the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 23 Sep 2025
Recent intensification of extreme precipitation over Antarctica driven by increases in greenhouse gases
Abstract. Extreme precipitation is a major contributor to the total precipitation over Antarctica, as well as its variability. However, it's still poorly understood whether any recent trends in extreme precipitation over Antarctica have occurred, and if so, whether they are anthropogenically driven. Here we address this knowledge gap by using ERA5 data from 1979 to 2023 to identify six Antarctic drainage basins with significant positive trends in total and extreme precipitation. These basins include one in the Antarctic Peninsula, one in West Antarctica, and four in East Antarctica. We show that these trends are partly due to an increased occurrence of atmospheric rivers. We subsequently perform a detection and attribution analysis of these trends using precipitation outputs from global climate model CESM2 ensembles that consider all external forcing (ALL), greenhouse gases only (GHG), and anthropogenic aerosols only (AAER). Five of the basins (one in West Antarctica and four in East Antarctica) have good agreement between the trends from the ALL ensemble and ERA5, as well as between the ALL and GHG ensembles, indicating that greenhouse gases are the primary driver of the present-day trends in total and extreme precipitation over these basins. The good agreement between the ALL ensemble and ERA5 trends is confirmed using a regression-based detection and attribution technique. However, regressing the ALL, GHG, and AAER ensembles against ERA5 did not yield robust attribution to any specific single-forcing for either total or extreme precipitation, which is likely due to limitations such as the relatively small ensemble size of the simulations.
- Preprint
(1700 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2025-4292', Anonymous Referee #1, 03 Dec 2025
-
RC2: 'Comment on egusphere-2025-4292', Anonymous Referee #2, 13 Dec 2025
This paper investigates whether recent trends in total and extreme precipitation over Antarctica are detectable and, if so, whether they can be attributed to anthropogenic forcing. Based on ERA5 precipitation data from 1979–2023, they perform a detection&attribution study using CESM2 large-ensemble simulations forced with all external forcings (ALL), greenhouse gases only (GHG), anthropogenic aerosols only (AAER). Because the precipitation trends over six basins where similar in both the ALL and GGH ensembles, they conclude that greenhouse gases are the primary driver of present-day trends in total and extreme precipitation in these regions. However, they note that robust attribution to any single forcing is complex because of the relatively small ensemble size. Finally, they also provide a brief discussion of the potential influence of atmospheric rivers on the observed precipitation changes even though this step is quite disconnected from the rest of the study.
The paper is well written overall, though some parts could read more smoothly. I expected this study to serve as a follow-up to previous work using the CESM1 and CanESM2 models (Dalaiden et al., 2022), with a particular emphasis on intense precipitation. Dalaiden et al (2022) concluded that: “We show that surface climate changes since the 1950s were driven by anthropogenic forcing, in particular the greenhouse gas forcing and stratospheric ozone depletion”. Since the contribution of the Antarctic ozone hole to changes in snow accumulation is well established (e.g., Lenaerts et al., 2018), I found it surprising that the authors rely directly on CESM2 simulations, given that no CESM2 simulations including stratospheric ozone depletion are currently available (Simpson et al., 2023). This choice effectively neglects the influence of the ozone hole, leading the analysis to focus solely on other forcings. Since the authors themselves acknowledge that a robust attribution to any single forcing is not possible, the resulting conclusions on GHG are neither particularly novel nor fully convincing. This is disappointing, as the research group has a good expertise for a D&A analysis, and the study could have been substantially strengthened by including a comparison with CESM1 or CanESM2 simulations. Given that Q. Dalaiden is a co-author of the present study, such a comparison would seem feasible.
Therefore, I have difficulty in identifying the main new contribution of the paper. The link with atmospheric rivers could have been interesting, but it is curretly too limited, and providing a D&A analysis on the changes in atmospheric river frequency and intensity would constitute an entirely new study.
I believe that this paper cannot be published in its present form and requires either an additional study or a clear comparison with previous work to properly account for ozone depletion in the analysis. The simplest approach would likely be to directly apply their method to the simulations presented by Dalaiden et al. (2022) and compare the results with the current findings. I am aware that this represents a significant amount of work, but as it stands, the conclusions are comparatively fragile relative to those of earlier studies.
