• Abhilash, S., and Coauthors, 2014: Prediction and monitoring of monsoon intraseasonal oscillations over Indian monsoon region in an ensemble prediction system using CFSv2. Climate Dyn., 42, 28012815, doi:10.1007/s00382-013-2045-9.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Adler, R. F., and Coauthors, 2003: The version-2 Global Precipitation Climatology Project (GPCP) monthly precipitation analysis (1979–present). J. Hydrometeor., 4, 11471167, doi:10.1175/1525-7541(2003)004<1147:TVGPCP>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Boyle, J. S., and T. J. Chen, 1987: Synoptic aspects of the wintertime East Asian monsoon. Monsoon Meteorology, C. P. Chang and T. N. Krishnamurti, Eds., Oxford University Press, 125–160.

  • Chan, J. C. L., and C. Y. Li, 2004: The East Asian winter monsoon. East Asian Monsoon, C. P. Chang, Ed., World Scientific Series on Meteorology of East Asia, Vol. 2, World Scientific Publishing, 54–106.

    • Crossref
    • Export Citation
  • Chang, C. P., and K. M. W. Lau, 1980: Northeasterly cold surges and near-equatorial disturbances over the winter MONEX area during December 1974. Part II: Planetary-scale aspects. Mon. Wea. Rev., 108, 298312, doi:10.1175/1520-0493(1980)108<0298:NCSANE>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chang, C. P., and K. M. Lau, 1982: Short-term planetary-scale interactions over the tropics and midlatitudes during northern winter. Part I: Contrasts between active and inactive periods. Mon. Wea. Rev., 110, 933946, doi:10.1175/1520-0493(1982)110<0933:STPSIO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chen, S. J., and L. Dell’Osso, 1987: A numerical case study of a cold air outbreak over East Asia during late spring. Meteor. Atmos. Phys., 37, 237244, doi:10.1007/BF01031045.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chen, T. C., 2002: A North Pacific short-wave train during the extreme phases of ENSO. J. Climate, 15, 23592376, doi:10.1175/1520-0442(2002)015<2359:ANPSWT>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chen, T. C., W. R. Huang, and J. H. Yoon, 2004: Interannual variation of the East Asian cold surge activity. J. Climate, 17, 401413, doi:10.1175/1520-0442(2004)017<0401:IVOTEA>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ding, Y. H., 1994: Monsoon over China. Kluwer Academic Publishers, 432 pp.

  • Ding, Y. H., and T. N. Krishnamurti, 1987: Heat budget of the Siberian high and the winter monsoon. Mon. Wea. Rev., 115, 24282449, doi:10.1175/1520-0493(1987)115<2428:HBOTSH>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ek, M. B., K. E. Mitchell, Y. Lin, E. Rogers, P. Grunmann, G. Gayno, and J. D. Tarpley, 2003: Implementation of Noah land surface model advances in the National Centers for Environmental Prediction operational mesoscale Eta model. J. Geophys. Res., 108, 8851, doi:10.1029/2002JD003296.

    • Search Google Scholar
    • Export Citation
  • Gong, D. Y., and C. H. Ho, 2002: The Siberian high and climate change over middle to high latitude Asia. Theor. Appl. Climatol., 72, 19, doi:10.1007/s007040200008.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Griffies, S. M., M. J. Harrison, R. C. Pacanowski, and A. Rosati, 2004: A technical guide to MOM4. GFDL Ocean Group Tech. Rep. 5, NOAA/GFDL, 342 pp., https://www.gfdl.noaa.gov/bibliography/related_files/smg0301.pdf.

  • Hong, C.-C., H.-H. Hsu, H.-H. Chia, and C.-Y. Wu, 2008: Decadal relationship between the North Atlantic Oscillation and cold surge frequency in Taiwan. Geophys. Res. Lett., 35, L24707, doi:10.1029/2008GL034766.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Huang, W.-R., S. Y. Wang, and J. C. L. Chan, 2011: Discrepancies between global reanalysis and observations in the interdecadal variations of Southeast Asian cold surge. Int. J. Climatol., 31, 22722280, doi:10.1002/joc.2234.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jeong, J.-H., and C.-H. Ho, 2005: Changes in occurrence of cold surges over East Asia in associated with Arctic Oscillation. Geophys. Res. Lett., 32, L14704, doi:10.1029/2005GL023024.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jeong, J.-H., B.-M. Kim, C.-H. Ho, D. Chen, and G.-H. Lim, 2006: Stratospheric origin of cold surge occurrence in East Asia. Geophys. Res. Lett., 33, L14710, doi:10.1029/2006GL026607.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jeong, J.-H., B.-M. Kim, C.-H. Ho, and Y.-H. Noh, 2008: Systematic variation in wintertime precipitation in East Asia by MJO-induced extratropical vertical motion. J. Climate, 21, 788801, doi:10.1175/2007JCLI1801.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jiang, X. W., S. Yang, Y. Li, A. Kumar, X. Liu, Z. Zuo, and B. Jha, 2013a: Seasonal-to-interannual prediction of the Asian summer monsoon in the NCEP Climate Forecast System version 2. J. Climate, 26, 37083727, doi:10.1175/JCLI-D-12-00437.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jiang, X. W., S. Yang, Y. Li, A. Kumar, W. Wang, and Z. Gao, 2013b: Dynamical prediction of the East Asian winter monsoon by the NCEP Climate Forecast System. J. Geophys. Res. Atmos., 118, 13121328, doi:10.1002/jgrd.50193.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kanamitsu, M., W. Ebisuzaki, J. Woollen, S. K. Yang, J. J. Hnilo, M. Fiorino, and G. L. Potter, 2002: NCEP–DOE AMIP-II Reanalysis (R-2). Bull. Amer. Meteor. Soc., 83, 16311643, doi:10.1175/BAMS-83-11-1631.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kumar, K. K., M. Hoerling, and B. Rajagopalan, 2005: Advancing dynamical prediction of Indian monsoon rainfall. Geophys. Res. Lett., 32, L08704, doi:10.1029/2004GL021979.

    • Search Google Scholar
    • Export Citation
  • Kuriyama, K., and M. Yamamoto, 2011: Interannual and synoptic-scale features of two types of cold-air outbreaks over the East China Sea during 1988–2006. Theor. Appl. Climatol., 103, 291304, doi:10.1007/s00704-010-0303-x.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lau, K. M., and C. P. Chang, 1987: Planetary scale aspects of winter monsoon and teleconnections. Monsoon Meteorology, C. P. Chang and T. N. Krishnamurti, Eds., Oxford University Press, 161–202.

  • Lau, N. C., and K. M. Lau, 1984: The structure and energetic of midlatitude disturbances accompanying cold-air outbreaks over East Asia. Mon. Wea. Rev., 112, 13091327, doi:10.1175/1520-0493(1984)112<1309:TSAEOM>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lee, J. Y., and Coauthors, 2010: How are seasonal prediction skills related to models’ performance on mean state and annual cycle? Climate Dyn., 35, 267283, doi:10.1007/s00382-010-0857-4.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Li, Q. P., Y. H. Ding, W. J. Dong, and G. H. Yang, 2007: A numerical study on the winter monsoon and cold surge over East Asia. Adv. Atmos. Sci., 24, 664678, doi:10.1007/s00376-007-0664-y.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lu, F. C., H. M. H. Juang, and C. C. Liao, 2007: A numerical case study of the passage of a cold surge across Taiwan. Meteor. Atmos. Phys., 95, 2752, doi:10.1007/s00703-006-0192-9.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Luo, X., and B. Wang, 2017: How predictable is the winter extremely cold days over temperate East Asia? Climate Dyn., 48, 25572568, doi:10.1007/s00382-016-3222-4.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Moorthi, S., H.-L. Pan, and P. Caplan, 2001: Changes to the 2001 NCEP operational MRF/AVN global analysis/forecast system. NWS Tech. Procedures Bull. 484, 14 pp., http://www.nws.noaa.gov/om/tpb/484.pdf.

  • Nakamura, H., and T. Murakami, 1983: Orographic effects on cold surges and lee-cyclogenesis as revealed by a numerical experiment, Part I: Time mean aspect. J. Meteor. Soc. Japan, 61, 524546, doi:10.2151/jmsj1965.61.4_524.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Park, T. W., J. H. Jeong, C. H. Ho, and S. J. Kim, 2008: Characteristics of atmospheric circulation associated with cold surge occurrences in East Asia: A case study during 2005/06 winter. Adv. Atmos. Sci., 25, 791805, doi:10.1007/s00376-008-0791-0.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Park, T. W., C. H. Ho, S. Yang, and J. H. Jeong, 2010: Influences of Arctic Oscillation and Madden-Julian Oscillation on cold surges and heavy snowfalls over Korea: A case study for the winter of 2009–2010. J. Geophys. Res., 115, D23122, doi:10.1029/2010JD014794.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Park, T. W., C. H. Ho, and S. Yang, 2011a: Relationship between the Arctic Oscillation and cold surges over East Asia. J. Climate, 24, 6883, doi:10.1175/2010JCLI3529.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Park, T. W., C. H. Ho, S. J. Jeong, Y. S. Choi, S. K. Park, and C. K. Song, 2011b: Different characteristics of cold day and cold surge frequency over East Asia in a global warming situation. J. Geophys. Res., 116, D12118, doi:10.1029/2010JD015369.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Park, T. W., C. H. Ho, and Y. Deng, 2014: A synoptic and dynamical characterization of wave-train and blocking cold surge over East Asia. Climate Dyn., 43, 753770, doi:10.1007/s00382-013-1817-6.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Saha, S., and Coauthors, 2006: The NCEP Climate Forecast System. J. Climate, 19, 34833517, doi:10.1175/JCLI3812.1.

