 100-year return levels (L124).
L127: Please correct the typographical errors in your manuscript.
L140: The article could be improved by showing the monthly and spatial differences in the results, including whether the increases at high elevations are expected for winter and whether these trends are consistent across seasons. Finally, there are several lines where you mentioned specific massifs (i.e., Mercantour massif). For a non-local reader, it can be challenging to follow.
In the methodology section, you described the S2M process as occurring in 300-meter increments. However, at a certain point, you changed to four single elevations (as shown in Figures 2 and 3). Could you please provide an explanation for this change and ensure consistency in the methodology?. A statement in the methodology section or results would be appreciated.
Figure 3, it would be beneficial to include the values inside the massifs.
Discussion: Data and methods, section 5.1 and 5.2.
It is recommended that you acknowledge the limitations of RCMs, quantile mapping, and the lack of observations in mountainous regions. It would be essential to better discuss how you overcame these limitations, as well as any differences between previous extreme values metrics. I reccomend to provide more information on the ADAMONT method and the spatial differences, such as the number of weather types categorized (and if the bias correction was the same all massifs and months). Additionally, you used quantile mapping for statistical adjustment and an analogous approach for sub-daily disaggregation. It is suggested that you at least mention the limitations and uncertainties of a simple quantile mapping approach. Please see a review by Maraun et al. (2017) in Nature Climate Change.
Since one of the principal conclusions of this study is the projected increase in extreme snowfall at high elevations, the manuscript can be enhanced by including a section where the authors describe the limitations of the S2M reanalysis and how it could impact the obtained results. The authors should further explain the differences in irradiance as reported by Quéno et al. (2017), precipitation (Vernay et al., 2019) or snow (Vionnett et al., 2019). Additionally, the differences between high-resolution non-hydrostatic models used in other studies, such as Musselman et al. (2017) in Nature, could be discussed.
Discussion: Results, section 5.3
It is recommended to expand the discussion and compare the results with previous studies. There are relevant differences between snow indicator (i., snowfall and snow melt rate) see Musselman et al. (2017) in Nature that will be worth to include.
In addition, the authors should briefly describe previous findings in the Alps, such as those reported by Marty et al. (2017) in The Cryosphere, Piazza et al. (2014) in Climatic Change, Terzago et al. (2017) in The Cryosphere, and Steger et al. (2014) in Climate Dynamics, among others. The authors should also address precipitation seasonality, as presented by Kotlarski et al. (2022) in Climate Dynamics, and explain how snow changes are exceptional within a long-term perspective, as shown by Carrer et al. (2023) in Nature Climate Change. Differences between CMIP 5 and 6 and by scenario should be noted. The authors should mention how recent increases in winter extreme snowfall events counterbalanced summer glacier ablation in Italian alpine regions, as reported by Colucci et al. (2021) in Water.
L215: You can include a review by Faranda et al. (2020) in Weather and Climate Dynamics, and references therein, who discuss recent snowfall changes due to thermodynamics and changes in the atmospheric circulation.
L225: It is therefore essential to acknowledge previous research on natural climate variability at high elevations in the Swiss Alps, such as the work of Willibald et al. (2020) in The Cryosphere and the references therein.
L232: Additionally, it is highly recommended that the potential environmental and social impacts of such projections be discussed.
L237: It is advised to remove quotes from the conclusion section. Finally, to promote transparency and reproducibility, it is recommended to include a statement on the availability of data and code.
Citation: https://doi.org/10.5194/egusphere-2023-335-RC1 - AC1: 'Reply on RC1', Erwan Le Roux, 05 Jul 2023
-
RC2: 'Comment on egusphere-2023-335', Anonymous Referee #2, 25 May 2023
Le Roux et al. analyze extreme future snowfall in the European Alps based on a recent developed methodology, which the authors already applied to snow loads. Overall, the paper is well-written and has high-quality figures. The aims of the study are highly relevant from a societal point of view. The author’s analysis offers novel insights into future snowfall changes, their spatial and elevational patterns. In summary, it is worthy of publication in TC after addressing some major issues as outlined below.
Major points
Methodology in general is lacking a lot of key information:
-
Did you take the snowfall variable directly from the RCMs or how?
-
What exactly was ADAMONT used for and how?
-
What did you use S2M for: as past reference, or only in the bias adjustment? It seems like you concatenated S2M with a GCM-RCM, but please correct me if I’m wrong. And if yes, why did you not use the historical data of the GCM-RCM instead? I would expect some kind of break when merging a reanalysis and a climate model.
