Future changes in daily snowfall events over China based on CMIP6 models

Huopo Chen ,Jianqi Sun

a Nansen-Zhu International Research Centre, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, China

b National Institute of Natural Hazards, Ministry of Emergency Management of China, Beijing, China

Keywords:Snowfall Anthropogenic influence Projection CMIP6

ABSTRACT China has been frequently affected by severe snowstorms in recent years that have particularly large economic and human impacts.It is thus of great importance to increase our understanding of the underlying mechanisms of and future changes in snowfall occurrences over China.In this study,the effects of anthropogenic influences on snowfall and the associated future changes are explored using new simulations from CMIP6 (phase 6 of the Coupled Model Intercomparison Project) models.Observational evidence reveals a decrease in the annual total snowfall days and an increase in intense snowfall days over the snowfall-dominated regions in China during recent decades.Fingerprints of anthropogenic influences on these changes are detectable,especially the impacts of increased greenhouse gas emissions.During the winter seasons,low temperatures still cover the regions of northern China,and the associated precipitation days show an increase due to anthropogenic warming,which substantially benefits the occurrence of snowfall over these regions,particularly for intense snowfall events.This is also true in the future,despite rapid warming being projected.By the end of this century,approximately 23%of grids centered over northern China are projected to still experience an increase in daily intense snowfall events in winters.Additionally,the length of the snowfall season is projected to narrow by nearly 41 days compared to the current climate.Thus,in the future,regions of China,especially northern China,are likely to experience more intense snowfall days over a more concentrated period of time during the winter seasons.

1.Introduction

Snowfall is an important type of weather event with physical,ecological,and societal impacts that is expected to change in a warming climate (Krasting et al.,2013 ;O’Gorman,2014 ;Chen et al.,2020a).Intense snowfall events have frequently hit the Northern Hemisphere in recent winters and have caused large economic losses and homelessness.There have also been considerable disruptions to the economy,ecosystems,and human health across China associated with severe snowfall events,such as the widespread severe snowfall over eastern China in early 2018 (Sun et al.,2019).Thus,past and future changes in snowfall are of particular concern.

Numerous observational studies have shown an obvious increase in intense snowfall events over some parts of the snowfall-dominated regions in the Northern Hemisphere,such as North America,northern Europe,and northern China (Seager et al.,2010 ;Liu et al.,2012 ;Zhou et al.,2018 ;Cohen et al.,2020),even though the global mean surface air temperature is known to have significantly increased in recent decades.These changes may be not only connected with the increase in atmospheric water content in response to warming but also linked to climate system anomalies,including Arctic sea-ice reduction,Hadley circulation intensification,strengthened stratospheric polar vortices,and North Atlantic Oscillation anomalies (Sun et al.,2009 ;Cohen et al.,2010 ;Seager et al.,2010 ;Liu et al.,2012 ;Wang and He,2013 ;Zhou et al.,2017 ;Bai et al.,2019).

However,until now,studies have mostly sought to explain the physical processes influencing climate system anomalies,with relatively few studies having examined the anthropogenic influence concerning snowfall changes.Anthropogenic fingerprints have been detected in the increasing risks of climate extremes at regional to global scales,including heatwave events,heavy precipitation,floods,and droughts,albeit with low confidence levels (Sun et al.,2014 ;Schaller et al.,2016 ;Chen and Sun,2017,2021 ;Dong et al.,2020 ;Li et al.,2020).The potential roles of anthropogenic influences in the decreasing trend of snowfall events during recent decades have also been detected across land areas of the Northern Hemisphere,and intense snowfall occurrences have been reported to have increased due to the increased anthropogenic activity over some parts of Asia,North America,and Greenland (Chen et al.,2020a).However,the role of anthropogenic influences on changes in snowfall events over China has not yet been explored in these earlier studies and therefore remains an open issue that we seek to address in this study.

In response to further warming in the future,the potential roles of anthropogenic influences on changes in climate extremes are expected to be amplified,leading to more occurrences of heat waves,heavy precipitation,floods,and droughts across the world (Zhou et al.,2014 ;King and Karoly,2017 ;Gao et al.,2018 ;Xu et al.,2018 ;Wang et al.,2019 ;Zhang et al.,2019,2020).In addition,daily snowfall events are projected to decrease across much of the Northern Hemisphere due to increasing anthropogenic warming in the future (Krasting et al.,2013 ;Danco et al.,2016 ;Chen et al.,2020a).Over China,early studies indicated that the days of intense snowfall events are projected to decrease over southern China but initially increase and then decrease over northern China in the future (Sun et al.,2010 ;Zhou et al.,2018).Recently,the outputs of CMIP6 (phase 6 of the Coupled Model Intercomparison Project) models have been progressively released.However,the changes in snowfall events over China according to CMIP6 model projections are still not clear.