  • Saha, S., and Coauthors, 2010: The NCEP Climate Forecast System Reanalysis. Bull. Amer. Meteor. Soc., 91, 10151057, doi:10.1175/2010BAMS3001.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Saha, S., and Coauthors, 2014: The NCEP Climate Forecast System version 2. J. Climate, 27, 21852208, doi:10.1175/JCLI-D-12-00823.1.

  • Simmons, A., S. Uppala, D. Dee, and S. Kobayashi, 2006: ERA-Interim: New ECMWF reanalysis products from 1989 onwards. ECMWF Newsletter, No. 110, ECMWF, Reading, United Kingdom, 25–35, https://www.ecmwf.int/sites/default/files/elibrary/2006/14615-newsletter-no110-winter-200607.pdf.

  • Takaya, K., and H. Nakamura, 2005: Mechanism of intraseasonal amplification of the cold Siberian high. J. Atmos. Sci., 62, 44234440, doi:10.1175/JAS3629.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Vitart, F., A. W. Roberston and S2S Steering Group, 2015: Sub-seasonal to seasonal prediction: Linking weather and climate. Seamless Prediction of the Earth System: From Minutes to Months, G. Brunet, S. Jones, and P. M. Ruti, Eds., WMO-1156, World Meteorological Organization, 385–401, http://library.wmo.int/pmb_ged/wmo_1156_en.pdf.

  • Vitart, F., and Coauthors, 2017: The Subseasonal to Seasonal (S2S) Prediction project database. Bull. Amer. Meteor. Soc., 98, 163173, doi:10.1175/BAMS-D-16-0017.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wang, B., and Q. Ding, 2008: Global monsoon: Dominant mode of annual variation in the tropics. Dyn. Atmos. Oceans, 44, 165183, doi:10.1016/j.dynatmoce.2007.05.002.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Woo, S. H., B. M. Kim, J. H. Jeong, S. J. Kim, and G. H. Lim, 2012: Decadal changes in surface air temperature variability and cold surge characteristics over northeast Asia and their relation with the Arctic Oscillation for the past three decades (1979–2011). J. Geophys. Res., 117, D18117, doi:10.1029/2011JD016929.

    • Search Google Scholar
    • Export Citation
  • Wu, B., and J. Wang, 2002: Winter Arctic oscillation, Siberian high and East Asian winter monsoon. Geophys. Res. Lett., 29, 1897, doi:10.1029/2002GL015373.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wu, M. C., and J. C. L. Chan, 1997: Upper-level features associated with winter monsoon surges over South China. Mon. Wea. Rev., 125, 317340, doi:10.1175/1520-0493(1997)125<0317:ULFAWW>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhang, Y., and W. C. Wang, 1997: Model-simulated northern winter cyclone and anticyclone activity under a greenhouse warming scenario. J. Climate, 10, 16161634, doi:10.1175/1520-0442(1997)010<1616:MSNWCA>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhu, J., and J. Shukla, 2013: The role of air–sea coupling in seasonal prediction of Asia–Pacific summer monsoon rainfall. J. Climate, 26, 56895697, doi:10.1175/JCLI-D-13-00190.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhu, J., B. Huang, Z. Z. Hu, J. L. Kinter III, and L. Marx, 2013: Predicting U.S. summer precipitation using NCEP Climate Forecast System version 2 initialized by multiple ocean analyses. Climate Dyn., 41, 19411954, doi:10.1007/s00382-013-1785-x.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • View in gallery

    Domains of East Asia (box outlined by dashed lines) and the SH (box outlined by solid lines). Black dots show the stations in eastern China. Shadings indicate the climatological values of SLP during the period from November to March. Blue contour shows the Tibetan Plateau terrain of 1500 m.

  • View in gallery

    SLP (contours) and 850-hPa winds (vectors) for (a) observations, (b) CFSv2 predictions for LD 0–14, and (c) differences in SLP (shading) and winds (vectors) between ensemble-mean predictions and observations.

  • View in gallery

    As in Fig. 2, but for T2m (contours) and precipitation (shading).

  • View in gallery

    Five-day moving averages of daily precipitation and T2m averaged over region 1 (35°–50°N, 105°–125°E) and region 2 (20°–35°N, 105°–125°E) in observations and CFSv2 predictions at different lead times.

  • View in gallery

    Five-day moving averages of daily SLP averaged over the SH region (35°–55°N, 90°–115°E) in observations and CFSv2 predictions at different lead times.

  • View in gallery

    (a) Number of cold surge days and (b) number of cold surge events in observations and CFSv2 predictions at different lead times.

  • View in gallery

    Composite anomalies of SLP (shading) and 850-hPa winds (vectors) for observations and CFSv2 predictions for LD 0–4 and LD 5–9 during day −3, day 0, and day +3 relative to all cold surge occurrences.

  • View in gallery

    Composite anomalies of geopotential height at 500 hPa (shading) and winds (vectors) at 200 hPa for (a)–(c) observations and (d)–(f) CFSv2 predictions for LD 0–14.

  • View in gallery

    Composite anomalies of precipitation (shading) and T2m (contours) for observations and CFSv2 predictions for LD 0–4 and LD 5–9.

  • View in gallery

    Correlation between observations and predictions at different lead times (horizontal axis) for (a) SLP over the SH area (35°–55°N, 90°–115°E) and (b) winds averaged over East Asia (20°–50°N, 80°–140°E); (c) precipitation and (d) T2m averaged over eastern China (20°–55°N, 105°–135°E), northern region, and southern region, respectively. Shown are three-point running averages along forecast lead days. Black dashed lines denote a statistically significant correlation at the 99% confidence level.

  • View in gallery

    Spatial distributions of TCCs between observations and CFSv2 predictions for LD 0–14 for (a) SLP, (b) meridional wind at 850 hPa, (c) precipitation, and (d) T2m. Shaded areas denote statistically a significant correlation at the 99% confidence level.

  • View in gallery

    Number of cold surge days that occurred in (a) region 1 and (b) region 2 for observations and CFSv2 predictions for LD 0–4 and LD 5–9.

  • View in gallery

    Composite anomalies of SLP (shading) and horizontal winds at 850 hPa (vectors) for the cold surge days in (top) northern and (middle) southern region (a),(b) for observations and (d),(e) for CFSv2 predictions for LD 0–14. (c),(f) Differences between northern and southern China and (g),(h) differences between prediction and observation.

  • View in gallery

    As in Fig. 13, but for geopotential height at 500 hPa (shading) and winds at 200 hPa (vectors).

  • View in gallery

    As in Fig. 13, but for precipitation (shading) and T2m (contours).

  • View in gallery

    Differences between ensemble-mean predictions for LD 0–14 and ERA-Interim for (a) SLP (shading) and 850-hPa winds (vectors) and (b) T2m. Also shown are the composite anomalies from the ERA-Interim for (c) SLP (shading) and 850-hPa winds (vectors) and (d) Z500 (shading) and 200-hPa winds (vectors) on day 0 of a cold surge occurrence.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 75 75 19
PDF Downloads 74 74 17

Subseasonal Dynamical Prediction of East Asian Cold Surges

View More View Less
  • 1 National Climate Center, China Meteorological Administration, Beijing, and Zhuhai Joint Innovative Center for Climate-Environment-Ecosystem, Zhuhai Key Laboratory of Dynamics Urban Climate and Ecology, Future Earth Research Institute, Beijing Normal University, Zhuhai, China
  • 2 Zhuhai Joint Innovative Center for Climate-Environment-Ecosystem, Zhuhai Key Laboratory of Dynamics Urban Climate and Ecology, Future Earth Research Institute, Beijing Normal University, Zhuhai, and School of Atmospheric Sciences, and Guangdong Province Key Laboratory for Climate Change and Natural Disaster Studies, Sun Yat-sen University, Guangzhou, China
  • 3 National Climate Center, China Meteorological Administration, Beijing, China
© Get Permissions
Full access

Abstract

Predictability of East Asian cold surges is studied using daily data from the hindcasts of 45-day integrations by the NCEP Climate Forecast System version 2 (CFSv2). Prediction skills of the CFSv2 in forecasting cold surges, their annual variation, and their physical links to large-scale atmospheric circulation patterns are examined. Results show that the climatological characteristics of the East Asian winter monsoon can be reasonably reproduced by the CFSv2. The model can well capture the frequency, intensity, and location of cold surges at a lead time of about two weeks. Obviously, fewer-than-observed cold surge days are found in the predictions when the lead time is above 14 days. The spatiotemporal evolutions of high-, mid-, and low-level circulation patterns during cold surge occurrences are all accurately indicated in the CFSv2 prediction. Except for precipitation, the other variables associated with cold surges, such as geopotential height, wind, sea level pressure, and surface air temperature, exhibit higher skills. The lead time of skillful prediction of precipitation is limited to around 1 week, with systematic wet biases over the South China Sea, the Philippine Islands, and the northwest Pacific, but dry biases over India, the Indo-China Peninsula, and most high-latitude regions. Wave train–like patterns of geopotential height and wind differ distinguishably when cold surges occur in northern and southern regions (using 35°N as the dividing line), and the CFSv2 gives a consistent prediction to these anomalous patterns. A weaker-than-observed Siberian high and weaker northerly winds over eastern China are found in the predictions especially at longer lead times.

