Research progress on inter-monthly winter temperature variation in East Asia and climate prediction

Ke Fn ,Hongqing Yng ,Hixi Di

a School of Atmospheric Sciences,Sun Yat-sen University and Southern Marine Science and Engineering Guangdong Laboratory(Zhuhai),Zhuhai,China

b Key Laboratory of Polar Science,MNR,Polar Research Institute of China,Shanghai,China

Keywords: Inter-monthly variation Winter temperature East Asia Winter monsoon Predictability

ABSTRACT Inter-monthly winter temperature variation in East Asia has been remarkable in recent years,showing reversed or alternating extreme cold and extreme warm events in different months or in different stages of the winter.There are many challenges in climate prediction in the winter months because the inter-monthly climate variation is often within the seasonal mean variation.It is therefore urgent to understand the variation of inter-monthly winter temperatures in East Asia,to identify their predictability and predictive sources,and to propose effective prediction methods and prediction models for the inter-monthly winter climate.This paper reviews progress in research during the last five years on the main characteristics,physical processes,mechanisms,predictability,and prediction of inter-monthly winter temperatures in East Asia,considering several related systems including the winter monsoon,Siberian high,and stratospheric polar vortex.The authors also discuss future research prospects.

1.Introduction

The inter-monthly winter temperature anomaly in East Asia has become more significant in recent years,showing a reversal or alternation of extreme cold to extreme warm events in different months or at different stages of the winter.For instance,it was extremely warm in China in December 2013—January 2014,but extremely cold in February 2014 (Si et al.,2014).While a super El Ni?o event occurred in 2015—2016,the seasonal mean air temperature in winter over China was much higher than normal.However,there was an extremely warm event in December 2015 and an extremely cold event in January 2016,followed by a warm February 2016.In 2020—2021,it was extremely cold from December to mid-January and then became extremely warm in mid-January to February 2021 over East Asia.The anomalous lower temperature in early winter 2021 was linked to the reduction of Arctic sea ice in Autumn,while the sudden stratospheric warming in early January caused a weak stratospheric polar vortex (SPV) and Siberian high(SH),resulting in the extreme warm event in the later winter(Yang and Fan,2022).The inter-monthly climate variation is often masked by the seasonal mean variation,which reduces the accuracy of predictions of the seasonal mean climate.Intraseasonal climate prediction,including monthly climate prediction,is becoming much more challenging.The object of the international Subseasonal-to-Seasonal (S2S) Climate Prediction Research Project is to understand climate processes with the aim of enhancing climate prediction(Vitart et al.,2017).

There have been many achievements in understanding the interannual and interdecadal variability of the winter seasonal mean climate in East Asia.Wang and Fan (2013) indicated that,beginning from the mid-1980s,the decline of the East Asian winter monsoon (EAWM) led to a rise in winter temperatures throughout China,along with an increase in the sea surface temperature (SST) along the East Asian coast and the volume of water vapor transported to Northeast China.As a result,winter snowfall in Northeast China increased after the mid-1980s.Ding et al.(2014) concluded that the interdecadal variation of the EAWM is connected to global climate change,atmospheric circulation,and the Pacific SST.The positive phase of the Pacific Decadal Oscillation and North Atlantic Oscillation(NAO)corresponds to a weak EAWM and high winter temperatures in China,and vice versa.Furthermore,the negative phase of the Atlantic Multi-decadal Oscillation is associated with cold events over East Asia,and vice versa.The significant warming of winter temperatures over East Asia in the 1980s was partly influenced by both the Eurasian teleconnection at 500 hPa and the eastward expansion of the SH connected to the polar circulation (Fan and Liu,2013).The SH has intensified in the last 20 years (Jeong et al.,2011).The Arctic Oscillation (AO) and the Antarctic Oscillation tended to be in a positive phase after the 1980s(Fan and Wang,2004;Gong et al.,2004;Wu and Wang,2002),which favored a weakening of the EAWM,a rise in temperature and a decline in the frequency of dust weather in East Asia during winter—spring.

