An operational statistical downscaling prediction model of the winter monthly temperature over China based on a multi-model ensemble
Introduction
China is located in East Asia, which is greatly influenced by the East Asian monsoon system. The East Asian monsoon brings abundant water vapor in summer and strong cold surges in winter to East Asia, including China. Abnormal East Asian monsoon variability is usually related to disasters, such as floods, droughts and heatwaves in summer or low temperatures, heavy snow and freezing rain in winter (e.g., Tao et al., 1998; Wang et al., 2000; Ding and Chan, 2005; Ding et al., 2008; Wang and He, 2012). Against the background of winter climate changes in recent decades, the wintertime average climate variability has caused widespread concern and raised an important scientific question, upon which numerous studies have sought to address (e.g., Li, 1989; Chen et al., 2000; Wu et al., 2006; Wang et al., 2008; Wang et al., 2009;Wang et al., 2010; Wang et al., 2011). Specifically, the East Asian winter monsoon, which is always closely associated with the winter surface air temperature (SAT) over the Eurasian continent (Wang and Chen, 2010a), is primarily affected by the variability of El Niño–Southern Oscillation (ENSO) and Arctic Oscillation (Wu and Wang, 2002; Wang and Chen, 2010b). The activity of the East Asian winter monsoon could be affected by the joint action of ENSO and the Arctic Oscillation (Cheung et al., 2012), both of which have linear and nonlinear combined impacts on the winter climate over northern and southern China (Chen et al., 2013), but the relationship between ENSO and the East Asian winter monsoon is unstable, with interdecadal variations (He and Wang, 2013). Besides, numerous studies have also focused on the sea-ice anomalies in the Arctic, which tend to induce an abnormal phase of the Arctic Oscillation (Deser et al., 2004; Deser et al., 2007; Honda et al., 2009; Tang et al., 2013), and cold temperature anomalies over northern Eurasia during winter (Zuo et al., 2016).
Above all, the winter temperature anomaly will make impact on safety of lives, transportation, agricultural production and so on. However, compared to the winter season, less attention has been paid to inter-month variation in winter, which is usually inconsistent and more severe than that of the wintertime average. For instance, during 2007/08, the wintertime temperature was relatively cold, but in December specifically the temperature was relatively warm over most parts of China. In January and February, the temperature decreased considerably, which resulted in the occurrence of a rare ice-storm extreme event over south-central China (Wei et al., 2014; Zhou et al., 2011). In contrast with the seasonal mean temperature anomaly in winter 2011/12 of about −4 °C, the intraseasonal temperature anomaly over China had reached −10 °C (Miao et al., 2016). Huang and Hu (2006) demonstrated a difference between the trends of early and late winter temperatures in northern or southern China. As shown in Fig. 1, a salient high-pressure ridge extends from the Siberian–Mongolian High (SMH) down to the northern South China Sea, which reflects the tracks of cold airmass outbreaks from its source region—the Siberian cold dome over the East Asian region (Ding and Krishnamurti, 1987; Zhang et al., 1997). The corresponding boundary layer northerlies prevail from high latitudes to the tropics. However, the intensity of the month-to-month SMH and cold airmass experience a weaker–stronger–weaker inter-monthly variation (Fig. 1a–c), and the SMH anomalies also have an out-of-phase relationship between previous winter and post-winter (Chang and Lu, 2012). One of the reasons for the inconsistent variability among the individual winter months is that the key factors are different. For instance, a large intraseasonal variation exists in the relationship between the winter Arctic Oscillation and different monthly air temperature anomalies over southern China (Zuo et al., 2015), and even the effect of regions of sea ice on the month-to-month variability of winter temperature anomalies over Northeast China are distinct from each other (Dai et al., 2019).
Due to large spacial-temporal difference existing in individual winter month, the out-of-phase variability of winter monthly SAT in China represents a considerable challenge for its climate prediction. No prediction skill exists over China, except for Northeast China in a statistical forecast model (named the snow-cast model) developed by Cohen and Christopher (2006). Dynamical prediction of monthly SAT remains a great challenge. In the Climate Forecast System version 1 (CFSv1) from the National Centers for Environmental Prediction (NCEP) retrospective forecast, the anomaly correlation coefficient of the monthly mean temperature for a 10-day lead is around 0.2 and non-significant over continental northern Eurasia, and the prediction skill vanishes completely as the lead time increases (Chen et al., 2010). Compared with the observation, the standard deviation of the SAT prediction skill in a multi-model ensemble (MME) comprising five state-of-the art models involved in phase 5 of the Coupled Model Intercomparison Project, was much lower, and the variation trend opposite, over most regions of China (Choi et al., 2016).