Denotes content that is immediately available upon publication as open access.

© 2017 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Song Yang, yangsong3@mail.sysu.edu.cn

Abstract

Predictability of East Asian cold surges is studied using daily data from the hindcasts of 45-day integrations by the NCEP Climate Forecast System version 2 (CFSv2). Prediction skills of the CFSv2 in forecasting cold surges, their annual variation, and their physical links to large-scale atmospheric circulation patterns are examined. Results show that the climatological characteristics of the East Asian winter monsoon can be reasonably reproduced by the CFSv2. The model can well capture the frequency, intensity, and location of cold surges at a lead time of about two weeks. Obviously, fewer-than-observed cold surge days are found in the predictions when the lead time is above 14 days. The spatiotemporal evolutions of high-, mid-, and low-level circulation patterns during cold surge occurrences are all accurately indicated in the CFSv2 prediction. Except for precipitation, the other variables associated with cold surges, such as geopotential height, wind, sea level pressure, and surface air temperature, exhibit higher skills. The lead time of skillful prediction of precipitation is limited to around 1 week, with systematic wet biases over the South China Sea, the Philippine Islands, and the northwest Pacific, but dry biases over India, the Indo-China Peninsula, and most high-latitude regions. Wave train–like patterns of geopotential height and wind differ distinguishably when cold surges occur in northern and southern regions (using 35°N as the dividing line), and the CFSv2 gives a consistent prediction to these anomalous patterns. A weaker-than-observed Siberian high and weaker northerly winds over eastern China are found in the predictions especially at longer lead times.

Denotes content that is immediately available upon publication as open access.

© 2017 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Song Yang, yangsong3@mail.sysu.edu.cn

1. Introduction

Cold surges are the most conspicuous weather events during winter in East Asia, especially those associated with intensified Siberian high (SH), and exert a major impact on socioeconomical human activities. During the occurrence of a cold surge, cold air breaks out and moves southward along the edge of the expanded SH, leading to severe weather events such as a large temperature drop, strong wind, and heavy freezing rain or snowfall (Boyle and Chen 1987; Ding 1994; Chan and Li 2004; Jeong et al. 2008). Early winter monsoon experiments and studies have revealed numerous observational facts about the East Asian winter monsoon (EAWM) and cold surges. The large-scale circulation background of cold surges has also been understood reasonably well (Chang and Lau 1980, 1982; Ding and Krishnamurti 1987). Lau and Lau (1984) and Chen (2002) found that cold surges originated from the upper tropospheric shortwave trough over Lake Baikal and grew as the trough traveled toward East Asia. Jeong et al. (2006) observed large anomalies of stratospheric negative potential vorticity and an increase in geopotential height over northern Eurasia about 1 week before a cold surge occurrence.

Previous studies have focused on the factors that affect the occurrence of cold surges and addressed the influence of large-scale climate phenomena on cold surges. The amplification of the SH and its expansion into East Asia are well known as an essential factor for the generation and maintenance of cold surges (Zhang and Wang 1997; Gong and Ho 2002; Takaya and Nakamura 2005). Several studies have indicated that the Arctic Oscillation (AO) influences cold surge development significantly through the variations of SH and upper-tropospheric circulation (Jeong and Ho 2005; Hong et al. 2008; Park et al. 2010, 2011a). In addition, the variation of upper-level circulation is affected by tropical variability such as El Niño–Southern Oscillation (ENSO; Chen et al. 2004) and the Madden–Julian oscillation (MJO; Jeong et al. 2008). Cold surges also exhibit significant interannual and decadal variations in frequency (Chen et al. 2004; Jeong and Ho 2005; Kuriyama and Yamamoto 2011), and these variations are well correlated with the AO and the North Atlantic Oscillation (NAO; Jeong and Ho 2005; Hong et al. 2008; Woo et al. 2012; Park et al. 2014).

Numerical studies of cold surges have been mainly focused on sensitivity experiments and case studies about the impacts of orographic forcing and latent heating (Nakamura and Murakami 1983; Chen and Dell’Osso 1987; Lu et al. 2007). Model capability of simulating the EAWM and the SH that are related to the occurrence frequency and location of cold surges has also been evaluated (Li et al. 2007).

Dynamical prediction of winter mean temperature remains a great challenge in spite of the prominent progress that has been made in understanding and predicting the EAWM. Compared with weather forecasting and seasonal climate prediction, subseasonal forecasting is generally less skillful, especially for the prediction of winter extremes. The subseasonal time scale is considered a difficult time range since the lead time is sufficiently long for the loss of much of the memory of atmospheric initial conditions and may be too short for the impact of oceans to become effective. Nevertheless, previous studies have indicated potentially important sources of predictability for this time range (e.g., Vitart et al. 2015). The subseasonal predictability originates from the natural modes of variability (e.g., ENSO, MJO, quasi-biennial oscillation, and Indian Ocean dipole), slowly varying processes (e.g., involving ocean heat content, soil moisture, snow, and sea ice), and elements of external forcing (e.g., aerosols; Vitart et al. 2015, 2017). Previous studies have also indicated that the anomalous lower boundary forcing from oceans and land surfaces is an important source of the predictability for extreme cold days over East Asia (Luo and Wang 2017).

Capability of dynamical models in the subseasonal prediction of extreme cold conditions has been seldom assessed in the past. Indeed, very few studies on the dynamical prediction of East Asian cold surges can be found. The Climate Forecast System version 2 (CFSv2; Saha et al. 2014) of the National Centers for Environmental Prediction (NCEP), one of the most popular operational climate prediction models, exhibited considerable skill in predicting the Asian monsoon (e.g., Jiang et al. 2013a,b; Zhu and Shukla 2013) and North American climate (e.g., Zhu et al. 2013). In this study, we examine the skill of CFSv2 in forecasting wintertime cold surges in East Asia. The capability of CFSv2 in capturing cold surges, their variations, and their physical links to large-scale atmospheric circulation patterns are discussed. The following questions are addressed using the CFSv2. What is the model capability of forecasting winter climatological features of temperature and precipitation in East Asia? What are the skills of prediction of the frequency and intensity of cold surges? How different are the biases of predicting cold surges at various leads of time?

In section 2, model output, observational data, and analysis methods are briefly described. The climatology of precipitation, surface air temperature, and atmospheric circulation predicted by the CFSv2 is examined in section 3. Skills of prediction of cold surge frequency and variation with different time leads are evaluated in section 4. Regional features of cold surges are discussed in section 5. Summary and discussion are given in section 6.

2. Data and analysis methods

a. Model output and observational data

In this study, daily output from the hindcast experiments by the CFSv2 is analyzed. The CFSv2 (Saha et al. 2010) is a fully coupled atmosphere–ocean–land–sea ice system for dynamical seasonal prediction, which replaced the CFS version 1 (Saha et al. 2006) for climate operation at the NCEP in March 2011. The atmospheric component is the NCEP Global Forecast System (Moorthi et al. 2001) with a T126 horizontal resolution and 64 hybrid sigma-pressure layers. The ocean component is the Geophysical Fluid Dynamics Laboratory Modular Ocean Model version 4 (MOM4; Griffies et al. 2004). The land surface model is the NCEP, Oregon State University, Air Force, and Hydrologic Research Laboratory model (Noah; Ek et al. 2003). Each component model is coupled without flux adjustment, and initial conditions are obtained from the Climate Forecast System Reanalysis (CFSR; Saha et al. 2010). The CFSv2 has greatly improved forecast skills in many aspects, such as the MJO, surface air temperature (at 2 m) over the United States, and global sea surface temperature compared to the CFS version 1 (Saha et al. 2014).

We analyze the outputs from the retrospective forecast of 45-day integrations by the CFSv2 initiated from every 0000, 0600, 1200, and 1800 UTC cycle from 1999 to 2010. Specifically, the daily outputs of 500-hPa geopotential height, 850- and 200-hPa winds, sea level pressure (SLP), surface air temperature at 2 m (T2m), and precipitation are analyzed. All hindcasts were integrated for 45 days. Thus, for a specific target day, we have forecasts of 45 time leads. Lead 0 is defined as the model run initialized on the target day, lead 1 denotes the model run initialized on the previous day of the target day, and so on. In sections 35, the ensemble predictions for each day are divided into several groups according to the length of lead time, that is, 0–4 days, 5–9 days, and so on.

