Search

This study evaluated the performance of 23 models participating in Phase 6 of the Coupled Model Intercomparison Project (CMIP6) in simulating snowfall variation across the Eurasian continent. The results show that the CMIP6 multi‐model ensemble (MME) can reasonably capture the climate means of snowfall indices for 1995–2014 period and has high pattern correlations with observations. However, most of CMIP6 models generally overestimate snowfall amounts, snowfall days and snowfall duration, and the overestimation is much clearer in spring. The performance of these models in simulating snowfall events with different grades (including light snowfall, moderate snowfall, heavy snowfall and snowstorm) becomes gradually poorer with higher snowfall levels. On the whole, the comprehensive model ranks implied that NorESM2‐MM, NorESM2‐LM and CESM2‐WACCM models showed the best performances. In addition, the CMIP6 MME outperforms individual models in terms of both spatial–temporal features and interannual variability; thus, the MME is used to project the snowfall change in the 21st century. Further evaluations show that by the end of this century, snowfall will decrease in most of Eurasia with an average value of 16.7% (31.3%) under the SSP2‐4.5 (SSP5‐8.5) scenarios, and snowfall days will decrease by 12.6 days (25.0 days); the snowfall duration will shorten further due to the first date of snowfall occurrence will postpone by 12.7 days (23.0 days) and the last date further advance by 15.0 days (26.7 days). Additionally, heavy snowfall, snowstorms and their number of days will increase over part of Northern Asia, the Tibetan Plateau and Central Asia, where the risk of snowfall extremes will increase. However, large inter‐model and scenario uncertainties also exist in the CMIP6 model projection of snowfall events, which requires further improvement. [ABSTRACT FROM AUTHOR]

It has been widely reported that anthropogenic influences are detectable with high confidence in global warming. However, whether human activities have an impact on snowfall changes is still unclear. Here we based on phase 6 of the Coupled Model Intercomparison Project (CMIP6) multi-forcing dataset and the regularized optimal fingerprint method, the detection and attribution of various grades of snowfall (including annual snowfall (>0.1 mm/day), light snowfall (<2.5 mm/day), intense snowfall (>5 mm/day), and their corresponding days) changes in Eurasia (20 ° -90 °N , 10 °W -180 °E) were carried out. Results show that anthropogenic activity forcing (ANT) and greenhouse gas forcing (GHG) well reproduced the spatial-temporal characteristics of snowfall indices. The ANT influence is robustly detected in the decreased trend of snowfall days (snowday), light snowfall, and light snowfall days (light_day) at the 90% confidence interval, clearly separated from the natural forcing. Moreover, the GHG signals are detectable for decreases in the three snowfall indices, which only could be distinguished from the natural and aerosol forcings for snowday and light snowfall. Thus, anthropogenic activities may considerably account for the decreases in snowday, light snowfall, and light_day across Eurasia, wherein the changes in the first two indices dominated by GHG emissions. However, human influence detections fail for intense snowfall, and it is hard to detect on regional scales, except for North Asia. Finally, by the end of this century (2081–2100), the observation-constrained projections based on the detection and attribution analysis under two SSPs (new Shared Socioeconomic Pathway) scenarios exhibit that the scaled snowday, light snowfall, and light_day are expected to decrease by approximately 13.9 days (28.3 days), 24.8% (48.7%), and 4.3 days (8.8 days) under the SSP2–4.5 (SSP5–8.5) scenario with reference to the current climate (1995–2014). Our study highlights the need to improve climate model performance in simulating extreme snowfall to clear whether and to what extent human influence impacts it. In the past 35 years, the decrease in snowfall days (snowday), light snowfall, and light snowfall days (light_day) can be attributed to human influence. However, CMIP6 simulations may underestimate these observed changes. We utilize the observationally constrained techniques based on detection results to constrain the projections for snowfall changes. Our results show that by the end of the 21st century, the scaled snowday, light snowfall, and light_day are expected to decrease by approximately 13.9 days (28.3 days), 24.8% (48.7%), and 4.3 days (8.8 days) under the SSP2–4.5 (SSP5–8.5) scenario with reference to the current climate. [Display omitted] • In the past decades, the observed snowfall decreased while extreme snowfall increased across most of Eurasia. • The decrease in snowfall days and light snowfall could be attributed to anthropogenic influence dominated by GHG emissions. • CMIP6 models may underestimate the observed changes in snowfall. • The observation-constrained projections show a larger decrease for snowday and light snowfall than the raw CMIP6 outputs. [ABSTRACT FROM AUTHOR]