El Ni?o—Southern Oscillation (ENSO) is an important factor in the East Asian climate and climate prediction(Wang et al.,2000).However,Wang and He(2012)revealed that the relationship between the EAWM and the ENSO has diminished in the years following the mid-1970s.Chen et al.(2019) indicated that different combinations of ENSO and the AO have different effects on the EAWM.Since 2007,with the reduction in Arctic sea ice,the effect of sea ice over the Arctic on the Eurasian winter climate has noticeably augmented.The decrease in Arctic sea ice may make more cold airmasses move southward and much more water vapor be transported from the open ocean via a wave-like circulation from middle to high latitudes.This causes an increase in snow and extreme cold events (Comiso et al.,2008;Liu et al.,2012;Wu,2018;Xu and Fan,2020,2022).Xu and Fan (2020) found that the interannual variability of sea ice over the Barents Sea intensified significantly after the mid-1990s,showing a five-to eight-year periodic oscillation.This changed the periodicity of the North Pacific Oscillation in January from two to three years to five to eight years with an eastward shift.The effect of the North Pacific Oscillation on the North American air temperature in has therefore increased since the mid-1990s.

The connection between the interannual variation of winter sea ice over the Barents Sea and spring dust weather frequency in North China has also significantly enhanced since the mid-1990s.A decrease in winter sea ice over the Barents Sea favors the occurrence of spring dust weather over North China because the stronger intensity of the interannual variation of Arctic sea ice induces much larger climate effects(Fan et al.,2018).Therefore,skillful statistical and hybrid statistical—dynamic climate models to predict spring dust weather in northern China have been constructed,taking into consideration the Arctic sea ice in winter (Ji and Fan,2019).Particularly,the hybrid prediction model has the high accuracy for the prediction of the frequency of abnormal dust weather.Therefore,we need to understand how these changes affect the subseasonal variabilities of winter temperatures in East Asia as the sea—land—atmosphere system undergoes further changes.

The atmospheric intraseasonal oscillation (ISO,a 10—90-day periodic oscillation) partly accounts for the intraseasonal variations in the Earth’s climate.The interannual variabilities of the ISO intensity are associated with the monsoon and ENSO(Guo et al.,2021;Li et al.,2020).The Madden—Julian Oscillation (MJO),an important atmospheric 30—60-day oscillation in the tropics,is a crucial predictor of the intraseasonal climate over East Asia (He et al.,2011;Jia et al.,2011;Li et al.,2020;Song and Wu,2019;Stan et al.,2017).Statistical prediction models of the subseasonal climate in China are effective and include spatiotemporal projection predictions(Hsu et al.,2015)and low-frequency oscillation predictions(Liang and Ding,2013).

Operational seasonal climate prediction is still complex.For instance,low-frequency oscillations cannot totally capture changes in the actual variation of the climate anomaly,which are sometimes not significant.Chen and Wei (2012) indicated that if 20—50 days of a lowfrequency oscillation are used to predict persistent summer precipitation over eastern China,this can only explain one-fifth of the actual variation in summer rainfall.The positive prediction skill of coupled general circulation models/oceanic general circulation models(CGCMs/OGCMs)is mostly distributed in the tropics,where the intensity of the MJO is generally underestimated(Kumar et al.,2011;Slingo et al.,1996).We therefore urgently need to have a thorough understanding of the mechanism of the ISO’s influence on winter climate in East Asia and its predictability to identify forecasting sources from air—land—ocean processes,as well as tropical—extratropical factors.

The main idea of the year-to-year increment prediction method is to take the year-to-year increment of a variable (the climate variable of the current year minus that of the previous year) as the predictand instead of the traditional anomaly relative to the climatological mean value(Fan et al.,2008).The year-to-year increment of the climate variable is predicted and then added to the observed value of the preceding year to acquire the climate or its anomaly.The year-to-year increment prediction method can amplify the signal of climate prediction,including high-latitude climate systems and tropospheric quasi-biennial oscillation (QBO).In addition,the tropical analog prediction method has therefore been proposed based on historical analogs of the tropical climate and a tropical analog prediction scheme for East Asian—western Pacific precipitation in summer has been developed(Wang and Fan,2009).According to the year-to-year increment method and tropical analog prediction,hybrid prediction models have been constructed,based on previous observations and synchronous factors from skillful outputs from CGCMs/OGCMs or multi-model ensemble (MME) results(Dai and Fan,2020;Dai et al.,2018;Fan et al.,2008,2012;Liu and Fan,2012,2013,2014).The prediction skill of these hybrid models for the East Asian climate,such as the East Asian monsoon,precipitation,temperature,typhoons,and dust weather,has higher prediction skills than the raw prediction of CGCMs/OGCMs and MME results.