Compared with CMIP5,the models in CMIP6 have been substantially improved in many aspects,including their dynamic processes and horizontal and vertical resolutions (Eyring et al.,2016).These new simulations generally show higher performance in simulating climate means and climate extremes (Chen et al.,2020b ;Jiang et al.,2020).Thus,two key issues are discussed in this study using the new outputs of CMIP6 models: (1) the potential role of anthropogenic influences on snowfall changes across China,and (2) their associated future changes.

2.Data and methods

2.1. Data

The observed gridded daily dataset across China developed by Wu and Gao (2013) are employed here,including variables of daily precipitation and daily maximum temperature.This dataset spans from 1961 to 2020 till now and the period of 1961–2014 is the period of interest here.To detect the fingerprints of different forcings on snowfall changes,several experiments conducted in the CMIP6 project are employed here,including historical (ALL),natural (NAT),aerosol (AER),and greenhouse gas (GHG) forcing simulations.To date,there are 10 CMIP6 models (Table S1) in which both daily precipitation and daily maximum temperature are available for all these forcing simulations.For future changes in snowfall,the daily outputs under the 21st century scenario of SSP5-8.5 (SSP: Shared Socioeconomic Pathway) are also used here.Changes for three future periods–the near-term (2021–2040),mid-term (2041–2060),and far-future (2081–2100)–are calculated with respect to the current period taken as 1995–2014.For comparison,both the observational and model datasets are interpolated to a common grid of 1.5° × 1.5°using the first-order conservative remapping procedure via the Climate Data Operator software.

2.2. Methods

Snowfall is difficult to measure and is generally distinguished as precipitation days on which the surface air temperature is below 0°C because snow forms when the surface air temperature is close to freezing point (Sun et al.,2010 ;Krasting et al.,2013 ;Ding et al.,2014 ;O’Gorman,2014 ;Danco et al.,2016 ;Ohba and Sugimoto,2020).However,the rain–snow transition does not occur precisely at the freezing point,and an empirical relationship between them is generally applied for snowfall calculations,which has been widely used (e.g.,Krasting et al.,2013 ;Danco et al.,2016) and is detailed as follows:

whereSis the estimated daily snowfall andPis the precipitation.The fraction of rain–snow transition is determined by the surface air temperature (T) and is replaced by the daily maximum surface air temperature in this study for snowfall statistics,which can ensure that the precipitation falling to the ground is solid snow (Chen et al.,2020a).Subsequently,snowfall days are defined as days with daily snowfall exceeding 0.1 mm,and intense snowfall days as days with daily snowfall greater than 5.0 mm.

Two attribution metrics proposed by Stott et al.(2004)–the probability ratio (PR) and fraction of attributable risk (FAR)–are used here to explore the potential anthropogenic influence on snowfall change over China.PR is calculated asP1/P0,which indicates the change in probability of an event,while FAR is defined as 1–1/ PR,which quantifies the fraction attributable to humans.Here,P1is the event probability for the period of interest,andP0is the corresponding probability for the preindustrial (PI) period.Due to the limited daily outputs of PI simulations,NAT simulations are employed here instead,which can also isolate anthropogenic influences well (Diffenbaugh et al.,2018).

3.Results

Observational evidence suggests that snowfall events mainly center over northeastern China,northern Xinjiang,and the eastern Tibetan Plateau across China (Fig.S1),which is in line with early studies(Sun et al.,2010 ;Zhou et al.,2018).Over these regions,especially over northern Xinjiang and northeastern China,significant increasing trends are apparent for the annual snowfall amount and intensity during the past few decades (Fig.1).However,the number of snowfall days shows an overall decreasing trend across China,particularly over the eastern Tibetan Plateau,while the observed intense snowfall days are clearly increasing across northern China.This implies that the regions of northern China have suffered from increased intense snowfall events in recent decades,especially the regions of northeastern China,northern Xinjiang,and the eastern Tibetan Plateau.Additionally,in response to warming,the timing of the first occurrence of snowfall shows a delayed tendency,while the last occurrences have advanced in recent decades over the snowfall-dominated regions in China (Fig.S2).Thus,the length of the snowfall season over China presents a shortened tendency in response to warming during recent decades.