GEV approach:
-
I would have expected a few more references on GEV in your description. For readers it would be useful to have an overview paper for general non-stationary GEV, which is what most people use. Then an adaptation to GEV with covariates (what you call non-stationary, but in principle this works with any covariate) and has been developed years ago.
-
Your application to GCM-RCM pairs seems like a prime example of using a random effects specification to reduce the number of parameters while accounting for model variability and interdependence. I’m not up-to-date on recent GEV literature and developments, but I guess there should be some random effects GEV models (not necessarily in your specific field). Would be nice to see this as a discussion point at least.
-
Since almost no natural physical process follows a piecewise linear function with abrupt breakpoints, I wonder why you chose this method? What is the benefit to more flexible approaches like a spline basis, which has a similar number of parameters but does not depend on choosing breakpoints and number of breakpoints (which you note as limitation in the discussion)? Then, even if you did some validation in your snow load paper, I would expect a goodness-of-fit test also here, since snowfall might behave differently than snow load. So as to justify why a piecewise linear function is fitting better than a simple linear model (or one without covariates, by-the-way).
-
Have you assessed the shape parameter of your GEV functions? The shape parameter decides on Gumbel, Weibull, or Frechet type, and each of these have a very different behaviour on tails and expectance of extremes, and your use of adjustment parameters might change the type of GEV in the future – so I was wondering whether you have analyzed this also?
Annual maxima from daily values (? I guess, this is not specified in the manuscript) are notoriously unstable. This deserves some discussion on the challenges in using annual maxima of snowfall, which already suffers from high spatial and temporal variability.
Finally, from a user and climatological point of view, I would say that besides maximum 1-day snowfall, 3-day and 5-day snowfall are of particular interest, even more than the 1-day one. In fact, many assessments of future climate change focus not only on 1day annual maxima but also on 3, and 5 day precipitation extremes (cf. Rx*day indices from ETCCDI). This is beneficial in multiple ways: First it does not introduce an arbitrary break at time=0 for the accumulation of daily values. Second, it also considers changes in circulation regimes, such as increased persistence. Third, problematic recent snowfall extremes in the Alps were often multi-day events. This study could benefit strongly from also including 3 and 5 day extremes in the analysis, making the results both scientifically more robust and with more societal relevance.
Minor points:
-
L58: I suggest not calling a reanalysis an “observation”, not even for simplicity. This can be very misleading for a reader who does not carefully read every detail.
-
Fig1c: Smoothed, how exactly? Please indicate.
-
L127: missing sentence?
-
L129: -40% is not a substantial change?
-
L205: “huge step forward” is not a justified conclusion. Just using a more complicated model does not automatically imply better results. To justify this sentence, I’d expect some tests of goodness-of-fit (see also related comment above).
Citation: https://doi.org/10.5194/egusphere-2023-335-RC2 - AC2: 'Reply on RC2', Erwan Le Roux, 05 Jul 2023
-
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-335', Anonymous Referee #1, 29 Mar 2023
Comments to the Authors
This work presents an analysis of the impact of high-emission scenarios on future extreme snowfall in the French Alps. The results are analyzed with respect to elevation and global warming levels, using a recent methodology based on non-stationary extreme value models. The authors have observed a general increase in extreme snowfall at elevations above 3600 m, but a decrease at lower elevations. This study is interesting and useful in helping to understand how snowfall patterns may respond to climate change. Overall, the methods employed in this research are robust, and the results are well-founded. However, the authors have noted data uncertainties at high elevations, which are difficult to overcome, given that the S2M data-assimilation scheme has already assimilated the best snow observations in the French Alps.
The manuscript is well-conceived and organized, and the findings are worthy of publication in The Cryosphere after undergoing an English proofreading. I am willing to accept the manuscript for publication as soon as you address the following suggestions.
Major comments:
I would appreciate a more in-depth discussion of your results. I recommend discussing your values with the results found by other authors using different snow scenarios, projections, and indicators. You should also discuss the seasonality of snowfall, the physical reasons for the changes you detected, the implications of your results, and why it is important to analyze extreme snowfall. A thorough review of the results found by other authors in the Alps, Pyrenees, Apennines, and nearby mountain ranges could complete the study and make it more relevant.
Comments:
Abstract:
L8: Please specify the duration (months) of the season you analyzed.
L8 – L9: I suggest briefly showing the projected changes in temperature and precipitation under RCP8.5.
L9: Please determine the minimum and maximum elevation range you analyzed.
L9: Please avoid abbreviations such as w.r.t. in the abstract.
L10: Please specify what massifs are (i.e., mountain zones with similar meteorological conditions) when you mention "available massifs."
L12: Please briefly show the spatial differences found (if any) in your study.
L14: Please avoid mentioning absolute values here.