This paper reviews the mechanisms,predictability and prediction of inter-monthly winter temperatures in East Asia over the last five years.We review the primary characteristics of inter-monthly East Asian temperatures in winter and related climate systems including the EAWM,SH and SPV in Section 2.In Section 3,we review the physical processes of the variation of the inter-monthly winter temperature,and the roles of ENSO and Arctic sea ice,respectively.In Section 4,we review the predictability and prediction of monthly winter temperatures and related climate systems.Finally,we discuss future research prospects in Section 5.

2.Inter-monthly winter temperatures in East Asia and related climate systems

2.1.EAWM and SH

The EAWM is modulated by multiple processes and factors.For instance,the EAWM and the SST anomaly in the Northwest Pacific exhibit an intraseasonal covariance (Wu,2016),which is connected to the intraseasonal NAO modulated by ENSO (Geng et al.,2017).Jiao et al.(2019) showed that the intraseasonal changes of the EAWM are closely related to two wave trains from the North Atlantic.The variation of the SPV is also an important factor affecting the intraseasonal variation of the EAWM(Chen et al.,2013).The EAWM and ENSO have significant connections in November—December,but this relationship is weaker in January—February.ENSO can affect the anomalous anticyclone in the Northwest Pacific by influencing the Walker circulation and can therefore influence the EAWM.This is because that with weakening ENSO variability in late winter,the corresponding Northwest Pacific anticyclone is in the south of 30°N and its influence on the EAWM is weakened(Tian and Fan,2020).

As one of the critical systems of the EAWM,the inter-monthly reversal of the winter SH affects the intraseasonal variabilities of winter temperature over East Asia (Hori et al.,2011).It is known that the winter SH is directly influenced not only by local thermodynamic processes and large-scale atmospheric circulation anomalies,but also by Arctic sea ice,Eurasian snow cover,and the SST over the North Atlantic(Cohen et al.,2001;Wu et al.,2011;Zeng et al.,2015).The stratospheric circulation has an increasing role in the winter SH and winter temperatures in East Asia(Cohen et al.,2001).

Furthermore,how does the inter-monthly reversal of the SH affect intraseasonal winter temperatures in East Asia? Chang and Lu(2012)showed that,between November and December—January,the SH exhibited an in-phase variation during 1958—1978 and an out-ofphase variation during 1979—2008,which may have been respectively responsible for the intraseasonal variation of the Pacific and Ural blocking highs.The extent of September sea ice in the Arctic can change both the storm track intensity in northeastern Europe and the frequency of the Ural blocking high via eddy—mean flow interactions,resulting in the reversal of the November SH and December—January SH during 1979—2015 (Lü et al.,2019).Furthermore,out-of-phase variations of the SH in December and January also occurred.The reversal frequency of the monthly SH in December and January increased after 2000,from 30%in 1981—2000 to 63%in 2001—2019(Yang and Fan,2021b).This might be because the effect of snow cover in November over Siberia also increased after 2000.Increased Siberian snow cover in November makes a strong SH in December via the effect of radiative cooling and induces troposphere—stratosphere propagation of planetary waves in November—December.This weakens the SPV and SH in January.Fig.1 shows the processes of out-of-phase variations of the EAWM and SH on the monthly scale.

Fig.1.Illustration of the out-of-phase variations of the EAWM and SH on a monthly scale.