For forecast evaluation, daily means of T2m, SLP, geopotential height, and winds from the NCEP–Department of Energy (DOE) Global Reanalysis 2 (R-2; Kanamitsu et al. 2002) are used. Global Precipitation Climatology Project (GPCP) version 2.1 is a merged precipitation dataset (Adler et al. 2003) and is also used as a validation dataset here. In addition, T2m and precipitation from station observations, obtained from the China Meteorological Administration, are used in cold surge definition and model evaluation. Considering the integrity of observational data, we choose 431 stations in eastern China without missing data. Limited by model hindcast output, our analyses are carried out for 11 winters (November–March) from 1999/2000 to 2009/10.

b. Definition of cold surges

Objective criteria to identify cold surges are required in this study. Generally, the typical characteristics of a cold surge in East Asia are represented as a steep drop of T2m with an expansion of the SH (Lau and Chang 1987; Zhang and Wang 1997; Chen et al. 2004; Jeong and Ho 2005; Park et al. 2010). However, the definitions in earlier studies varied depending on the regions of focus, among others. Here, we focus on the cold surges that affect eastern China, and thus we adopt the criteria in Park et al. (2010, 2011a,b) with slight modifications as follows.

The expansion of the SH to southern Siberia is related to cold surge occurrence in East Asia (Zhang and Wang 1997). The region of southern Siberia indicated in Fig. 1 (35°–55°N, 90°–115°E) is regarded as the domain of the SH (Park et al. 2010, 2011a,b). As the first criterion, we identify the days of strong SH when SLP and the magnitude of relative vorticity at the center of the surface anticyclone over the SH domain exceed 1035 hPa and 1.0 × 10−5 s−1, respectively (Park et al. 2010). The center of the surface anticyclone is defined as the grid point where the geopotential height at 1000 hPa is larger than the values of the eight surrounding grid points (Zhang and Wang 1997).

Fig. 1.
Fig. 1.

Domains of East Asia (box outlined by dashed lines) and the SH (box outlined by solid lines). Black dots show the stations in eastern China. Shadings indicate the climatological values of SLP during the period from November to March. Blue contour shows the Tibetan Plateau terrain of 1500 m.

Citation: Weather and Forecasting 32, 4; 10.1175/WAF-D-16-0209.1

Different criteria of T2m drop were used in previous studies. Zhang and Wang (1997) required T2m to drop 8°–10°C in central China or 5°–7°C in southern China within 24–48 h of cold surge onset. Huang et al. (2011) required a T2m drop of at least 4°C. Here, we use the standard of T2m drop following Park et al. (2010) as the second criterion. We first divide the 431 China stations into 5 × 5 grid boxes in the domain of East Asia (20°–55°N, 105°–135°E; see the dashed box in Fig. 1) and calculate the average T2m values in these boxes. We then define that both the day-to-day drop of T2m (T2mt − T2mt−1) and the averaged daily T2m anomaly should exceed 1.5σ (σ is the standard deviation of T2m during the winters of our analysis) in more than one grid box.

3. Climatological features of temperature, precipitation, and circulation in observations and CFSv2

Previous studies have indicated that cold surges occur frequently from November to March of the following year and that the period can be considered the same as the EAWM season (e.g., Zhang and Wang 1997), the same period used in this study. Before analyzing the model’s capability of forecasting cold surge occurrence, ensemble means of winter predictions at different leads (days) from 1999 to 2010 are examined. Figures 2a–c present 11-yr means of SLP and 850-hPa wind in the observations, CFSv2 predictions, and model biases, respectively. The CFSv2 predictions are obtained from the ensemble means of time leads of 0–14 days (see section 2 for specification of initial conditions). It can be seen from Fig. 2 that a strong high pressure exists in both CFSv2 and observations, with central pressure exceeding 1032 hPa over central-eastern Siberia and Mongolia (around 45°N). At the lower level (850 hPa), northwesterly flow appears along the eastern edge of the SH, northern China, and the Sea of Japan, with a maximum speed of 8 m s−1. Southerly flow is seen ahead of a trough near the northern South China Sea. The averaged bias of CFSv2 with respect to observations (Fig. 2c) indicates that the SH is more extensive but its central pressure is slightly lower than the observed. Moreover, weaker-than-observed northwesterly flow at the lower level and anticyclonic wind bias over the northwestern Pacific are found in the CFSv2. The large northerly wind speed in the lower troposphere is also one of the important factors to depict cold surges (Chang and Lau 1980). However, further analysis indicates that the maximum meridional wind speed and the frequency of strong northerly wind (with υ > 8 m s−1) at 850 hPa in the CSFv2 are smaller than those in observations. Overall, the spatial patterns of biases from the ensemble predictions are similar with those of other lead days (not shown), implying a stable maintenance of prediction biases from the beginning of the forecast. In observations, strong westerly winds are found at the upper level (200 hPa, figure not shown) over East Asia. The westerly jet stream is located near 32°N, with a maximum speed of 68 m s−1. The predicated location and intensity of the jet center are consistent with observations, except that stronger-than-observed maximum wind speed is found in the CFSv2. The coefficients of temporal correlation of geopotential height (at 1000 and 500 hPa), SLP, and zonal and meridional winds (at 850 hPa) between observations and predictions for different time leads have been analyzed (figures not shown). Significant correlation can be found over the analysis domain when the lead time is within 20 days. The differences and root-mean-square errors between observations and predictions present noticeable bias over the Tibetan Plateau and Okhotsk (e.g., SLP and geopotential height at 500 hPa). The standard deviations of the above variables in the observations and predictions indicate that the CFSv2 can reasonably reproduce the interannual variability as observed (figures not shown).

Fig. 2.
Fig. 2.

SLP (contours) and 850-hPa winds (vectors) for (a) observations, (b) CFSv2 predictions for LD 0–14, and (c) differences in SLP (shading) and winds (vectors) between ensemble-mean predictions and observations.

Citation: Weather and Forecasting 32, 4; 10.1175/WAF-D-16-0209.1

Figure 3 shows the wintertime mean precipitation and T2m in observations and CFSv2 predictions, along with their differences. The main precipitation belt is over the northwest Pacific and to the east of the Philippines. Comparison between Figs. 3a and 3b indicates that the CFSv2 captures the observed features in general, including the location and magnitude of precipitation centers. However, the CFSv2 overestimates the precipitation over southern China, the Philippine islands, and northwest Pacific, together with a wet bias south of the Tibetan Plateau. The observed T2m demonstrates that the temperature gradient between land and oceans is large, and the temperature contours are almost parallel to the latitude circles, with a mean temperature of 16°C over southern China but −20°C over northern China. The hindcasts using the CFSv2 reproduce the spatial pattern of T2m, although cold biases are found over northern regions, especially in western Siberia and northern China. On the contrary, warm biases are found over eastern China, the Korean Peninsula, and Japan (Fig. 3c). Cold biases can also be found over the western Tibetan Plateau. Because of the complex topography of the Tibetan Plateau, it is difficult for climate models to produce realistic simulations and decent predictions of regional precipitation and T2m. In addition, because of the lack of adequate observations over the western Tibetan Plateau, it is quite difficult to confirm the model’s biases over this region.

Fig. 3.
Fig. 3.

As in Fig. 2, but for T2m (contours) and precipitation (shading).

Citation: Weather and Forecasting 32, 4; 10.1175/WAF-D-16-0209.1

Figures 3a and 3b show apparent regional differences in precipitation and T2m between observations and CFSv2, with a wet–warm condition south of 35°N and a dry–cold condition north of it. Meanwhile, over the Tibetan Plateau few station observations are available and large uncertainties exist in observational data. Thus, we divide eastern China (east of 105°E) into two subregions (northern region and southern region, divided by the latitude of 35°N) to gain further understanding of the skills of subseasonal prediction of precipitation and T2m at different leads. Figures 4a–d show daily precipitation and T2m averaged over region 1 (35°–50°N, 105°–125°E) and region 2 (20°–35°N, 105°–125°E) for observations and CFSv2 ensemble mean predictions, respectively. To remove daily fluctuations, the time series are smoothed by a 5-day moving average. Slightly more precipitation over region 1 is found in mid-November and March of the following year. In observations, precipitation over region 2 increases steadily since mid-February. The precipitation from the CFSv2 forecasts with leads of 0–4, 5–9, and 10–14 days shows a good agreement with the observations, with stronger correlation over region 2. The CFSv2 overestimates precipitation over both regions, and the consistency between observations and predictions drops obviously with increasing lead time. The ensemble averaged predictions beyond the 25-day lead time can hardly reflect the observed intraseasonal variation. Figures 4c and 4d indicate that the minimum values of observed T2m in both regions occur in late January. Like precipitation, the T2m from the hindcasts at leads from 0–4 days to 20–24 days yields similar evolution characteristics as the observation, in spite of a warm bias during the entire winter.

Fig. 4.
Fig. 4.

Five-day moving averages of daily precipitation and T2m averaged over region 1 (35°–50°N, 105°–125°E) and region 2 (20°–35°N, 105°–125°E) in observations and CFSv2 predictions at different lead times.