The Climate Forecast System version 2 (CFSv2) is a fully coupled dynamical prediction system in the form of a global ocean–atmosphere coupled general circulation model (CGCM) that is the successor to CFSv1 (Saha et al., 2006). Yuan et al. (2011) analyzed the capability of CFSv2 for global predictions of SAT and precipitation and found that CFSv2 shows a significant skill enhancement compared to CFSv1. Temperature predictive skill is much higher than precipitation predictive skill in China (Lang et al., 2014). CFSv2 is limited by the coarse resolution that can be improved through spatial downscaling (Yuan and Liang, 2011). BCC_CSM1.1 m is version 1.1 of the seasonal climate prediction model developed at the BCC, China Meteorological Administration (CMA), in which the seasonal prediction of ENSO is very good (Ren et al., 2017), while the prediction skill of winter precipitation over South China is limited (Lu et al., 2017). Furthermore, we also evaluated the TCC for December, January and February SAT prediction in each grid using the two abovementioned state-of-the-art climate models' MME (Fig. 2). As can be seen, the prediction skill that is significant at the 90% confidence level only occurs over western China and a small part of the southeast coast.
Since the various modes of winter temperature in China show distinct characteristics among different months (Jia and Jian, 2015), and the winter monthly SAT prediction in dynamical models is poor, it is urgent that we investigate to what extent the winter monthly SAT over China is predictable and can be improved. In the present study, we focus on the improvement of the prediction of winter monthly SAT over China compared with CGCMs, as well as finding the key factors that influence SAT in China. The remainder of the paper is organized as follows. Section 2 briefly introduces the models, data and methods employed in this study. The selection and confirmation of the predictors is described in section 3. Results and the estimations of the statistical downscaling prediction model are presented in section 4. Section 5 summarizes our major findings.
Section snippets
Coupled models' hindcast data
CFSv2 has been an important component of the monthly to seasonal prediction system of the NCEP's climate prediction. The atmospheric component of CFSv2 is from the NCEP's Global Forecast System at T126 (~0.937°) and the oceanic component is the Modular Ocean Model, version 4.0, at 0.25°–0.5° grid spacing coupled with an interactive three-layer sea-ice model (Saha et al., 2014). For 1982–2010, hindcasts were initiated at 0000, 0600, 1200, and 1800Z on every fifth day, starting from 0000Z on 1
Major modes of winter monthly mean SAT over China and selection of predictors
Before the establishment of SD model, several essential assumptions should be met, which are the close and stationary relations between predictors and predictand, relatively high prediction skill of the predictors from CGCM, and so on (Schoof, 2013). First, the relationships between predictors and SAT over China have been identified. The physical meaning of the two EOF modes of December (or January, February) SAT over China and how they link to lower boundary anomalies will be discussed in this
Validation and prediction with SD models
In order to predict winter monthly SAT over China, a set of SD models is established through the SVD-linear regression method using the predictors introduced in section 3. The idea is to first predict each time expansion coefficient of the predictand and then use the observed SVD pattern and the corresponding predicted time expansion coefficients to make an SAT prediction at each station.
In our case, besides the MME SD models, the potential attainable forecast skill for winter monthly SAT over
Summary and discussion
The 2 (3, 4)-month‑lead monthly forecast skills of dynamical and statistical downscaling models were investigated for the December (January, February) SAT anomaly over China and compared with the potential attainable predictability obtained from the “realistic” SST predictors based the SD model and dynamical MME model for the 36-year period of 1983/84–2018/19.
The winter monthly dynamical prediction for China SAT was achieved by the MME of CFSv2 and BCC_CSM1.1 m with the October initial
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
This work was jointly supported by the National Natural Science Foundation of China (Key Program) (41730964) and National Key Research and Development Program on Monitoring, Early Warning and Prevention of Major Natural Disasters (2018YFC1506000; 2017YFC1502302).
References (76)
- et al.
A review on Northern hemisphere Sea-ice, storminess and the North Atlantic Oscillation: observations and projected changes
Atmos. Res.
(2011) - et al.
Generalizing uncertainty decomposition theory in climate change impact assessments
J. Hydrol. X
(2019) - et al.
Tropical–Extratropical interactions associated with East Asian Cold Air Outbreaks. Part I: interannual variability
J. Clim.
(2017) - et al.
Downscaling precipitation to river basin in India for IPCC SRES scenarios using support vector machine
Int. J. Climatol.
(2008) - et al.
Influence of the Indian Ocean Dipole on the Australian winter rainfall
Geophys. Res. Lett.
(2003) - et al.
Intraseasonal predictability of Siberian high and East Asian winter monsoon and its interdecadal variability
J. Clim.
(2012) - et al.
The interannual variability of East Asian winter monsoon and its relation to the summer monsoon
Adv. Atmos. Sci.