2.2.SPV

The SPV provides predictability to the intraseasonal tropospheric climate via stratosphere—troposphere interactions (Baldwin et al.,2003,Butler et al.,2019).As planetary waves propagate vertically from the troposphere to the stratosphere and then break up,it leads to convergence of the wave activity flux and eventually weakens the SPV (Andrews et al.,2019;Matsuno,1970).Subsequently,weak SPV signals can propagate downwards into the troposphere and even reach the surface (Baldwin and Dunkerton,2001;Kidston et al.,2015),resulting in a strong EAWM (Chen and Wei,2009;Chen et al.,2013).This process is particularly evident after sudden stratospheric warming events,which exhibit the strongest stratosphere—troposphere coupling(Charlton and Polvani,2007;Chen et al.,2015;Deng et al.,2008).In contrast,strong SPV events lead to the opposite characteristics in stratosphere—troposphere interaction processes (Chen and Wei,2009;Limpasuvan et al.,2005).Previous studies have shown that the SPV intensity shows ISO characteristics in winter with a period of about 120 days (Hardiman et al.,2020;Kuroda and Kodera,2004).The SPV might therefore undergo transitions from weak to strong or vice versa in winter.

Shan and Fan (2022b) described the 1987/88 winter case in which the transition of the SPV intensity between early and late winter was the most significant during 1979—2019.It was indicated that the weak (strong) SPV anomalies in early (late) winter 1987/88 might have resulted from strong (weak) upward planetary waves,especially wavenumber-1,propagating into the stratosphere.These anomalous signals of the upward stratospheric waves can be traced back to the lower troposphere.Furthermore,the reverse SPV anomalies between early and late winter can both propagate downward into the troposphere.The transition of the SPV might therefore contribute to the reversal of tropospheric zonal wind anomalies,presenting dipole-like anomaly modes between the North Atlantic and Asia—Arctic regions.Shan and Fan (2022a) further analyzed the predictability of the above SPV transition case and indicated that the ability of the model in capturing the intensity of upward planetary waves entering the stratosphere might be key to predicting the SPV.Peng et al.(2019) revealed that the impact of weak SPV events in December on Eurasian temperature can be divided into a stratosphere—troposphere interaction stage and a troposphere fluctuation stage.The first stage persists for about 25 days and is characterized by downward propagation of the SPV anomaly,which can trigger a negative phase of the AO.After 25 days,as the downward SPV signal gradually disappears,a tropospheric wave train can propagate eastward from the North Atlantic and continue to influence Eurasian temperature.

In addition,the SPV modulates the linkage between the AO and precipitation over Asia (40—56°N,75—120°E) in winter (November—February)on the interannual scale(Zhou and Fan,2021).During 2002—2017 (1979—1999),winter stratospheric positive (negative) geopotential height anomalies over high latitudes of the North Atlantic strongly(weakly) propagated downward,causing a negative (positive) AO with different spatial patterns,which may have further caused the significant(non-significant) negative relationship between the AO and precipitation over Asia.Moreover,Zhou and Fan(2022a)showed that the sea-ice concentration(SIC)over the Barents—Kara sea(SIC_BKS)in November—December was positively (negatively) correlated with the SPV in subsequent January during 1979—1995 (1996—2009) on the interannual scale.The variations are directly related to the interannual variabilities of upward-propagating planetary waves associated with low SIC_BKS years modulated by the decadal variations of the shift in the interannual variability’s center locations of the SIC_BKS and the intensities of polar front jet waveguides.Zhou and Fan (2022b) found that the correlation between the tropical stratospheric QBO at 30 hPa in August—September and the SPV in December—January was significantly positive during 1998—2017 on the interannual time scale,whereas the correlation was insignificant before 1998.Compared with 1979—1997,the anomalous deep convection over the tropical western Pacific and Indian oceans in October—November were stronger and shifted westward under the easterly phases of the stratospheric QBO in August—September during 1998—2017.This anomalous convection can generate and propagate Rossby wave train anomalies to mid-to-high latitudes of the Northern Hemisphere and further enhance the intensity of upward-propagating tropospheric planetary waves entering the stratosphere,leading to a weak SPV in December—January.