Citation: Weather and Forecasting 32, 4; 10.1175/WAF-D-16-0209.1

The semipermanent SH spanning the Eurasian continent is a key element of EAWM, which is linked to cold surges (Lau and Chang 1987; Chan and Li 2004). The cold surge frequency varies coherently with SH fluctuations on both interannual and decadal time scales (Wu and Chan 1997; Zhang and Wang 1997; Jeong and Ho 2005). Compared with the winter AO, the SH shows a more direct and significant influence on the EAWM, and the impact of the SH on surface air temperature is primarily seen to the south of 50°N over East Asia (Wu and Wang 2002). The correlations of T2m and the SH intensity index in both observations and CFSv2 predictions have also been calculated (figure not show). The SH intensity is defined by the SLP averaged over the SH region (35°–55°N, 90°–115°E). Significantly negative correlation can be found over the Asian continent, and the CSFv2 reproduces the observed relationship. Figure 5 shows that the predictions at short leads exhibit high skills for the intraseasonal oscillations of SLP. Not surprisingly, the skill decreases with lead time, but still remains significant when the lead time is beyond 20 days.

Fig. 5.
Fig. 5.

Five-day moving averages of daily SLP averaged over the SH region (35°–55°N, 90°–115°E) in observations and CFSv2 predictions at different lead times.

Citation: Weather and Forecasting 32, 4; 10.1175/WAF-D-16-0209.1

4. Prediction skill of cold surges

Using the criteria of East Asian cold surge mentioned in section 2, 94 cold surge days are identified from observations in the 11 winters (November–March) from 1999/2000 to 2009/10. Approximately 10 cold surges sweep across East Asia each winter, which is consistent with the results from previous studies (e.g., Park et al. 2010). The mean occurrence frequency identified by the present study may be somewhat lower than that in other studies because of the use of different criteria (Zhang and Wang 1997; Chen et al. 2004; Jeong and Ho 2005). Figure 6 shows the year-to-year variations of cold surge days in observations and CFSv2 ensemble mean predictions at different leads. Apparent interannual differences existed during the 11 winters, with 16 cold surges in 2001/02 but only one cold surge in 2005/06. Figure 6 suggests that the CFSv2 captures about 90% cold surge days only when lead time is shorter than 5 days. The cold surge days predicted by the model drop gradually with increasing lead time, and they decrease to below 50% when the lead time is longer than 14 days. The occurrence days of cold surges in the ensemble-averaged predictions are obviously less than those in observations. Five extremely strong cold surges are detected in the observations on 1 January 2000, 12 December 2001, 5 December 2005, 21 December 2008, and 7 March 2010. When these cold surges occurred, the maximum SLP at the center exceeded 1055 hPa and the day-to-day drop of T2m exceeded 8°C. The CFSv2 successfully captures all the strong cold surge cases. The occurrence time and intensity of the cold surges are also well predicted by the model at short lead times.

Fig. 6.
Fig. 6.

(a) Number of cold surge days and (b) number of cold surge events in observations and CFSv2 predictions at different lead times.

Citation: Weather and Forecasting 32, 4; 10.1175/WAF-D-16-0209.1

A strong cold surge process usually lasts for a few days. An examination of the date of each cold surge in observations and predictions indicates that the dates identified in the predictions are identical to those in the observations in most cases. In other cases, however, the observed cold surges persisted longer than the predicted cold surges by several days. In other words, although the CFSv2 can capture the same cold surge process, sometimes the persistent time is shorter than that in the observations, leading to less-than-observed cold surge days. Then, we take the cold surges that persisted more than 2 days as one process or event. Figure 6b shows the annual number of cold surge events in both observations and predictions. A total of 94 cold surge days are detected from the observations, while only 84 days detected from the CFSv2 with leads (LD) of 0–4 days. If the persistent days are not considered, a total of 62 and 60 events are detected from the observations and the predictions, respectively. It is also noted that the CFSv2 captures more than 60% of the cold surge events when lead time is shorter than 9 days. For example, the ensemble predictions with leads of 0–4 days successfully forecast 60 events, with an interannual variation that agrees with the observations. Unfortunately, obviously fewer-than-observed cold surge events are predicted at longer leads. The predicted cold surge frequency decreases rapidly to about 50% of the observed at the lead time of 10 days.

The large-scale circulation patterns and climate anomalies related to cold surges have been investigated previously (e.g., Zhang and Wang 1997; Chen 2002; Park et al. 2008, 2010). Closely associated with the circulation pattern of EAWM, strong northerly wind is one of the main characteristics of cold surge occurrences (Ding and Krishnamurti 1987; Zhang and Wang 1997). Amplification of the SH and the Aleutian low is another important feature. To evaluate the skill of CFSv2 in predicting the associated atmospheric circulation and its temporal evolution, we analyze the composite anomalies of several key variables in observations and ensemble forecasts. Here, the composite anomalies of the forecasts in each lead time are for the cold surge days detected from the observations. A comparison between the observations and the forecasts is helpful for understanding the possible causes for prediction bias of cold surges in the model. The anomalies of SLP and horizontal wind at 850 hPa in observations and predictions on day −3, day 0, and day +3 relative to the date of cold surge occurrence are shown in Fig. 7. In the observations, a stronger-than-normal SH controls all of northern Eurasia on day −3, with a positive anomalous center over 50°–70°N, 60°–110°E. At 850 hPa, the northerly flow along the edge of the SH meets the southerly wind near 40°N. Meanwhile, a cyclone–anticyclone couplet appears over the midlatitudes of East Asia (Fig. 7a). On the day of cold surge occurrence, the SLP gradient along the south coast of China intensifies and the SH shifts to Inner Mongolia, while the low over the oceans strengthens and extends westward, forming a couplet. The amplified and southeastward shifted high-pressure system leads to an enhancement of northwesterly wind over the ocean to the east of China. Furthermore, the previously existing cyclone–anticyclone couplet at the lower levels over East Asia is replaced by an anticyclone–cyclone (opposite sign) couplet, and eastern China is controlled by a strong northerly wind (Fig. 7b). After 3 days of the cold surge occurrence, the high-pressure center weakens sharply and shifts further southeastward, accompanied by weakening northerly winds. The circulation patterns and their evolutions are successfully predicted by the CFSv2 14 days in advance, and the predicted location and intensity of the SH, the Aleutian low, and the lower-level dipole mode are in good agreement with the observations (Figs. 7d–i, which show the ensemble predictions at 0–4 and 5–9 lead days). The CFSv2 also captures the observed circulation pattern even for the forecasts at much longer leads (15–39 days), although weaker-than-observed SH is found in the predictions with increasing lead time (not shown).

Fig. 7.
Fig. 7.

Composite anomalies of SLP (shading) and 850-hPa winds (vectors) for observations and CFSv2 predictions for LD 0–4 and LD 5–9 during day −3, day 0, and day +3 relative to all cold surge occurrences.

Citation: Weather and Forecasting 32, 4; 10.1175/WAF-D-16-0209.1

Figure 8 shows composite anomalies to depict the time evolution of geopotential height at 500 hPa (Z500) and upper-level wind at 200 hPa during cold surges. A positive anomaly center of Z500 (or a ridge; shading) is found around western Russia (50°–100°E and 45°–75°N), a negative center (or a trough) over Mongolia (80°–130°E and 40°–55°N), and another positive center over the Korean Peninsula, Japan, and the North Pacific (120°–160°E and 25°–45°N). The positive–negative–positive centers are aligned in the northwest–southeast direction and form a wave train on day −3 relative to the occurrence day of cold surges (Fig. 8a). The wave train propagates southeastward on day 0 with the trough deepening and extending along a northeast–southwest direction. On day +3, the wave train continues to shift southeastward. In particular, the positive anomaly center, which shifts toward Mongolia, weakens obviously or even disappears (Fig. 8c). The upper-level wind also exhibits wave train features, consistent with the tropospheric trough and ridge. The air temperature pattern well matches the features in geopotential height (i.e., warm–cold–warm wave; not shown). Lower-level cold advection (at 850 hPa) exits on the back of the tropospheric trough, leading to expansion of the SH and development of a cold trough. The CFSv2 shows successful ensemble predictions when the leads are 0–4, 5–9, and 10–14 days. Figures 8d–f present the ensemble predictions with leads of 0–14 days, which indicates that the CFSv2 predicts accurately the location and propagation feature of the wave train. However, small bias is found in the predicted intensity of the tropospheric trough and ridge during the entire process of cold surges.

Fig. 8.
Fig. 8.

Composite anomalies of geopotential height at 500 hPa (shading) and winds (vectors) at 200 hPa for (a)–(c) observations and (d)–(f) CFSv2 predictions for LD 0–14.

Citation: Weather and Forecasting 32, 4; 10.1175/WAF-D-16-0209.1

Figure 9 presents the composite anomalies of precipitation (shading) and T2m (contour) of observations and CFSv2 ensemble mean predictions at different lead times. On the day of cold surge occurrence, abnormally cold temperatures control China. A negative center is located from eastern Inner Mongolia to the northern Yangtze River basin, with values below −5°C. Meanwhile, the eastern oceanic region is obviously warmer than normal. Precipitation occurs mainly in southern China, coastal areas, and the eastern oceanic region. Comparison of the predictions and observations indicates that the CFSv2 ensembles with leads of 0–4 and 5–9 days give successful forecasts for the spatial characteristics of temperature and precipitation, with abnormal wet–dry and warm–cold centers agreeing well with the observations (Fig. 9b). However, the predictions underestimate the extent of precipitation and temperature anomalies when lead time is longer than 10 days, for both positive and negative anomalies.