Major comments:
- A primary concern is the misuse of the term ‘extreme’ in this paper. This issue is not unique to the present work but is widespread in the Antarctic literature. For instance, Turner et al. (2019) or Adusumilli et al. (2021) used this term for precipitation exceeding the 95th percentile, but it should not be considered as extreme events. This is misleading from a statistical perspective, as true extremes follow different statistical laws. Events at the 95th percentile are not rare, they occur roughly 18 times per year. In contrast, annual maxima could reasonably be considered extreme. I recommend replacing the term “extreme” with “intense”. Including a remark on this aspect in the text would also be welcome.
- This comment is related with my first remark in the introduction of this report. The detection step leads to the consideration of only six basins where CESM2 and ERA5 trends are in agreement, whereas basins with negative trends are excluded. However, this exclusion could reflect the model’s inability to reproduce the real processes driving the spatial pattern of accumulation changes in Antarctica. What result woould be obtained if the same approach was based on simulations from Dalaiden et al. (2022). in the latter publication, they show clearly more negative and significant trends in these regions. Overall, the current approach gives the impression of cherry-picking, or that CESM2 misrepresents the impact of anthropogenic forcings on recent Antarctic changes. The latter could reflect that the model induces an artificial overestimation of the precipitation/temperature relationship in CESM2 (Clausius–Clapeyron)? If so, could this in turn bias or exaggerate the inferred link with greenhouse gases? The overall increase in Antarctic snowfall (in figure 2) could reflect that. Could such an effect potentially mask the impact of circulation changes? How are precipitation changes correlated with temperature changes? This issue is further more complex to interpret because, although the trends in the selected basins are similar in absolute value, they occur in basins where the average precipitation is roughly twice as high as in ERA5. If the data were standardized, the trends would appear approximately half as large as those in ERA5.
- The spatial distribution of observed SMB variations is largely linked to trends in the Southern Annular Mode (SAM; Meddley and Thomas, 2019), which is itself strongly influenced by the Antarctic ozone hole. Lenaerts et al. (2018) and, later, Dunmire et al. (2020) also suggested this role of the ozone hole. Moreover, in ALL, if I am not wrong, the impact of ozone is considered, but its effect on precipitation distribution is weaker than in ERA5 ; it is also the case if compared to the simulations from Dalaiden et al. (2022). Could you justify why these significant differences are observed ? I suggest performing the same D&A analysis using CESM1 or CanESM2 simulations provided by Dalaiden et al. (2022), particularly since Q. Dalaiden is a co-author. At least providing more accurate comparisons with previous studies is necessary.
- Atmospheric rivers (ARs) are only briefly addressed in the paper, which focuses on quite different aspects. Important methodological information on AR precipitation calculations is missing in the text. Calculation of their role in precipitation trends should be improved (or better explained). Moreover, the authors did not analyze whether changes in trends resulted from changes in frequency, intensity (or both), and what was the cause of these changes. Consequently, in its current form, the study of ARs in the paper appears somewhat artificial. It would have been interesting to perform a D&A analysis on AR changes, but this would constitute a significantly different paper. I rather suggest to remove this part of the study.
Other specific comments:
Line 27: “Extreme precipitation events (…) have increased in intensity and frequency globally in recent decade (…) which has been attributed to anthropogenically-induced climate change”. I find this sentence a bit too affirmative. In chapter 11 from the IPCC, AR6, they write that “Regional increases in the frequency and/or intensity of heavy precipitation have been observed with at least medium confidence for nearly half of AR6 regions” they also write that “The assessment focuses on land regions excluding Antarctica.”. This indicates that there is still considerable work to be done, particularly in regions where the time series are too short to clearly demonstrate changes or trends in extremes, more particularly in Antarctica.
reference : Seneviratne, S.I., X. Zhang, M. Adnan, W. Badi, C. Dereczynski, A. Di Luca, S. Ghosh, I. Iskandar, J. Kossin, S. Lewis, F. Otto, I. Pinto, M. Satoh, S.M. Vicente-Serrano, M. Wehner, and B. Zhou, 2021: Weather and Climate Extreme Events in a Changing Climate. In Climate Change 2021: The Physical Science Basis. Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change[Masson-Delmotte, V., P. Zhai, A. Pirani, S.L. Connors, C. Péan, S. Berger, N. Caud, Y. Chen, L. Goldfarb, M.I. Gomis, M. Huang, K. Leitzell, E. Lonnoy, J.B.R. Matthews, T.K. Maycock, T. Waterfield, O. Yelekçi, R. Yu, and B. Zhou (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, pp. 1513–1766, doi: 10.1017/9781009157896.013.