3.Mechanisms of Inter-monthly winter temperatures over East Asia

3.1.Role of ENSO

The variations of the inter-monthly East Asian winter temperatures show a reversal or alternation in different months or in different stages of the winter.There are both in-phase and out-of-phase variations in the temperature anomalies over China between 16 November—15 January and 16 January—15 March (Wei et al.,2020).ENSO has a crucial role in the inter-monthly anomalies of winter temperatures over East Asia.Geng et al.(2017)indicated that super El Ni?o events are accompanied by out-of-phase inter-monthly temperatures in winter over East Asia,which interact with the subtropical jet to induce an out-of-phase NAO.Central Pacific—type ENSO events(CP-ENSO)have had a significant impact on the out-of-phase mode of December and January temperature over China since the late 1990s (Li et al.,2021).With warm CP-ENSO events occurring since 1997,the South China Sea and the Kuroshio extension have presented negative SST anomalies.The westward extension of the descending branch of the Walker circulation weakens the Hadley and Ferrel circulations,leading to a stronger SH and colder conditions in China in December.In January,as the descending branch of the Walker circulation moves to the north at about 30°N,with a stronger Ferrel circulation and weaker SH,China is warmer than the climatology.

CP-ENSO can also affect the out-of-phase variations of Central Asian temperature in December and January(Li et al.,2022).In December,the cold SST anomalies over the central tropical Pacific generates eastwardpropagating Rossby waves from the North Pacific,leading to positive geopotential and high-temperature anomalies over Central Asia.As a result of the stronger and more southward climatological subtropical jet stream over the North Pacific in January,the anomalous synoptic transient eddy activity induced by CP-ENSO can propagate eastward and lead to local positive geopotential anomalies high over North Atlantic.This results in negative geopotential anomalies over Central Asia with low-temperature anomalies.

3.2.Role of Arctic sea ice

With the intensified interannual variability of Arctic sea ice in recent years (Fan et al.,2018),Arctic sea ice significantly more contributes to the inter-monthly variabilities of winter temperatures in East Asia than before.Dai et al.(2019) indicated that the reversal of temperatures in December and January—February over Northeast China is a combined result of anomalies in the SIC_BKS and the SIC over the Davis Strait—Baffin Bay(SIC_DSBB)in November.The SIC_DSBB anomalies in November induce SST anomalies over the North Atlantic through eddy feedback mechanisms and further propagate eastward Rossby waves to East Asia,affecting December temperatures over Northeast China.By contrast,the SIC_BKS anomalies in November affect temperatures over Northeast China in January—February via stratospheric processes.This may be attributable to the differences in the month-to-month increases in the SIC_DSBB and SIC_BKS in November.The increase in the SIC_DSBB in November is relatively small,corresponding to a weak Rossby wave source.This means that the eastward Rossby waves generated by the SIC_DSBB are only sustained until December.By contrast,the large increase in the SIC_BKS in November could affect the upward-propagating stationary Rossby waves,inducing an AO-like pattern in the stratosphere.This AO-like pattern propagates downward to the troposphere and affects temperatures in Northeast China later in January—February.

These crucial mechanisms have also been confirmed on the synoptic scale,as illustrated by the example of 2014—2015.The temperatures in Northeast China changed from cold conditions in December 2014 to warm conditions in January—February 2015,with a temperature difference of>3°C.The SIC_DSBB and SIC_BKS in November 2014 were anomalously low and high,respectively.The evolution of temperatures in winter 2015 was therefore related to the synergistic impact of eastward-propagating Rossby waves in the troposphere during the first 10 days of December 2014 and downward-propagating planetary wavenumber-1 from the stratosphere from late December 2014(Dai and Fan,2022).Warm SST anomalies in the equatorial central Pacific modulated by the positive Pacific Decadal Oscillation phase,together with the Arctic SIC in autumn,also contributed to the temperature reversal in winter 2015(Xu et al.,2018).

3.3.Case study:reversal of East Asian temperatures in winter 2021 connected to Arctic sea ice and the SPV

Several cold surges occurred over East Asia before mid-January 2021 in early winter,followed by extreme warm surges after mid-January in late winter.Yang and Fan (2022) showed that the monthly alternations of winter temperatures in East Asia in 2021 corresponded to a reversal of the intensity of the SH.They also showed the role of the Barents—Laptev sea ice in the previous September and sudden stratospheric warming in January in the out-of-phase variations of the SH.The reduction of Barents—Laptev sea ice in September decreased the Arctic—East Asia temperature gradient.This strengthened the SH,resulting in colder conditions in early winter 2021.However,sudden stratospheric warming happened on 5 January 2021,weakening the SPV.The SPV shifted towards North America and propagated planetary-scale fluctuations downward from the stratosphere.Consequently,the negative AOlike pattern shifted towards North America and the SH was weakened,causing extremely warm conditions in late winter 2021.Fig.2 illustrates the out-of-phase variations in the inter-monthly temperature over China in winter.