Fig. 9.
Fig. 9.

Composite anomalies of precipitation (shading) and T2m (contours) for observations and CFSv2 predictions for LD 0–4 and LD 5–9.

Citation: Weather and Forecasting 32, 4; 10.1175/WAF-D-16-0209.1

We further depict the forecast skills of the CFSv2 for several key variables at different leads of time. Figures 10a–d show the temporal correlation between observations and forecasts during cold surge days for regionally averaged SLP over the SH domain (35°–55°N, 90°–115°E), horizontal winds at 850 hPa (U850 and V850) over East Asia (20°–50°N, 80°–140°E), and precipitation and T2m over eastern China (20°–55°N, 105°–135°E), northern region (35°–55°N, 105°–125°E), and southern region (20°–35°N, 105°–125°E) (see section 3), respectively. The prediction skill for SLP is significant when the lead time is less than 11 days, with the correlation coefficient above the 99% confidence level (Fig. 10a). However, the forecast skill decreases with increasing lead time. It takes about 10–14 days of lead time for U850 and V850 to fall into the range of unskillful prediction (Fig. 10b). The correlations for precipitation and T2m are lower than those for the other variables, with skill dropping below the 99% confidence level within 1 week of lead time for precipitation and less than 2 weeks for T2m. The forecast skill shows some differences between region 1 and region 2, and for precipitation higher skill is found over the northern region (significantly within lead time of 8 days) than over the southern region (5 days). For T2m, however, forecast skills are higher for the southern region (14 days) than the northern domain (11 days), which is consistent with the climatological result (see Figs. 4a–d). Prediction skills for SLP, meridional wind at 850 hPa (V850), precipitation, and T2m measured by the temporal correlation coefficient (TCC) between observations and predictions with different time leads are also analyzed. Only the results with leads of 0–14 days are shown in Fig. 11. Significant TCCs for SLP, V850, and T2m can be found over most Asian regions. Skill of precipitation is lower than that of the other variables, with the skillful region mainly in the low latitudes. Indeed, poor skills are found in northwestern China and the higher latitudes when lead time is longer than 7 days. Skill of T2m also drops quickly over northeastern China and the western Tibetan Plateau in the predictions with a lead time of 20 days (not shown).

Fig. 10.
Fig. 10.

Correlation between observations and predictions at different lead times (horizontal axis) for (a) SLP over the SH area (35°–55°N, 90°–115°E) and (b) winds averaged over East Asia (20°–50°N, 80°–140°E); (c) precipitation and (d) T2m averaged over eastern China (20°–55°N, 105°–135°E), northern region, and southern region, respectively. Shown are three-point running averages along forecast lead days. Black dashed lines denote a statistically significant correlation at the 99% confidence level.

Citation: Weather and Forecasting 32, 4; 10.1175/WAF-D-16-0209.1

Fig. 11.
Fig. 11.

Spatial distributions of TCCs between observations and CFSv2 predictions for LD 0–14 for (a) SLP, (b) meridional wind at 850 hPa, (c) precipitation, and (d) T2m. Shaded areas denote statistically a significant correlation at the 99% confidence level.

Citation: Weather and Forecasting 32, 4; 10.1175/WAF-D-16-0209.1

5. Regional features of cold surges

Cold surges of different origins and moving paths exert different impacts on East Asian weather and climate. Does the skill of predicting cold surge occurrence vary with region? Here, we divide all detected cold surges into two types according to the origin of cold surge: northern and southern regions. Based on the second criterion described in section 2, the center of maximum daily T2m drop is distinguished according to both day-to-day drop of T2m and daily T2m anomaly on cold surge days, and the location of the maximum cooling center is treated as a cold surge occurrence region. Figure 12 shows the annual variation of cold surge days over region 1 and region 2 (see section 3) in observations and CFSv2 ensemble-mean predictions of different leads of time. In total, 41 cold surge days are found over region 1 in observations during the 11 winters (Fig. 12a), which occurred most frequently in 1999 with a total of seven times. At the lead time of 0–4 days, the CFSv2 predicts similar cold surge days (46 days) as observed and captures their interannual variation. However, just 26 cold surge days are captured in the predictions at 5–9 lead days, and the cold surge frequency depicted by the CFSv2 drops obviously with increasing lead time. For region 2, 53 cold surge days are found in observations, which are slightly more than those for region 1 (Fig. 12b). Only 39 cold surge days are predicted by the CFSv2 at leads of 0–4 days, which is about 74% percent of the observed. Furthermore, the cold surge days predicted by the CFSv2 decrease to below 55% when lead time is beyond 9 days. The CFSv2 especially cannot predict the number of cold surges in 2005 accurately. Ten cold surge days were observed in that year, the most among the 11 winters. However, the predictions from 0 to 4 and 5 to 9 lead days just depict four cold surges, which is far fewer than the observed.

Fig. 12.
Fig. 12.

Number of cold surge days that occurred in (a) region 1 and (b) region 2 for observations and CFSv2 predictions for LD 0–4 and LD 5–9.

Citation: Weather and Forecasting 32, 4; 10.1175/WAF-D-16-0209.1

Figure 13 shows the composite anomalies of SLP and horizontal wind at 850 hPa for the cold surge days over the northern and southern regions in observations and ensemble predictions from the leads of 0–14 days. When cold surges occur over the northern region, a stronger-than-normal SH can be observed (Fig. 13a). It not only covers the entire Eurasian continent, but also extends eastward to the Bering Sea. Moreover, the location of the SH (near 40°–65°N) is more northward than the observed. For example, Fig. 7b indicates a combination of low pressure near Japan with the Aleutian low, but Fig. 13a shows a separation of deep low-pressure trough into two parts due to the high pressure that expands southeastward and extends between Japan and the Aleutian Islands. Although the CFSv2 at leads of 0–14 days forecasts the active range of the high pressure, the intensity of the high is weaker than that in the observations (Fig. 13g), leading to weakening of the anomalous anticyclone at 850 hPa. Thus, the dipole mode cannot be found in the predictions, and the northerly wind at 850 hPa over eastern China is distinctly weaker than the observed.

Fig. 13.
Fig. 13.

Composite anomalies of SLP (shading) and horizontal winds at 850 hPa (vectors) for the cold surge days in (top) northern and (middle) southern region (a),(b) for observations and (d),(e) for CFSv2 predictions for LD 0–14. (c),(f) Differences between northern and southern China and (g),(h) differences between prediction and observation.

Citation: Weather and Forecasting 32, 4; 10.1175/WAF-D-16-0209.1

Distinct from the abovementioned characteristics, the SH is less extensive when cold surges occur over southern China, and the location of its center shifts clearly southward to around 45°N (Fig. 13e). On the contrary, a stronger-than-normal low pressure controls northern Russia and extends from northeast to southwest. Between the strong SH and Aleutian low anomalies, the northerly wind anomaly at 850 hPa from the North Pole brings strong cold air to the south, reaching as far south as 15°N. The CFSv2 well predicts the anomalous pattern, but the SH is weaker than the observed. Furthermore, the predicted horizontal wind at 850 hPa is also obviously weaker than that in the observations.

The differences in atmospheric circulation between the two types of cold surges indicate that the observed and predicted SLP anomalies both show an apparent boundary at 45°N (Figs. 13c,f). When positive anomalies cover the whole northern region, negative anomalies are to the south. In the meantime, a strong anticyclone appears at 850 hPa over the northern region, with southwesterly wind from the South China Sea to southern China, the Korean Peninsula, and Japan. Thus, when the SH is strong and expands to the east with its center at high latitudes, cold surges usually occur in the northern region. Contrarily, when the SH expands southward along with a deepened and southwestward-extended Aleutian low, accompanied by a strong lower-level northerly wind, a cold surge is more likely to occur in the southern region.

Figure 14 shows the composite anomalies of geopotential height at 500 hPa and horizontal wind at 200 hPa for the cold surge days over the two regions in observations and the predictions at 0–14 days of lead time. Differences are clearly seen between the two types of cold surges. When cold surges occur over the northern region, anomalous geopotential height forms a positive–negative dipole pattern in the meridional direction. Large positive Z500 anomalies cover the subarctic region and Russia and extend from the Sea of Okhotsk to Japan (Fig. 14a). Significant negative anomalies are found from Mongolia to eastern China (25°–50°N, 90°–140°E). Unusually, the southward expansion of the SH is apparent from day −3 to day +3 of the cold surge days (not shown), forming a strong anticyclone–cyclone couplet. Northern China is under the control of an upper-tropospheric anticyclonic pattern.

Fig. 14.
Fig. 14.

As in Fig. 13, but for geopotential height at 500 hPa (shading) and winds at 200 hPa (vectors).

