Elsevier

Atmospheric Research

Volume 249, February 2021, 105262
Atmospheric Research

An operational statistical downscaling prediction model of the winter monthly temperature over China based on a multi-model ensemble

https://doi.org/10.1016/j.atmosres.2020.105262Get rights and content

Highlights

  • China winter monthly surface air temperature (SAT) prediction is improved by statistical downscaling.

  • Correlation analysis of empirical orthogonal function principal components is applied to explore the predictors from Arctic sea ice and sea surface temperature.

  • The temporal correlation coefficient skill is improved compared with the original MME prediction.

  • China SAT in December is less predictable than that in January and February by the dynamical and statistical downscaling models, which may be due to the complex predictability source of December SAT.

Abstract

The interannual variation of East Asian winter surface air temperature (SAT) exhibits considerable differences between individual months. However, the prediction skill of dynamical models is rather low; therefore, a statistical downscaling (SD) model was employed to improve the skill for 36 winters during 1983/84–2018/19 in a cross-validated method and for 6 winters during 2013/14–2018/19 in an independent verification. The first two observed empirical orthogonal function (EOF) modes of monthly mean SAT variability explain about 60%–70% of the total variance. Correlation analysis of EOF principal components (PCs) is applied to explore the predictability from key lower boundary anomalies, i.e., Arctic sea ice and sea surface temperature (SST), for SD models. Two previous or simultaneous predictors over distinct areas are identified by the PC correlation analysis for individual monthly China SAT. It is assumed that the simultaneous SST predictors from the multi-model ensemble (MME) of Climate Forecast System version 2 (CFSv2) and BCC_CSM1.1 m involved in the SD model are perfectly predicted, and the potential attainable predictability can be obtained to qualify the highest prediction skills of the SD model. During 1983/84–2018/19, the area-averaged temporal correlation coefficient (TCC) skill of the MME model is 0.10, 0.26 and 0.30 (0.27 at the 90% confidence level) for December, January and February, respectively. The significantly high skill of the cross-validated SD forecast mainly covers extensive domains in central China, and the area-averaged TCC skill is improved compared with the original MME prediction. The corresponding area-averaged attainable TCC skills are 0.26, 0.46 and 0.50 in the SD model, with reference to significant TCC regions in most parts of China. The pattern correlation coefficient skill of the original MME model is much lower than that of the SD model prediction. Independent SD forecasting for a recent 6-year period further reveals that the winter monthly China SAT is highly predictable by the SD model. It should be noted that the China SAT in December is less predictable than that in January and February by the dynamical and SD models, which may be due to the complex predictability source of December SAT. During January and February, the El Niño–Southern Oscillation signal and Indian Ocean dipole mode can be considered as one of the prominent sources of predictability leading to a better prediction skill.

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)

  • J. Bader et al.

    A review on Northern hemisphere Sea-ice, storminess and the North Atlantic Oscillation: observations and projected changes

    Atmos. Res.

    (2011)
  • Y. Kim et al.

    Generalizing uncertainty decomposition theory in climate change impact assessments

    J. Hydrol. X

    (2019)
  • M.R. Abdillah et al.

    Tropical–Extratropical interactions associated with East Asian Cold Air Outbreaks. Part I: interannual variability

    J. Clim.

    (2017)
  • A. Anandhi et al.

    Downscaling precipitation to river basin in India for IPCC SRES scenarios using support vector machine

    Int. J. Climatol.

    (2008)
  • K. Ashok et al.

    Influence of the Indian Ocean Dipole on the Australian winter rainfall

    Geophys. Res. Lett.

    (2003)
  • C. Chang et al.

    Intraseasonal predictability of Siberian high and East Asian winter monsoon and its interdecadal variability

    J. Clim.

    (2012)
  • W. Chen et al.

    The interannual variability of East Asian winter monsoon and its relation to the summer monsoon

    Adv. Atmos. Sci.

    (2000)
  • P.Y. Chen 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)
  • M. Chen et al.

    Prediction of monthly-mean temperature: the roles of atmospheric and land initial conditions and sea surface temperature

    J. Clim.

    (2010)
  • W. Chen et al.

    The combined effects of the ENSO and the Arctic Oscillation on the winter climate anomalies in East Asia

    Chin. Sci. Bull.