Fig.2.Illustration of the out-of-phase variations of the inter-monthly temperature over the whole of China(left)and Northeast China(right).

4.Monthly predictability and prediction of winter temperatures in East Asia and related climate systems

4.1.EAWM and SH

The prediction skill of the EAWM and winter SH shows inter-monthly differences.For example,the National Center for Environmental Prediction Coupled Forecast System version 2 (CFSv2) model predicts the EAWM more accurately in early winter (November—December) than in late winter (January—February) (Tian and Fan,2019).This is partly attributable to the relation between ENSO and the EAWM being much closer in early winter.Tian et al.(2017)applied the year-to-year increment prediction method to improve the prediction skill for the EAWM,in which the Arctic sea ice and the SST over the North Pacific in the previous autumn were used as predictors.Yang et al.(2021)found that the intensity of the November SH can be predicted best among the individual months of the winter season (November—February) by CFSv2.Their results showed that the thermal factors related to the SH (e.g.,surface upward longwave radiation),dynamical factors (e.g.,convergent downward motion in the troposphere),and snow cover in Siberia partly determined the inter-monthly difference of prediction skill for the SH.The consistent variations of the December—January SH have been shown to be connected to the Arctic sea ice in the previous autumn for the period 1981—2000.By contrast,the reversal of the SH between December and January has been shown to be related to the Siberian snow cover in November 2001—2019(Yang and Fan,2021b).Yang and Fan (2021a) therefore selected the predictors of Siberian snow cover and Arctic SIC in developing their prediction model of the winter SH with the year-to-year increment approach to improve the accuracy in predicting the SH.

4.2.NAO

The NAO,a dominant atmospheric oscillation in the Northern Hemisphere,plays a crucial role in the Eurasian climate.However,it is challenging to predict the winter and monthly NAO using most CGCMs/OGCMs (Fan et al.,2016).Tian and Fan (2019) showed that CFSv2 skillfully predicts the monthly NAO better in December than in other months of winter.The SST anomalies over the North Atlantic and stratospheric variations are therefore both key predictability sources of the NAO.Using the year-to-year increment prediction method,the two preceding predictors (SST over the Atlantic and snow cover over Eurasia)were selected and a skillful statistical prediction model for the winter NAO was constructed(Tian and Fan,2015).According to the skillful prediction of the year-to-year increment of the NAO from state-of-the-art CGCMs/OGCMs and the MME,a hybrid statistical—dynamical prediction model for the winter NAO was established that showed a higher prediction skill for an extreme winter NAO than the output of the raw coupled model(Fan et al.,2016).

4.3.Winter temperatures in East Asia

There are multiple factors and processes influencing the intraseasonal variabilities of East Asian temperatures in winter,including the Arctic SIC,Eurasian snow cover,ENSO,the NAO and the EAWM.Considering the physical mechanism of temperatures over China in winter and the year-to-year increment prediction method,the winter sea level pressure (SLP) over pan-Eurasia from the CFSv2 model,the observed North Pacific SST from the preceding July—September,and Arctic SIC in the preceding August were chosen to construct an operational hybrid prediction model for winter temperatures over China(Dai and Fan,2020).The domain of the simultaneous SLP predictor covers the EAWM,such as the SH,the East Asian trough,and the Indian—Burma trough,which can have an impact on winter temperatures over China.The SST predictor over the North Pacific contributes to ENSO events via wind speed—evaporation—SST feedback mechanisms and further influences the EAWM by modifying anticyclones over the Philippine Sea.The decreased Arctic sea ice strengthens the SH and the Aleutian low in winter,favoring cold conditions over China (Dai and Fan,2020).