Citation: Weather and Forecasting 32, 4; 10.1175/WAF-D-16-0209.1

When cold surge occurs over the southern region, the features of the wave train in the zonal direction resemble those shown in Fig. 9c. An upper-tropospheric ridge–trough–ridge pattern of Z500 anomalies in a northwest–southeast direction can be found in Fig. 14b, with the positive anomaly center (a ridge) around southwestern Russia and western Mongolia (45°–65°N, 70°–110°E). A negative center (a trough) is seen from the Sea of Okhotsk to the Korean Peninsula and north of the Yangtze River in eastern China (30°–65°N, 110°–140°E), and another positive center appears from the Aleutian Islands to the eastern Sea of Japan (25°–50°N, 140°E–180°). The anomalous wave train coincides with the cyclonic–anticyclonic–cyclonic pattern in the lower layer (Fig. 13b), leading to strong northeasterly flow between the cyclone and the anticyclone, which controls all of eastern China (110°–120°E).

Different high-level circulation patterns between the two types of observed cold surges are shown in Fig. 14c. A positive–negative–positive–negative wave train appears from the west to the east in the high-latitude region. When cold surges occur in the southern region, the SH is stronger and is located more southward, and the trough from the Aleutian Islands to Japan is also deeper compared with the case of high-latitude cold surges. Considering both higher-level and lower-level circulation patterns, the anomalous northerly wind from the high latitudes is much stronger over eastern China, leading to cold air over southern China.

The CFSv2 ensemble predictions at the lead time of 0–14 days show similar circulation patterns as the observed (Figs. 14d–f). The model well predicts the locations of wave train centers and reveals the difference in large-scale patterns between the two types of cold surges. However, the intensity of the tropospheric trough and ridge is slightly underestimated by the model (Figs. 14g,h).

Figure 15 shows the composite anomalies of precipitation and T2m for the two types of cold surge observations and ensemble predictions. When a cold surge occurs in the northern region, above-normal precipitation is found in China’s coastal areas, Japan, and the eastern oceanic region. Maximum precipitation is located right at the anomalous cyclonic center at 850 hPa (Fig. 13a). In contrast to Fig. 9a, obviously less precipitation is found in southern China, with more precipitation over northeast China and the Sea of Japan. Besides, negative anomalies of T2m cover almost all of China, with the central position being farther to the north by about 5° of latitude than that in Fig. 9a. Figure 15b shows that, when cold surges occur in the southern region, the anomalies of precipitation and T2m are similar to the features shown in Fig. 9a. Differences between the two types of cold surges also include less precipitation in southern China and the northwest Pacific, but more precipitation over the Korean Peninsula and Japan (Fig. 15c). Decreased temperature is mainly concentrated in the region north of 30°N on the Asian continent. The ensemble predictions at 0–14 days of lead time are shown in Figs. 15d–f. In spite of the small difference in intensity, the predicted patterns of precipitation and T2m in each type of cold surge coincide with the observed. Less-than-observed precipitation over the eastern oceanic regions and a smaller-than-observed drop of T2m are found in both types of cold surges (Figs. 15g,h).

Fig. 15.
Fig. 15.

As in Fig. 13, but for precipitation (shading) and T2m (contours).

Citation: Weather and Forecasting 32, 4; 10.1175/WAF-D-16-0209.1

The above analysis shows that similar systematic biases can be found in ensemble predictions of precipitation for both winter season and cold surge days, with more-than-observed precipitation over the South China Sea, the Philippine Islands, and the northwest Pacific, but less-than-observed precipitation over India, the Indo-China Peninsula, and most high-latitude regions. During the cold surge days, the CFSv2 predicts a weaker SH, weaker northerly wind over eastern China, and a weaker cyclone around the Sea of Japan at the lower levels with increasing lead time, compared with observations. These biases of prediction may result in underestimated precipitation over southern China, the Sea of Japan, and the northwest Pacific by the CFSv2. On the other hand, the CFSv2 exhibits a better skill for T2m prediction, although biases still exist when lead time exceeds 5 days, namely, with warm bias over East Asia and cold bias over the northwest Pacific. The underestimated land–ocean temperature contrast may influence the prediction of the EAWM, which influences the precipitation over East Asia significantly. Several projects of coupled model hindcast intercomparison have shown that state-of-the-art coupled models perform reasonably in capturing the gross features of the subseasonal–seasonal variability of the Asian monsoon (Kumar et al. 2005; Wang and Ding 2008; Lee et al. 2010). Our analysis shows that the CFSv2 can also give a decent prediction of the interannual variability of EAWM (not shown). However, high skill of EAWM prediction does not necessarily lead to high skill in prediction of cold surges, even though the EAWM and cold surges are closely related.

6. Summary and discussion

In this study, the prediction capability of wintertime East Asian cold surges is investigated using the daily data of retrospective forecasts by the NCEP CFSv2. The skills of the model in capturing cold surges, their variations, and their physical links to large-scale atmospheric circulation patterns are discussed. The variations of prediction skill and bias, as well as the regional features, with respect to prediction period and lead time are also explored.

Several major features of the EAWM are well predicted by the CFSv2, which include the SH and the low-troposphere wind and T2m in East Asia. However, like other models, the CFSv2 shows low skill in predicting precipitation. Systematic biases are found in the predictions when lead time is beyond 1 week, with more-than-observed precipitation over the South China Sea, the Philippine Islands, and the northwest Pacific, but less-than-observed precipitation over the Indian Peninsula, the Indo-China Peninsula, and most high-latitude regions. The CFSv2 can well capture the frequency, intensity, and location of cold surges at leads of about 2 weeks. However, the cold surge frequency detected by the model is obviously less than the observed. The interannual features of cold surge frequency are well predicted by the model at short leads. The spatial–temporal evolution of the large-scale circulation associated with cold surges is accurately indicated in the predictions. The CFSv2 also well predicts the observed patterns of temperature and precipitation anomalies during cold surge days.

Distinguishable differences in large-scale circulation, precipitation, and T2m are found between the two types of cold surges that occur over northern and southern regions, respectively. When cold surges occur over the northern region, a dipole pattern of anomalous Z500 appears in a meridional direction. On the contrary, when cold surges occur over the southern region, a zonally oriented upper-tropospheric wave train is found, with a strong and southward-expanded SH and a strengthened and southwestward-extended Aleutian low, leading cold air from the high latitudes to southern China. The CFSv2 gives a reasonable prediction of these anomalous patterns; however, the SH and the northerly wind over eastern China are underpredicted with increasing lead time.

When cold surges occur in southern China, the atmospheric patterns are similar to those of cold surges during the positive AO phase (see Park et al. 2011a). Conversely, the patterns of cold surges in northern China are similar to those during the negative AO phase. However, the frequency anomalies in southern (northern) China during the positive (negative) AO phase are not in agreement with those in Park et al. (2010). Thus, the location of cold surge occurrence is not dependent on any single impacting factor. Previous studies have also addressed the influences of NAO, ENSO, and MJO on cold surge occurrence (Jeong and Ho 2005; Hong et al. 2008; Park et al. 2010, 2011a; Woo et al. 2012). Future research should be focused on not only the model skill of forecasting cold surges, but also the skill of predicting the main factors that affect cold surges.

Subseasonal forecasting is generally less skillful, especially for forecasting extreme events, which are remarkably influenced by both atmospheric conditions and boundary forcing. The significant bias growth with increasing lead time reflects the distinction of the initial memory capability of the climate system, and this initialization issue is a difficult problem in numerical forecast studies. We analyze the prediction bias features of cold surges and related circulations. The possible causes of these biases and the impacts of initial atmospheric and oceanic conditions in the CFSv2 need to be further explored. The ensemble-mean method is another important factor that affects dynamical prediction skills. Here we only use a simple arithmetic average method, and the cold surge events are detected from ensemble forecasts. The ensemble mean is inevitably smoothed and may lose some important signals for extreme events. In a supplementary analysis, we detect cold surge events from each single ensemble member and then compute the average values and reveal some differences from the results shown above. For example, in the additional analysis the detected numbers of cold surge events for LD 5–9 from 1999 to 2010 are 7, 10, 6, 10, 8, 7, 8, 3, 4, 6, and 11, respectively. The numbers of cold surge events and their interannual variations are closer to the observed, especially when the lead time is above 10 days (figures not shown). Thus, different ensemble methods and an ensemble mean from different sample numbers may lead to different biases, which need to be further evaluated. In addition, model resolution can affect the forecast of the frequency of cold surge events. Park et al. (2011b) found that cold events appeared more frequently in higher-resolution models. Lower-resolution models can only capture large-scale cold events. The variations of forecast skills with model resolutions should also be analyzed in future studies.

Evaluation of model performance is often based on observations or reanalysis datasets. However, differences exist between different reanalysis products, which lead to another uncertainty in climate model assessment. In addition to station data, this study also applies the NCEP–DOE R-2. The ERA-Interim (Simmons et al. 2006) is another reanalysis dataset produced by the European Centre for Medium-Range Weather Forecasts (ECMWF), and we repeated our analysis using this dataset. As shown in Fig. 16, there are discernible differences between the two sets of reanalysis data. Compared to the R-2, the climatic mean biases from the ERA-Interim are even smaller for SLP, low-level winds, and T2m (Figs. 16a,b). Although consistency is seen in detecting cold surge events and evaluating atmospheric circulation patterns (Figs. 16c,d), further evaluation is necessary for the forecasting uncertainty caused by the selection of reanalysis data.