CMIP6 models can reasonably reproduce many features of mean snowfall and snowfall days over China with respective pattern correlations of 0.37 and 0.77 (Fig.S3).However,the snowfall days are clearly underestimated,particularly in northeastern China and northern Xinjiang.Additionally,the increasing tendency of annual snowfall is also underestimated,but the spatial pattern of the linear trend is reproduced well,albeit with some grids in northern Xinjiang and the eastern Tibetan Plateau presenting a decreasing tendency (Fig.S4).CMIP6 models also present a high degree of skill in simulating the tendencies of snowfall days and intensity,with increasing trends of intensity and decreasing trends of snowfall days across most grids of China,as the observations show in Fig.1.Additionally,CMIP6 models can also reasonably reproduce the trends of intense snowfall and intense snowfall days,but with relatively weak trends compared with observations.Furthermore,the changing characteristics referring to the delayed trends of the first occurrence of snowfall and the advanced trends of the end times in recent decades are both simulated well for regions with the most snowfall in China (Fig.S5).These high degrees of similarity between models and observations thus substantially improve the robustness of our following analyses.

Fig.1.Observed linear trends of (a) annual snowfall,(b) snowfall intensity,(c) snowfall days,and (d) intense snowfall days over China estimated during the period 1961–2014.Crosses denote that the linear trend is significant at the 0.05 level using the Mann–Kendall test.

Fig.2 displays attribution metric plots of daily snowfall events to different forcings across China.The ALL forcing decreases the annual snowfall amount over most grids of China with respect to the “counterfactual ”world without any anthropogenic influence of the NAT forcing simulation.In contrast,27% of grids experience increased snowfall amounts in China,and these grids are mainly located over northwestern and northeastern China,as shown in the observations.In response to GHG emissions,the decrease in the annual snowfall amount is much stronger than that in the ALL forcing simulation,and it prevails over approximately 88% of grids of China.An increasing tendency can be observed in some grids that are also mainly located over northwestern and northeastern China.For the AER forcing simulation,a significant increase in the annual snowfall amount is dominant over China (～76% of grids) compared to the NAT forcing simulation,except for some grids in northeastern China.This increasing tendency of snowfall is highly connected to the cooling effect of aerosols,which generally leads to a decrease in temperature.Further evaluations show that the average annual snowfall amount would increase by nearly 9.7% over China if considering only the effects of AER emissions with respect to the NAT simulation.However,the annual snowfall amount would decrease by approximately 14.9%,averaged over China,in response to GHG emissions.Finally,the snowfall amount is reported to decrease by approximately 8.2% in the ALL forcing simulation compared to the NAT simulation.

Changes in snowfall occurrences feature similar responses to the different forcings when compared with the results of the annual snowfall amount.Evaluations show that more than 92% of grids in China have PRs less than 1.0,implying that the occurrence probability displays an obvious decrease in response to GHG emissions across China.The corresponding metric of FAR also shows a similar pattern,with high values over southern China and low values over northern China,suggesting relatively strong impacts of GHG emissions on changes in snowfall occurrences over southern China.In contrast,the occurrence probability is reported to significantly increase in the AER forcing simulation,and the PR values are greater than 1.0 over most grids of China,especially over southern China.Correspondingly,the average FAR over China is up to 10.2% in the AER forcing simulation,implying that the effect of AER emissions contributes to the increase in snowfall days by approximately 10.2%.This value is much smaller than that in the GHG simulation,in which the average FAR reached -26.8%,suggesting that the increased GHG emissions decreased the snowfall days by approximately 26.8%with respect to the NAT simulation.When considering the effects of both anthropogenic and natural forcing,the results from the ALL forcing simulation show an obvious decrease in snowfall days over most grids of China,especially over southern China.Limited grids (approximately 20%) that are mainly located over northern China experience an increase in snowfall days.The impact of ALL forcing is estimated to contribute approximately 8.1% of the occurrence probability of snowfall averaged over China.

Fig.2.Attribution metrics of daily snowfall events to different forcings over China.The left-hand panels show the changes in annual snowfall in ALL (top)/GHG(middle)/AER (bottom) simulations with respect to that in the NAT simulations.The probability ratio (PR) shown in the middle column is calculated as the rate of occurrence probability in different forcing simulations to that in the NAT simulations,implying snowfall events occur more frequently when this metric is greater than 1.0.The fraction of attribution risk (FAR) shown in the right-hand column is defined as 1 -1/ PR,which indicates the fraction attributable to humans.ALL forcing,historical climate model simulations;GHG forcing,greenhouse gases simulations;AER forcing,aerosol forcing simulations;NAT forcing,the natural simulations.