    (2013)
  • H.N. Cheung et al.

    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.

    (2012)
  • J. Choi et al.

    Seasonal-to-interannual prediction skills of near-surface air temperature in the CMIP5 decadal hindcast experiments

    J. Clim.

    (2016)
  • J. Cohen et al.

    Improved skill of northern hemisphere winter surface temperature predictions based on land–atmosphere fall anomalies

    J. Clim.

    (2006)
  • H. Dai et al.

    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.

    (2019)
  • C. Deser et al.

    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.

    (2004)
  • C. Deser et al.

    The transient atmospheric circulation response to North Atlantic SST and sea ice anomalies

    J. Clim.

    (2007)
  • Y.H. Ding et al.

    The East Asian summer monsoon: an overview

    Meteorog. Atmos. Phys.

    (2005)
  • Y.H. Ding et al.

    Heat budget of the Siberian high and the winter monsoon

    Mon. Weather Rev.

    (1987)
  • Y.H. Ding et al.

    The unprecedented freezing disaster in January 2008 in Southern China and its possible association with the global warming

    J. Meteor. Res.

    (2008)
  • Y.Q. Gao et al.

    Arctic Sea ice and eurasian climate: a review

    Adv. Atmos. Sci.

    (2015)
  • E. Hawkins et al.

    The potential to narrow uncertainty in projections of regional precipitation change

    Clim. Dyn.

    (2011)
  • S. He et al.

    Oscillating relationship between the East Asian Winter Monsoon and ENSO

    J. Clim.

    (2013)
  • M. Honda et al.

    Influence of low Arctic Sea-ice minima on anomalously cold Eurasian winters

    Geophys. Res. Lett.

    (2009)
  • J.Y. Huang et al.

    Trends of winter temperature in China

    Acta Meteorol. Sin.

    (2006)
  • L. Ji et al.

    Model study on the interannual variability of Asian winter monsoon and its influence

    Adv. Atmos. Sci.

    (1997)
  • D. Jia et al.

    Interannual variability of wintertime monthly surface air temperature in China

    Clim. Environ. Res.

    (2015)
  • M. Kanamitsu et al.

    NCEP-DOE AMIP-II reanalysis (R-2)

    Bull. Am. Meteorol. Soc.

    (2002)
  • H.Y. Kao et al.

    Contrasting eastern-Pacific and Central Pacific types of El Niño

    J. Clim.

    (2009)
  • M.G. Kendall

    Rank Correlation Methods

    (1955)
  • T. Koenigk et al.

    Regional Arctic Sea ice variations as predictor for winter climate conditions

    Clim. Dyn.

    (2016)
  • Y. Lang et al.

    Evaluating skill of seasonal precipitation and temperature predictions of NCEP CFSv2 forecasts over 17 hydroclimatic regions in China

    J. Hydrometeorol.

    (2014)
  • J.Y. Lee et al.

    Seasonal prediction and predictability of the Asian winter temperature variability

    Clim. Dyn.

    (2013)
  • C.Y. Li

    Warmer winter in eastern China and El Niño

    Chin. Sci. Bull.

    (1989)
  • Z.X. Li et al.

    Relationship of East Asian winter monsoom and winter Kuroshio SSTA

    Jiangxi Meteorol. Sci. Technol.

    (2005)
  • F. Li et al.

    Relationship between Bering Sea ice cover andEast Asian winter monsoon year-to-year variations

    Adv. Atmos. Sci.

    (2013)
  • Y. Liu et al.

    A hybrid statistical downscaling model for prediction of winter precipitation in China

    Int. J. Climatol.

    (2015)
  • J.W. Liu et al.

    Two types of surface wind response to the East China Sea Kuroshio Front

    J. Clim.

    (2013)
  • B. Lu et al.

    Skillful seasonal predictions of winter precipitation over southern China

    Environ. Res. Lett.

    (2017)
  • Cited by (16)

    • Multi-predictor ensembles improving seasonal prediction of summer rainfall over the Bohai Sea Rim based on statistical downscaling of BCC_CSM1.1 m

      2022, Atmospheric Research
      Citation 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.

    View all citing articles on Scopus
    View full text