the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 03 Sep 2026
Shifts in rain-on-snow occurrence in northwestern Eurasia and their connection to Arctic sea ice loss
Abstract. This study explores precipitation, rain-on-snow (ROS) events and subsequent refreezing across northwestern Eurasia, and their relationship with Arctic sea ice. Daily outputs from nine CMIP6 models were analysed for 2015−2100 under three emission scenarios. ROS events were detected using precipitation, temperature, and snow cover after bias correction, and refreezing was identified when sub-freezing temperatures were found within four days after ROS. We find increasing winter precipitation across northwestern Eurasia, dominated by a shift towards more rainfall. The shift from snow to rainfall is more pronounced where temperatures remain near 0 °C, leading to changes in ROS frequency and timing under higher emissions. ROS events followed by refreezing occur less frequently than the total ROS events but indicate conditions favourable for ice-layer formation within the snowpack. The greatest changes occur between 55° N and 70° N, while northern regions remain less affected. ROS timing shifts from spring peaks towards later winter and early spring; at lower latitudes, the spring peak weakens as snow cover becomes less persistent. The Barents-Kara sea ice concentration anomalies are negatively correlated with accumulated precipitation and rainfall. Detrending of the correlations shows that the common long-term trend partly drives this relationship, but negative correlations between sea ice concentration anomalies and rainfall remain in several cases. Future ROS changes are driven mainly by increased rainfall rather than by total precipitation alone, while subsequent refreezing remains dependent on sub-freezing conditions following rainfall. The projected shifts in ROS timing and refreezing may affect snow conditions, socio-ecological systems, and infrastructure across northwestern Eurasia.
- Preprint
(1342 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 08 Nov 2026)
-
RC1: 'Comment on egusphere-2026-5279', Anonymous Referee #1, 09 Oct 2026
reply
The manuscript investigates changes in rain-on-snow events across Eurasia using outputs from nine CMIP6 models under three emission scenarios for 2015–2100. The analysis uses daily mean values and a temperature threshold of 0 °C to distinguish liquid from solid precipitation. Under all three scenarios, winter precipitation increases, with a greater proportion falling as rain as temperatures rise. However, warming also reduces the snow-covered area, limiting the occurrence of rain-on-snow events. Barents–Kara sea ice concentration is negatively correlated with rainfall, although sea ice loss alone does not explain the changes in precipitation.The manuscript provides important insights into future changes in rain-on-snow events across Eurasia. Overall, the analysis and presentation are clear, and I have small suggestions for clarification and improved accessibility.Please provide more information about the models used or include references describing them. Please check that all figures use accessible colors and, where appropriate, additional visual cues. The red dots against the green background in Figure 1 may be difficult to distinguish for readers with color vision deficiencies. Similar concerns apply to the color schemes in Figures 3–7. It would be interesting to evaluate how well the 0 °C temperature threshold identifies liquid precipitation compared with observations. Comparing results based on daily values with those using a finer temporal resolution for a single year would also help assess the sensitivity of rain-on-snow detection to temporal resolution. These analyses could be considered in future work but are not necessary for the present manuscript.ReplyCitation: https://doi.org/
10.5194/egusphere-2026-5279-RC1 -
RC2: 'Comment on egusphere-2026-5279', Anonymous Referee #2, 30 Oct 2026
reply
Summary:
The authors use members from nine CMIP6 models across three emissions pathways to quantify expected changes in rain-on-snow (ROS) frequency, rainfall, total precipitation, and refreezing following rainfall. The authors then use a correlation-based analysis to compare sea ice and precipitation anomalies, using both raw and detrended values. The authors then offer context as to how the expected changes in ROS and refreezing in particular may influence ecology and infrastructure, and how their potential impacts vary across emissions pathways.
General Comments to Author:
- The paper has clear and suitable methods and is concise and well written. The structure of the manuscript, especially the inclusion of a separate ‘Implications’ section, strengthened the paper. I think this work is a valuable contribution to the growing body of literature on the influence of climate change on snowpacks and ROS frequency. I only have minor concerns and suggestions, detailed in the line-by-line comments below.
Line-by-Line Specific Comments to Author:
Figure 1: information on what the basemap colors are showing and the pixelation (is that the simulation grid cell size?) would be helpful
Line 25–26, 178–179: Why did the authors choose to identify refreezing within four days after a ROS event?
Line 88–89: Specify the required variables here – I recommend moving the statement listing variables below up to line 90 following this sentence.
Line 90–91: Please clarify the spatial resolution of the climate model data.
Lines 91–93: The use of one realization from each of the nine CMIP6 models provides a nine-model multi-model ensemble, capturing differences among models but not characterizing internal climate variability. Given the regional nature of this analysis and the sensitivity of ROS occurrence to the timing of meteorological conditions, could the authors clarify why only r1i1p1f1 was selected and whether additional realizations were available? This limitation should also be acknowledged when interpreting the projected changes and associated uncertainty.
Line 148–159: The bias correction approaches seem suitable, but the authors are fairly vague in their description. To support the validity and effectiveness of the bias correction, the authors should consider including supplementary figures showing pre- and post- bias-correction distributions for the variables mentioned, as well as observed and simulated distributions during the reference periods.
Line 176: Based on the pr > 0 threshold, the magnitude of precipitation is not considered. Can the authors comment on how that may be relevant? The hypothesis being that large ROS events are much more impactful on snowpack properties than small events, which likely dominate in frequency. In other words, all ROS events are not created equal, and that should be noted.
Line 220–222: “3 events per winter per decade” wording seems awkward – consider “increasing by up to...” instead of “reaching up to...”
Line 264–267: This is a helpful definition in the Figure 3 caption – I recommend noting this in the text at the beginning of this section.
Figure 3: ROS frequency decreases are noted in several of these panels relative to the prior figures. I suspect this is due to reductions in the number of snow-covered days, meaning there are fewer opportunities for rain to fall on the snow because the snow has already disappeared. Did the authors explore this? I see it was noted in the discussion – consider noting it earlier when presenting these figures.
Line 283–287: Reporting the slopes of each of these lines to indicate the rate of precipitation increases would support these claims. These could be placed on the figure or in a supporting table.
Line 312–315: It is difficult to make sense of these shadings and determine the added value because the various colors overlap. I think it is wise to show the variability range in some form, but consider alternative visualization techniques, such as separating the plots into four panels. Consider alternate visualization for the following figure (Figure 7) as well.
Line 366–373: Did the authors consider exploring spatial correlations between sea ice and precipitation increases? I'd imagine coastal areas may be more highly correlated than inland areas in the region.
General Discussion: The discussion is generally strong and well written, but lacks comparisons on how the magnitude of changes identified by this work (of precipitation extremes/change and ROS + refreeze frequency) compare to previous studies.
Citation: https://doi.org/10.5194/egusphere-2026-5279-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 287 | 154 | 64 | 505 | 66 | 62 |
- HTML: 287
- PDF: 154
- XML: 64
- Total: 505
- BibTeX: 66
- EndNote: 62
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Julienne Stroeve
Ran Tao
Bruce C. Forbes
We examined how winter rain falling on snow may change across northwestern Eurasia as the climate warms. Using projections from nine climate models, we found more winter precipitation and a larger portion falling as rain. Rain-on-snow events become less concentrated in spring and occur over more of the winter, while some southern areas lose spring events as snow cover becomes less persistent. These changes could affect snow conditions, reindeer herding, ecosystems, transport and infrastructure.
We examined how winter rain falling on snow may change across northwestern Eurasia as the...