Line 34: In Yu et al. (2018) and Simon et al. (2024), the analysis focuses on the 95th, or even the 90th, percentile. These values should not be considered as “extreme” but rather as “intense”.
Line 35: Antarctic precipitation and EPEs are the major component of surface mass balance (SMB) => This remark depends strongly on the threshold used. The 95th percentile represents a substantial contribution to the total SMB (Turner et al., 2019), but if the 99th percentile is considered, the contribution becomes much smaller. I agree that the March 2022 heat-wave event contributed to “up to 90% of the March 2022 total precipitation in local areas of the East Antarctic Ice Sheet” (Wille et al., 2024b), indicating that such events can have a critical impact. However, in that case, please cite appropriate references (e.g., Wille et al., 2024b; Clem and Raphael, 2023).
references :
Clem, K. R. and M. N. Raphael, Eds., 2023: Antarctica and the Southern Ocean [in “State of the Climate in 2022“]. Bull. Amer. Meteor. Soc., 104 (9), S322–S365, https://doi.org/10.1175/BAMS-D-23-0077.1
Wille, J. D., and Coauthors, 2024: The extraordinary March 2022 East Antarctica “heat” wave. Part II: Impacts on the Antarctic ice sheet. J. Climate, 37, 779–799, https://doi.org/10.1175/JCLI-D-23-0176.1.
Line 47: “Moreover, an enhanced understanding of the drivers of these trends, such as anthropogenically-induced changes (i.e., external climate forcing), is also of critical importance to better highlight how EPEs” => the authors are attempting to address two distinct questions. Each question would warrant a dedicated study. The first part (on drivers) concerns potential changes in atmospheric circulation and their causes, such as links with lower latitudes or major modes of variability. This question is approached through an analysis of atmospheric rivers (ARs), but in the present paper the treatment of ARs is too superficial to adequately address it. I suggest to remove this analysis on ARs. The second part of the sentence pertains to D&A techniques, which is the actual focus of this study.
lines 51-57: The list of potential drivers of change provided here is far from being exhaustive and this reflects my earlier comment on the complexity of addressing the question of the drivers. This point could be introduced, as noted in line 47, but it would also be reasonable to exclude it from the study’s scope, provided this is clearly justified in the text.
Line 57-59: I don’t fully understand this sentence. Could you the directions of trends and the sign of feedback between the various interacting factors?
Line 60: “although some studies have investigated the drivers of trends of extreme temperatures over Antarctica (e.g., Blanchard-Wrigglesworth et al., 2023; Wille et al., 2024b)” => to my knowledge, these papers do not present trends in extreme temperatures.
Line 75: (Min Xu et al., 2024) is not in the reference list.
Line 78: “As ARs are an important contributor to EPEs” => please add a reference
Line 79: If I am not wrong, Wille et al. (2025) do not define a unique ARDT. Which ARDT did you use? How did you attribute precipitation to an AR? i.e., in which contour did you compute the precipitation attributed to an AR?
Line 86: “We use CESM2 ensembles that consider all external forcings (ALL; 50 members) and span the period from 1850 to 2100 under CMIP6 (Coupled Model Intercomparison Project Phase 6)”=> What is the difference between simulations used by Dunmire et al. (2022) and those selected in your study? Indeed, Dunmire et al. (2022) write that “AIS historical precipitation trends in CESM2 appear to be largely driven by the increasing SAM and intensifying Antarctic ozone depletion”. Consequently, could you clarify why they find a significant role for the ozone hole, whereas your study does not?
Line 92: “Here, the GHG and AAER concentrations are evolving in time in their respective simulations, while all other forcings are fixed at 1850 values” => It is written above that you consider prescribed stratospheric ozone depletion, please clarify.
Line 102: the definition of “extreme” precipitation should be presented sooner. Nevertheless, please replace extreme by intense.
Line 111: “can be detected”.
Line 112 : “we summed the 3-hourly precipitation values from the AR detection dataset over 24-hour intervals to produce daily AR-associated precipitation” => Over which area is it calculated? Is it the over the AR contour produced by Wille’s ARDT?
Line 118: “the relative contribution of AR-associated total and extreme precipitation trends to the ERA5-based total and extreme precipitation trends”=> how did you proceed? Did you compute precipitation trends with all the data including ARs, and then excluding ARs?