L15: Please include the statement "Finally, for each global warming level..." between L5 and L8, where you briefly explain the methods.
L16: "The elevation threshold is projected to rise between +1.5◦C and +4ºC: from 3000 m to 3350 m" to address the question. There are no changes if warming is <= 1.5ºC ? and for > 4ºC?, please rephrase.
Introduction
L35: I recommend describing the main conclusions of these works.
L49: A better justification is needed to highlight the innovative aspects of this study, distinguishing it from previous works, since a similar degree approach methodology has been previously applied (such as Verfaille et al. (2017) in The Cryosphere).
Considering that the introduction, Figure 2, and Table 1 are similar to LeRoux et al. (2021), I suggest providing a more comprehensive review of past and future snow trends in mid-latitude mountain areas, including complementing extreme snowfall trends with other snowpack projections. The questions that this study aims to answer should be introduced in the last sentences of the introduction.
Data and methods
L57. A detailed explanation is needed to clarify why the S2M works by 300 m and massifs.
The ADAMONT methods should be described more comprehensively, including the weather types included and the rationale for selecting quantile-mapping.
Results
The article's content needs further editing in several places. For instance, the description of the data and methods should avoid using the term "observations" since the data analyzed are not observed data (L103). Sections 4.1 and 4.2 should describe the maximum and minimum changes for both indicators, as well as any seasonal and spatial differences, and provide an explanation for why the study focuses on 100-year return levels (L124).
L127: Please correct the typographical errors in your manuscript.
L140: The article could be improved by showing the monthly and spatial differences in the results, including whether the increases at high elevations are expected for winter and whether these trends are consistent across seasons. Finally, there are several lines where you mentioned specific massifs (i.e., Mercantour massif). For a non-local reader, it can be challenging to follow.
In the methodology section, you described the S2M process as occurring in 300-meter increments. However, at a certain point, you changed to four single elevations (as shown in Figures 2 and 3). Could you please provide an explanation for this change and ensure consistency in the methodology?. A statement in the methodology section or results would be appreciated.
Figure 3, it would be beneficial to include the values inside the massifs.
Discussion: Data and methods, section 5.1 and 5.2.
It is recommended that you acknowledge the limitations of RCMs, quantile mapping, and the lack of observations in mountainous regions. It would be essential to better discuss how you overcame these limitations, as well as any differences between previous extreme values metrics. I reccomend to provide more information on the ADAMONT method and the spatial differences, such as the number of weather types categorized (and if the bias correction was the same all massifs and months). Additionally, you used quantile mapping for statistical adjustment and an analogous approach for sub-daily disaggregation. It is suggested that you at least mention the limitations and uncertainties of a simple quantile mapping approach. Please see a review by Maraun et al. (2017) in Nature Climate Change.
Since one of the principal conclusions of this study is the projected increase in extreme snowfall at high elevations, the manuscript can be enhanced by including a section where the authors describe the limitations of the S2M reanalysis and how it could impact the obtained results. The authors should further explain the differences in irradiance as reported by Quéno et al. (2017), precipitation (Vernay et al., 2019) or snow (Vionnett et al., 2019). Additionally, the differences between high-resolution non-hydrostatic models used in other studies, such as Musselman et al. (2017) in Nature, could be discussed.
Discussion: Results, section 5.3
It is recommended to expand the discussion and compare the results with previous studies. There are relevant differences between snow indicator (i., snowfall and snow melt rate) see Musselman et al. (2017) in Nature that will be worth to include.
In addition, the authors should briefly describe previous findings in the Alps, such as those reported by Marty et al. (2017) in The Cryosphere, Piazza et al. (2014) in Climatic Change, Terzago et al. (2017) in The Cryosphere, and Steger et al. (2014) in Climate Dynamics, among others. The authors should also address precipitation seasonality, as presented by Kotlarski et al. (2022) in Climate Dynamics, and explain how snow changes are exceptional within a long-term perspective, as shown by Carrer et al. (2023) in Nature Climate Change. Differences between CMIP 5 and 6 and by scenario should be noted. The authors should mention how recent increases in winter extreme snowfall events counterbalanced summer glacier ablation in Italian alpine regions, as reported by Colucci et al. (2021) in Water.
L215: You can include a review by Faranda et al. (2020) in Weather and Climate Dynamics, and references therein, who discuss recent snowfall changes due to thermodynamics and changes in the atmospheric circulation.
L225: It is therefore essential to acknowledge previous research on natural climate variability at high elevations in the Swiss Alps, such as the work of Willibald et al. (2020) in The Cryosphere and the references therein.
L232: Additionally, it is highly recommended that the potential environmental and social impacts of such projections be discussed.