(2000) - et al.
Diagnostic study on the impact of the global sea surface temperature anomalies on the winter temperature anomalies in eastern China in past 50 years
J. Trop. Meteorol.
(2001) - et al.
Prediction of monthly-mean temperature: the roles of atmospheric and land initial conditions and sea surface temperature
J. Clim.
(2010) - et al.
The combined effects of the ENSO and the Arctic Oscillation on the winter climate anomalies in East Asia
Chin. Sci. Bull.
(2013)
Relationship between Ural–Siberian blocking and the East Asian winter monsoon in relation to the arctic oscillation and the El Nino–Southern Oscillation
J. Clim.
Seasonal-to-interannual prediction skills of near-surface air temperature in the CMIP5 decadal hindcast experiments
J. Clim.
Improved skill of northern hemisphere winter surface temperature predictions based on land–atmosphere fall anomalies
J. Clim.
Month-to-Month variability of winter temperature over Northeast China Linked to Sea Ice over the Davis Strait–Baffin Bay and the Barents–Kara Sea
J. Clim.
The effects of North Atlantic SST and sea ice anomalies on the winter circulation in CCM3. Part II: Direct and indirect components of the response
J. Clim.
The transient atmospheric circulation response to North Atlantic SST and sea ice anomalies
J. Clim.
The East Asian summer monsoon: an overview
Meteorog. Atmos. Phys.
Heat budget of the Siberian high and the winter monsoon
Mon. Weather Rev.
The unprecedented freezing disaster in January 2008 in Southern China and its possible association with the global warming
J. Meteor. Res.
Arctic Sea ice and eurasian climate: a review
Adv. Atmos. Sci.
The potential to narrow uncertainty in projections of regional precipitation change
Clim. Dyn.
Oscillating relationship between the East Asian Winter Monsoon and ENSO
J. Clim.
Influence of low Arctic Sea-ice minima on anomalously cold Eurasian winters
Geophys. Res. Lett.
Trends of winter temperature in China
Acta Meteorol. Sin.
Model study on the interannual variability of Asian winter monsoon and its influence
Adv. Atmos. Sci.
Interannual variability of wintertime monthly surface air temperature in China
Clim. Environ. Res.
NCEP-DOE AMIP-II reanalysis (R-2)
Bull. Am. Meteorol. Soc.
Contrasting eastern-Pacific and Central Pacific types of El Niño
J. Clim.
Rank Correlation Methods
Regional Arctic Sea ice variations as predictor for winter climate conditions
Clim. Dyn.
Evaluating skill of seasonal precipitation and temperature predictions of NCEP CFSv2 forecasts over 17 hydroclimatic regions in China
J. Hydrometeorol.
Seasonal prediction and predictability of the Asian winter temperature variability
Clim. Dyn.
Warmer winter in eastern China and El Niño
Chin. Sci. Bull.
Relationship of East Asian winter monsoom and winter Kuroshio SSTA
Jiangxi Meteorol. Sci. Technol.
Relationship between Bering Sea ice cover andEast Asian winter monsoon year-to-year variations
Adv. Atmos. Sci.
A hybrid statistical downscaling model for prediction of winter precipitation in China
Int. J. Climatol.
Two types of surface wind response to the East China Sea Kuroshio Front
J. Clim.
Skillful seasonal predictions of winter precipitation over southern China
Environ. Res. Lett.
Cited by (16)
Research progress on inter-monthly winter temperature variation in East Asia and climate prediction
2023, Atmospheric and Oceanic Science LettersMulti-predictor ensembles improving seasonal prediction of summer rainfall over the Bohai Sea Rim based on statistical downscaling of BCC_CSM1.1 m
2022, Atmospheric ResearchCitation Excerpt :Except for BCC_CSM1.1 m, this deficiency can also be seen in other models like ECMWF System 4, NCEP CFSv2, and GloSea5, etc., in which the summer rainfall predictions are usually skillful over the Yangtze River basin and Southeast China, but low-skill or even unskillful over North China (e.g., BSR) (Kim et al., 2012; Liu et al., 2018; Wang et al., 2021b). These gaps are often bridged by dynamical or statistical downscaling strategies (Nikulin et al., 2012; Kang et al., 2014; Benestad, 2016; Liu et al., 2021). At present, continuous improvements of GCMs and tailored development of downscaling methods are conducted in parallel and complement each other, bringing about the new challenge and opportunity of regional climate prediction.
Hybrid Downscaling Models for Real-Time Predictions of Summer Precipitation in China on a Monthly–Seasonal Scale
2024, Chinese Journal of Atmospheric SciencesSources of the predictability of month-to-month variation of precipitation anomalies in East Asia during summer
2023, International Journal of Climatology