For the intense snowfall amount and days (Fig.S6),their responses to different forcings are quite similar to those of annual snowfall and snowfall days.However,there are more grids that experience an increase in intense snowfall amount and days,especially for the ALL forcing simulation.The intense snowfall amount shows an increase in response to the ALL forcing in approximately 30% of grids of China.This value is nearly 25% for intense snowfall days.If intense snowfall is defined as daily accumulated snowfall greater than 10 mm,a larger change can be observed for both the intense snowfall amount and intense snowfall days across China (Fig.S7).Furthermore,the grids that experience an increase in intense snowfall would even further increase.For example,an increase in daily intense snowfall days defined as exceeding 10 mm extends to nearly 27% of grids that are mainly located over northern China.Thus,relatively strong responses to anthropogenic influence are generally expected for intense snowfall events across China,especially in terms of the increasing trends of intense snowfall events over northern China.

As expected,the increased GHG emissions exerted a uniform effect on the dates of the first and last occurrences of snowfall events over China,with an increasing tendency of the first occurrence and a decreasing tendency of the last occurrence of snowfall events (Fig.S8).That is,the first dates of snowfall events have been delayed,and the last dates have been advanced in recent decades due to GHG emissions.The AER forcing generally shows an opposite effect on the trends of snowfall occurrence dates but with a relatively weak response when compared to the GHG forcing.Thus,anthropogenic influence has decreased the interval length of the snowfall season over China in recent decades,which is in line with observations.

With further warming in the future,annual snowfall is expected to significantly decrease with a high confidence level across China,especially over southern parts of China,with respect to the current state(Fig.S9).However,some grids over northern China are also projected to experience an increasing tendency of snowfall in the future.In the near-term future of 2021–2040,the snowfall is reported to increase in approximately 13% of grids,mainly located over northern China.In other regions,especially southern China,the annual snowfall is projected to be significantly decreased.Similar changes are featured in the middle of this century (2041–2060),but with a larger decrease in the magnitude of annual snowfall and fewer grids (9%) in which the annual snowfall will increase.By the end of this century (2081–2100),annual snowfall over almost all grids is estimated to significantly decrease.In terms of regional averages,the annual snowfall over China is projected to decrease by nearly 12.7%,23.1%,and 43.6% for the nearterm,middle,and end of this century with respect to the current climate,respectively.

Fig.3.Projected changes in winter snowfall (left-hand column),snowfall days (middle column),and intense snowfall days (right-hand column) for the future periods of 2021–2040 (top),2041–2060 (middle),and 2081–2100 (bottom) under the SSP5-8.5 scenario with respect to the current state (1995–2014).Crosses in the panels denote that the changes are significant at the 0.05 level using the Student’s t -test.

Similar patterns are seen for future changes in snowfall days and intense snowfall days,but intense snowfall days tend to increase in more grids in the future.By the beginning of this century,a larger decrease of at least 25% over most grids of southern China could be observed for both snowfall days and intense snowfall days,while a smaller decrease may prevail over northern China and the Tibetan Plateau.Even some grids over northern China are projected to experience an obvious increase in snowfall occurrences,especially intense snowfall events(13% of grids).This implies an increasing occurrence probability of intense snowfall events over these grids despite rapid warming in the future.However,increases in snowfall days and intense snowfall days over these grids will narrow and weaken with time.By the end of this century,almost all grids over China are covered by a large decreasing tendency for both snowfall days and intense snowfall days.Further estimations show that the average snowfall days over China are expected to decrease by approximately 14.2%,25.9%,and 47.9% for the three future periods (near-term,mid-term,and far-future),respectively.Correspondingly,the average intense snowfall days are estimated to decrease by approximately 13.3%,24.1%,and 45.6% for the near-term,mid-term,and far-future periods,respectively.

Different from annual snowfall,winter (December,January,and February) snowfall shows a stronger increasing response to future warming over most grids of northern China (Fig.3).Moreover,there are also more grids covered by the increasing tendency of winter snowfall.During the period 2021–2040,the estimation shows that winter snowfall is projected to increase in approximately 35% of grids across China with respect to the current climate.This value increases to nearly 39% in the middle of this century.By the end of this century,grids with increasing responses of winter snowfall are reported to decrease with respect to the early periods,especially over northwestern China,covering approximately 28% of grids.Although winter snowfall is projected to obviously increase over most grids of northern China,the average snowfall over China is estimated to decrease in response to future warming,mainly due to the significant decrease over southern China.The results show that winter snowfall will decrease by approximately 7.8%,11.8%,and 25.2% by the early part,middle,and end of this century,respectively,with respect to the current climate.