Line 188 and line 210: I already wrote that, but would this agreement still be real if accumulation were standardized? Since CESM2 accumulation is roughly twice as large as in ERA5, I would expect the relative trends to be about twice as large as well. This would affect also Figures 3 and 4. How do you explain the disagreement in Wilkes land and Marie Byrd Land ? I do not understand why this disagreement is less pronounced in the results of Dalaiden et al. (2022). Could the authors clarify the causes of the differences between CESM1 and CESM2? Indeed, if CESM2 misrepresents the Antarctic climate, the attribution based on it could be unreliable.
Line 220: What would happen if you considered larger areas, as in Dalaiden et al. (2022), rather than restricting the analysis only to basins where the trends are similar?
Figure 4: Caption and elsewhere in the text: the names of catchments are impossible to retrieve without figure 1. It would be helpful to include the name of the region each time you mention a catchment.
Line 118 and section 3.4, Line 293: role of AR in precipitation trends. The current calculation yields relative AR contributions to precipitation trends of up to 100% in some catchments, even where ARs are generally rare. Is this because the AR-related trend is driven by a single event or by very few events? In any case, it seems unexpected that trends in intense precipitation would be 100% attributed to ARs when ARs do not account for all intense precipitation events
Line 318: “this agrees with Casado et al. (2023)”=> Could this result simply be due to coincidence and indicate that the model is incorrectly reproducing, or is masking, the effects of SAM variability?
Line 324: There is no D&A analysis for AR changes. Consequently, the study of ARs appears fully disconnected from the rest of the analysis.
Citation: https://doi.org/10.5194/egusphere-2025-4292-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 793 | 56 | 30 | 879 | 30 | 23 |
- HTML: 793
- PDF: 56
- XML: 30
- Total: 879
- BibTeX: 30
- EndNote: 23
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The study is focused on examining trends in extreme precipitation events (EPE) over Antarctica in ERA5 reanalysis and a global climate model CESM2 and their attribution. The authors identify which Antarctic drainage basins exhibit statistically significant trends in precipitation and EPE. The trends are then related to atmospheric rivers. Further, the trends are identified in three CESM2 runs – including all external forcings (‘ALL’) and then separately for the model runs forced only with greenhouse gases (‘GHG’) and only with anthropogenic aerosols (‘AAER’) – with the latter two considering changes since 1850. Historical runs as well future projections under the SSP370 scenario are considered for the period 1850-2100 with the present-time period 1979-2023 together with ERA5 reanalysis. The results demonstrate that for the period 1979-2023 greenhouse gas forcing is the primary driver of the positive precipitation trends in the ALL ensemble, including EPE trends. The paper is well written and is based on a new methodology using single-forcing applied to earth system models in order to attribute the detected trends in precipitation to a specific forcing beyond natural climate variability. Before the paper can be accepted for publication, several major concerns must be addressed:
1) One of my major concerns is applying ERA5 and a global climate model for estimating trends in precipitation over Antarctica without demonstrating in the paper or citing previous studies how well the reanalysis and the model represent precipitation and particularly extreme precipitation events. I agree that ground-based observations are scarce and difficult but they do exist. There are also satellite observations. And regional climate models (such as MAR and RACMO2), which have been shown to much better represent Antarctic precipitation – not only due to their higher resolution (as noted in the Discussion, L354-360) but also improved physical parameterizations. I invite the authors to provide an evaluation of ERA5 and CESM2 before applying statistical analysis on trends and attribution – this will make the results of the paper more credible.
Here are some suggested references – the authors do refer to some of them already however lacking to present the conclusions about various biases found both in ERA5 and ESMs:
Turner, J., Phillips, T., Thamban, M., Rahaman, W., Marshall, G. J., Wille, J. D., et al. (2019). The dominant role of extreme precipitation events in Antarctic snowfall variability. Geophysical Research Letters, 46, 3502–3511. https://doi.org/10.1029/2018GL081517
Roussel, M.-L., Lemonnier, F., Genthon, C., and Krinner, G.: Brief communication: Evaluating Antarctic precipitation in ERA5 and CMIP6 against CloudSat observations, The Cryosphere, 14, 2715–2727, https://doi.org/10.5194/tc-14-2715-2020, 2020.
Gossart, A., S. Helsen, J. T. M. Lenaerts, S. V. Broucke, N. P. M. van Lipzig, and N. Souverijns, 2019: An Evaluation of Surface Climatology in State-of-the-Art Reanalyses over the Antarctic Ice Sheet. J. Climate, 32, 6899–6915, https://doi.org/10.1175/JCLI-D-19-0030.1.