Compared with the prediction results from raw CFSv2 outputs,the prediction models based on individual predictors all improved the predictability of winter temperatures over China,and different predictors contributed to the forecasting of winter temperatures over different regions.The multi-predictor model using all three predictors(MP scheme I)outperformed all the other models,including the CFSv2 outputs.The station-average root-mean-square error (RMSE) of MP scheme I was only 0.25,a decrease of about 48.0%compared with the CFSv2 model.However,MP scheme I still showed relatively poor prediction skill over Northeast China as a result of the difficulties in predicting winter temperatures in this region.Dai et al.(2018) therefore developed another hybrid downscaling scheme.This scheme focused on forecasting winter temperatures over Northeast China by selecting the winter SLP over the subtropical South Indian Ocean from the CFSv2 outputs released in November,the August SST over the North Pacific,and the November SIC_BKS to build a hybrid prediction scheme (HD-scheme).For the downscaling results during the period 1984—2016,the area-averaged anomaly correlation coefficient of the HD-scheme improved to 0.55(at the 99% confidence level) and the area-averaged RMSE reduced by>60% compared with the output of the CFSv2 model at most stations(Dai et al.,2018).

MP scheme II was proposed by combining the prediction results at 225 stations over Northeast China from the outputs of HD-scheme and the results at other stations from MP scheme I.In general,MP scheme I and MP scheme II showed comparable skill over China,but MP scheme II outperformed MP scheme I in predicting winter temperatures over Northeast China(Fig.3).The ratio of the same sign of anomalous years(the anomalous RSS)increased from 39%in MP scheme I to 57%in MP scheme II,and the RMSEP increased from 49% to 54% over Northeast China.Both MP schemes were shown to be expert in predicting anomalous winters in China.Liu et al.(2021)developed statistical—dynamical prediction models of monthly winter air temperatures over China based on the October SIC over the Arctic and the key regional simultaneous SST derived from the MME of OGCMs.In new research,prediction models of the inter-monthly temperature for a consistent mode and two inconsistent modes have been developed,which effectively improve the inter-monthly temperature prediction over China (Yang et al.,2023).Fig.4 shows the processes of statistical—dynamical prediction models of the winter temperature over China on the seasonal and monthly scales.

Fig.3 .Comparison of cross-validation downscaling results over Northeast China from CFSv2 outputs,the MP-I scheme and the MP-II scheme during 1984—2017.

Fig.4.Illustration of the statistical—dynamical prediction models for winter temperature over China on the seasonal and monthly scale.

5.Discussion and conclusions

The variation of inter-monthly winter temperatures over East Asia has been remarkable in recent years,showing reversals or alternations in different months or different stages of winter,leading to more alternating extreme cold and extreme warm events.It is therefore important to enhance the skill of inter-monthly to seasonal predictions of winter temperatures.We reviewed recent research progress in the variation of predictability and prediction of the inter-monthly temperatures over East Asia in winter and the related climate systems,such as the EAWM,the SH,and the SPV.

The processes of the inter-monthly variation in winter temperatures over East Asia are highly complex,and sea—land—atmosphere processes and extratropical climate systems are important.However,these extratropical processes or climate systems cannot be reproduced well by most state-of-the-art CGCMs/OGCMs,and synchronized factors cannot be used in advance in the statistical prediction.Fan et al.(2008) proposed the year-to-year increment prediction approach,which can amplify the prediction signal,especially for the Earth’s climate at high latitudes.An effective hybrid prediction method has been proposed based on the year-to-year increment prediction model by combining synchronized predictors from skillful CGCMs/OGCMs with previous predictors from accurate observations.This technique involves some of the key extratropical processes or climate systems.The SPV and QBO are key sources of predictability for inter-monthly winter temperatures over East Asia,and therefore the processes,simulation,predictability and prediction of the stratospheric atmospheric circulation need to be enhanced.

Funding

This study was supported by the National Key Research and Development Program of China[grant number 2022YFE0106800],the Natural Science Foundation of China(NSFC)[grant numbers 42230603 and 41730964],and the Innovation Group Project of the Southern Marine Science and Engineering Guangdong Laboratory (Zhuhai) [grant number 311021001].