For the winter snowfall days,the projected increasing tendency also prevails over northern China,especially over northeastern China.This implies an increasing risk of snowfall occurrence in winter in response to future warming over northern China.Increases in intense snowfall days are much higher than those in snowfall days across China,and there are more grids that are expected to experience increasing occurrences of intense snowfall in the future.Our estimations show that there are approximately 31%,36%,and 23% of grids across China in which intense snowfall days are projected to increase for the periods 2021–2040,2041–2060,and 2081–2100,respectively.Furthermore,these grids are mainly located over northern China,and increases in intense snowfall days in winter even exceed 25%.However,the regionally averaged results over China show a decrease of nearly 8.4%,13.4%,and 28.5% of intense snowfall days for the near-term,mid-term,and far-future periods,respectively,which are much smaller than those of snowfall days,with values of 10.0%,17.5%,and 34.4%,respectively.

Warming in the future would further delay the first occurrence of snowfall events across China (Fig.S10).In addition,dates of the last occurrence of snowfall are projected to significantly decrease,leading to a uniform decrease in the length of the snowfall season in response to warming in the future.This impact is expected to be amplified with time.By the end of this century,the average dates of the first occurrence are projected to be delayed by approximately 20.7 days,while the dates of the last occurrence are projected to be advanced by approximately 20.3 days.Correspondingly,the length of the snowfall season would be narrowed by nearly 41 days compared to that in the current climate.Thus,in the future,the regions of China,especially northern China,will experience more daily intense snowfall events.Furthermore,these intense snowfall events are expected to occur over more concentrated times during the winter seasons in the future.

4.Conclusion and discussion

Both observational and modeling evidence have generally revealed a decrease in the total snowfall days and an increase in the mean snowfall intensity during recent decades over the snowfall-dominated regions in China,including northeastern China,northern Xinjiang,and the eastern Tibetan Plateau,which suggests increased intense snowfall days over these regions.Further analyses indicate that the fingerprints of anthropogenic influences have been detected on changes in snowfall and intense snowfall events over China in recent decades,especially the impacts of GHG emissions.Due to the increased GHG emissions,the occurrence probabilities of snowfall and intense snowfall events are observed to decrease across China,but few grids over northern China are observed to experience an increase,especially for intense snowfall events.

With further warming in the future under the 21st century SSP5-8.5 scenario,the annual snowfall amount and days are expected to significantly decrease with a high confidence level across China with respect to the current climate,especially for regions of southern China.Intense snowfall events also feature a similar pattern of change but with more grids over northern China that experience an increasing tendency in response to future warming,particularly for the winter season.By the end of this century,there are approximately 23% of grids across China in which the occurrence probabilities of intense snowfall are projected to increase with respect to the current climate,which are mainly located over northern China.This value is much larger for the winter season at 25%.However,the intense snowfall days in winter are still projected to decrease by approximately 28.5%,averaged over China,by the end of this century.

Our analyses indicate a decrease in snowfall and intense snowfall days across China,but there is also an increase over some grids of northern China under a warming background from both observations and projections.However,one key issue is how climate warming causes such changes in snowfall.On the one hand,the warming tendency is significant across China due to increasing anthropogenic influences,but most parts of northern China are still covered by low temperatures during the winter seasons despite the increasing trend of temperature (Fig.S11 and S12).This is also true for the future change in surface air temperature over China (Fig.S13).That is,the regions of northern China still have a beneficial background of low temperatures to snowfall occurrences,even though the temperature is projected to rapidly increase in the future.On the other hand,according to the Clausius–Clapeyron relationship,climate warming can lead to an increase in atmospheric moisture,which favors an increase in precipitation intensity;that is,a high probability of high snowfall intensity in winter.Our estimations further indicate a significant increase in precipitation days over northern China due to future warming,especially for intense precipitation days during the winter season (Fig.S13).Thus,low temperatures and increased atmospheric moisture can result in an increase in snowfall occurrence,especially intense snowfall occurrence over northern China,even though warming will be significant in the future.Certainly,more discussion about the underlying mechanisms of snowfall changes is still needed in future studies.

Funding

This research was jointly supported by the National Natural Science Foundation of China [grant numbers 41991284,41922034,and 42088101 ] and the Strategic Priority Research Program of the Chinese Academy of Sciences [grant numbers XDA23090102 and XDA19070201 ].

Supplementary materials

Supplementary material associated with this article can be found,in the online version,at doi: 10.1016/j.aosl.2021.100137.