L237: It is advised to remove quotes from the conclusion section. Finally, to promote transparency and reproducibility, it is recommended to include a statement on the availability of data and code.
Citation: https://doi.org/10.5194/egusphere-2023-335-RC1 - AC1: 'Reply on RC1', Erwan Le Roux, 05 Jul 2023
-
RC2: 'Comment on egusphere-2023-335', Anonymous Referee #2, 25 May 2023
Le Roux et al. analyze extreme future snowfall in the European Alps based on a recent developed methodology, which the authors already applied to snow loads. Overall, the paper is well-written and has high-quality figures. The aims of the study are highly relevant from a societal point of view. The author’s analysis offers novel insights into future snowfall changes, their spatial and elevational patterns. In summary, it is worthy of publication in TC after addressing some major issues as outlined below.
Major points
Methodology in general is lacking a lot of key information:
-
Did you take the snowfall variable directly from the RCMs or how?
-
What exactly was ADAMONT used for and how?
-
What did you use S2M for: as past reference, or only in the bias adjustment? It seems like you concatenated S2M with a GCM-RCM, but please correct me if I’m wrong. And if yes, why did you not use the historical data of the GCM-RCM instead? I would expect some kind of break when merging a reanalysis and a climate model.
GEV approach:
-
I would have expected a few more references on GEV in your description. For readers it would be useful to have an overview paper for general non-stationary GEV, which is what most people use. Then an adaptation to GEV with covariates (what you call non-stationary, but in principle this works with any covariate) and has been developed years ago.
-
Your application to GCM-RCM pairs seems like a prime example of using a random effects specification to reduce the number of parameters while accounting for model variability and interdependence. I’m not up-to-date on recent GEV literature and developments, but I guess there should be some random effects GEV models (not necessarily in your specific field). Would be nice to see this as a discussion point at least.
-
Since almost no natural physical process follows a piecewise linear function with abrupt breakpoints, I wonder why you chose this method? What is the benefit to more flexible approaches like a spline basis, which has a similar number of parameters but does not depend on choosing breakpoints and number of breakpoints (which you note as limitation in the discussion)? Then, even if you did some validation in your snow load paper, I would expect a goodness-of-fit test also here, since snowfall might behave differently than snow load. So as to justify why a piecewise linear function is fitting better than a simple linear model (or one without covariates, by-the-way).
-
Have you assessed the shape parameter of your GEV functions? The shape parameter decides on Gumbel, Weibull, or Frechet type, and each of these have a very different behaviour on tails and expectance of extremes, and your use of adjustment parameters might change the type of GEV in the future – so I was wondering whether you have analyzed this also?
Annual maxima from daily values (? I guess, this is not specified in the manuscript) are notoriously unstable. This deserves some discussion on the challenges in using annual maxima of snowfall, which already suffers from high spatial and temporal variability.
Finally, from a user and climatological point of view, I would say that besides maximum 1-day snowfall, 3-day and 5-day snowfall are of particular interest, even more than the 1-day one. In fact, many assessments of future climate change focus not only on 1day annual maxima but also on 3, and 5 day precipitation extremes (cf. Rx*day indices from ETCCDI). This is beneficial in multiple ways: First it does not introduce an arbitrary break at time=0 for the accumulation of daily values. Second, it also considers changes in circulation regimes, such as increased persistence. Third, problematic recent snowfall extremes in the Alps were often multi-day events. This study could benefit strongly from also including 3 and 5 day extremes in the analysis, making the results both scientifically more robust and with more societal relevance.
Minor points:
-
L58: I suggest not calling a reanalysis an “observation”, not even for simplicity. This can be very misleading for a reader who does not carefully read every detail.
-
Fig1c: Smoothed, how exactly? Please indicate.
-
L127: missing sentence?
-
L129: -40% is not a substantial change?
-
L205: “huge step forward” is not a justified conclusion. Just using a more complicated model does not automatically imply better results. To justify this sentence, I’d expect some tests of goodness-of-fit (see also related comment above).
Citation: https://doi.org/10.5194/egusphere-2023-335-RC2 - AC2: 'Reply on RC2', Erwan Le Roux, 05 Jul 2023
-
Peer review completion
Journal article(s) based on this preprint
Erwan Le Roux et al.
Erwan Le Roux et al.
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 369 | 185 | 25 | 579 | 38 | 11 | 11 |
- HTML: 369
- PDF: 185
- XML: 25
- Total: 579
- Supplement: 38
- BibTeX: 11
- EndNote: 11
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2770 KB) - Metadata XML
-
Supplement
(56 KB) - BibTeX
- EndNote
- Final revised paper