Gilbert, E., Pishniak, D., Torres, J. A., Orr, A., Maclennan, M., Wever, N., and Verro, K.: Extreme precipitation associated with atmospheric rivers over West Antarctic ice shelves: insights from kilometre-scale regional climate modelling, The Cryosphere, 19, 597–618, https://doi.org/10.5194/tc-19-597-2025, 2025.
2) Methodology clarity: L129-131: as the method of scaling coefficients is central to the results, it has to be explained in more detail (can be also as a supplementary). Now to understand the methodology, the reader has to read Dalaiden et al 2022 paper and its supplement.
L169: “Over the Antarctic Peninsula, the ERA5 trends in total and extreme precipitation differ in direction, with total precipitation increasing and extreme precipitation decreasing, although only the trends over the northern Antarctic Peninsula are significant.” – this can be related to the increased occurrence of rainfall instead of snowfall. The variable “precipitation” used in the study – is this a total precipitation or only snowfall? Please specify this in the Data&Methods section
Also, as Fig 2e shows there seem to be a positive trend in extreme precipitation over the I-Ipp basin however difficult to see because of a very thick line denoting the basin. Maybe making the basin contour line thinner? This positive trend in the I-Ipp region is then mentioned on the next page (L180) so this contradicts the sentence above.
L90: The authors refer to Simpson et al 2023 paper, which describes a single-forcing methodoloy applied to CESM2 model. It will be very useful to provide key details which should help to better understand the results of the present study, especially how GHG forcing is defined in the CESM2 model.
The analysis in the present paper is done for 1979-2023, while the authors mention future simulations under SSP370 scenario (Lines 87-89). According to Simson et al (2023): “The CESM2 large ensemble, referred to hereafter as LENS2, is a 100-member ensemble of simulations run under CMIP6 historical forcings between 1850 and 2014 and forcings of the Shared Socioeconomic Pathway 3–7.0 (SSP3–7.0; Meinshausen et al. 2020) thereafter. “ - Were the same periods used in the present study, meaning 1979-2014 were from historical simulations and 2014-2023 from SSP370 scenario? This and other relevant details have to be clearly described in the Data/Methodology section.
3) L278-286: As the authors suggest, the presented results lack consistency stating “the signals from the GHG and AAER ensembles (and OTHERS) cannot be considered statistically significant for either total or extreme precipitation. “ - does this undermine the entire study conclusion regarding the GHG signal in positive precipitation trends? This differs from the results obtained by Dalaiden et al (2022) who found positive and greater than 1 scaling factors for snow accumulation for GHG forcing. Can the authors put their result in perspective and analyze why there is such a difference?
This is also highlighted in the Discussion section: “However, the three-signal regression-based D&A analysis using the ALL, GHG and AAER ensembles (resulting in scaling factors for GHG, AAER, and OTHERS) was unable to provide robust attribution for total or extreme precipitation to any specific single-forcing, including increased greenhouse gases (Fig. 7). This inconsistency between these results demonstrates that formally detecting the fingerprint of anthropogenic forcing on recent Antarctic precipitation changes remains highly challenging (Previdi and Polvani, 2016; Dalaiden et al., 2022).” – this statement is conflicting with the paper title and its conclusions that intensification of extreme precipitation is due to increases in greenhouse gases. I invite the authors to carefully revise the methodology and/or rethink major conclusions of the study as well as better discussing the results in comparison to previous publications (including Dalaiden et al 2020).
Minor comments:
L39: It is difficult to understand which statement is supported by which reference with 12 references… In this and other instances with more than 5 references listed, I would recommend to break it into separate statements each supported by several key references.
L62 and L187: the section titles 3.1 Identification of precipitation trends and 3.2 Detection of precipitation trends seem to say basically the same thing – identification or detection. The difference is that section 3.1 shows trends detected in ERA5 and section 3.2 then shows trends detected in CESM2 model and continues with a quantitative trend analysis focusing only on selected basins with positive trends according to ERA5. This is confusing. My recommendation will be to separate into “Spatial distribution of precipitation trends” or along the lines discussing Fig 2 for both ERA5 and CESM2, and then the next section “Analysis for basins with positive precipitation trends” or similar.
Figs 5&7: yaxis label: “sacling” – should be “scaling”
L656: Typo: reference Wille et al 2021 listed twice