Fig. 16.
Fig. 16.

Differences between ensemble-mean predictions for LD 0–14 and ERA-Interim for (a) SLP (shading) and 850-hPa winds (vectors) and (b) T2m. Also shown are the composite anomalies from the ERA-Interim for (c) SLP (shading) and 850-hPa winds (vectors) and (d) Z500 (shading) and 200-hPa winds (vectors) on day 0 of a cold surge occurrence.

Citation: Weather and Forecasting 32, 4; 10.1175/WAF-D-16-0209.1

At present, climate models have been an important tool for subseasonal forecasts in many agencies, in spite of the many limitations they encounter, such as the obvious uncertainty in initial conditions, quick growth of forecast bias, and limited skill in the forecasting of regional features (Abhilash et al. 2014). In addition, different models have different model biases, which lead to inaccurate, incomplete, and model-dependent estimates of signal and noise variability. Assessment of the impact of a particular process on forecast skill for cold surges is also model dependent. It is important to use a multimodel framework to reduce the uncertainty of cold surge forecasts. The World Weather Research Program and the World Climate Research Program have launched a Subseasonal to Seasonal (S2S) Prediction Project in recent years. A main deliverable of the project is the establishment of an extensive database, containing up to 60-day forecasts and reforecasts from 11 operational centers (Vitart et al. 2017). One of the key issues for the S2S Prediction Project is to address the predictability of extreme events. The project provides us a convenient opportunity for further assessment based on the multimodel forecasting of cold surges.

Acknowledgments

The authors thank the three anonymous reviewers who provided helpful comments and suggestions for improving the overall quality of the paper. This study was supported by the State Key Research Plan of China (Grant 2014CB953904), National Key R&D Program of China (Grant 2016YFA0602100), the National Natural Science Foundation of China (Grants 41375081 and 41661144019), the China Meteorological Special Project (Grant GYHY201406018), the LASW State Key Laboratory Special Fund (2016LASW-B01), and the CMA Institute of Tropical and Marine Meteorology/Guangdong Provincial Key Laboratory of Regional Numerical Weather Prediction.

REFERENCES

  • Abhilash, S., and Coauthors, 2014: Prediction and monitoring of monsoon intraseasonal oscillations over Indian monsoon region in an ensemble prediction system using CFSv2. Climate Dyn., 42, 28012815, doi:10.1007/s00382-013-2045-9.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Adler, R. F., and Coauthors, 2003: The version-2 Global Precipitation Climatology Project (GPCP) monthly precipitation analysis (1979–present). J. Hydrometeor., 4, 11471167, doi:10.1175/1525-7541(2003)004<1147:TVGPCP>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Boyle, J. S., and T. J. Chen, 1987: Synoptic aspects of the wintertime East Asian monsoon. Monsoon Meteorology, C. P. Chang and T. N. Krishnamurti, Eds., Oxford University Press, 125–160.

  • Chan, J. C. L., and C. Y. Li, 2004: The East Asian winter monsoon. East Asian Monsoon, C. P. Chang, Ed., World Scientific Series on Meteorology of East Asia, Vol. 2, World Scientific Publishing, 54–106.

    • Crossref
    • Export Citation
  • Chang, C. P., and K. M. W. Lau, 1980: Northeasterly cold surges and near-equatorial disturbances over the winter MONEX area during December 1974. Part II: Planetary-scale aspects. Mon. Wea. Rev., 108, 298312, doi:10.1175/1520-0493(1980)108<0298:NCSANE>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chang, C. P., and K. M. Lau, 1982: Short-term planetary-scale interactions over the tropics and midlatitudes during northern winter. Part I: Contrasts between active and inactive periods. Mon. Wea. Rev., 110, 933946, doi:10.1175/1520-0493(1982)110<0933:STPSIO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chen, S. J., and L. Dell’Osso, 1987: A numerical case study of a cold air outbreak over East Asia during late spring. Meteor. Atmos. Phys., 37, 237244, doi:10.1007/BF01031045.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chen, T. C., 2002: A North Pacific short-wave train during the extreme phases of ENSO. J. Climate, 15, 23592376, doi:10.1175/1520-0442(2002)015<2359:ANPSWT>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chen, T. C., W. R. Huang, and J. H. Yoon, 2004: Interannual variation of the East Asian cold surge activity. J. Climate, 17, 401413, doi:10.1175/1520-0442(2004)017<0401:IVOTEA>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ding, Y. H., 1994: Monsoon over China. Kluwer Academic Publishers, 432 pp.

  • Ding, Y. H., and T. N. Krishnamurti, 1987: Heat budget of the Siberian high and the winter monsoon. Mon. Wea. Rev., 115, 24282449, doi:10.1175/1520-0493(1987)115<2428:HBOTSH>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ek, M. B., K. E. Mitchell, Y. Lin, E. Rogers, P. Grunmann, G. Gayno, and J. D. Tarpley, 2003: Implementation of Noah land surface model advances in the National Centers for Environmental Prediction operational mesoscale Eta model. J. Geophys. Res., 108, 8851, doi:10.1029/2002JD003296.

    • Search Google Scholar
    • Export Citation
  • Gong, D. Y., and C. H. Ho, 2002: The Siberian high and climate change over middle to high latitude Asia. Theor. Appl. Climatol., 72, 19, doi:10.1007/s007040200008.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Griffies, S. M., M. J. Harrison, R. C. Pacanowski, and A. Rosati, 2004: A technical guide to MOM4. GFDL Ocean Group Tech. Rep. 5, NOAA/GFDL, 342 pp., https://www.gfdl.noaa.gov/bibliography/related_files/smg0301.pdf.

  • Hong, C.-C., H.-H. Hsu, H.-H. Chia, and C.-Y. Wu, 2008: Decadal relationship between the North Atlantic Oscillation and cold surge frequency in Taiwan. Geophys. Res. Lett., 35, L24707, doi:10.1029/2008GL034766.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Huang, W.-R., S. Y. Wang, and J. C. L. Chan, 2011: Discrepancies between global reanalysis and observations in the interdecadal variations of Southeast Asian cold surge. Int. J. Climatol., 31, 22722280, doi:10.1002/joc.2234.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jeong, J.-H., and C.-H. Ho, 2005: Changes in occurrence of cold surges over East Asia in associated with Arctic Oscillation. Geophys. Res. Lett., 32, L14704, doi:10.1029/2005GL023024.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jeong, J.-H., B.-M. Kim, C.-H. Ho, D. Chen, and G.-H. Lim, 2006: Stratospheric origin of cold surge occurrence in East Asia. Geophys. Res. Lett., 33, L14710, doi:10.1029/2006GL026607.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jeong, J.-H., B.-M. Kim, C.-H. Ho, and Y.-H. Noh, 2008: Systematic variation in wintertime precipitation in East Asia by MJO-induced extratropical vertical motion. J. Climate, 21, 788801, doi:10.1175/2007JCLI1801.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jiang, X. W., S. Yang, Y. Li, A. Kumar, X. Liu, Z. Zuo, and B. Jha, 2013a: Seasonal-to-interannual prediction of the Asian summer monsoon in the NCEP Climate Forecast System version 2. J. Climate, 26, 37083727, doi:10.1175/JCLI-D-12-00437.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jiang, X. W., S. Yang, Y. Li, A. Kumar, W. Wang, and Z. Gao, 2013b: Dynamical prediction of the East Asian winter monsoon by the NCEP Climate Forecast System. J. Geophys. Res. Atmos., 118, 13121328, doi:10.1002/jgrd.50193.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kanamitsu, M., W. Ebisuzaki, J. Woollen, S. K. Yang, J. J. Hnilo, M. Fiorino, and G. L. Potter, 2002: NCEP–DOE AMIP-II Reanalysis (R-2). Bull. Amer. Meteor. Soc., 83, 16311643, doi:10.1175/BAMS-83-11-1631.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kumar, K. K., M. Hoerling, and B. Rajagopalan, 2005: Advancing dynamical prediction of Indian monsoon rainfall. Geophys. Res. Lett., 32, L08704, doi:10.1029/2004GL021979.

    • Search Google Scholar
    • Export Citation
  • Kuriyama, K., and M. Yamamoto, 2011: Interannual and synoptic-scale features of two types of cold-air outbreaks over the East China Sea during 1988–2006. Theor. Appl. Climatol., 103, 291304, doi:10.1007/s00704-010-0303-x.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lau, K. M., and C. P. Chang, 1987: Planetary scale aspects of winter monsoon and teleconnections. Monsoon Meteorology, C. P. Chang and T. N. Krishnamurti, Eds., Oxford University Press, 161–202.

  • Lau, N. C., and K. M. Lau, 1984: The structure and energetic of midlatitude disturbances accompanying cold-air outbreaks over East Asia. Mon. Wea. Rev., 112, 13091327, doi:10.1175/1520-0493(1984)112<1309:TSAEOM>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lee, J. Y., and Coauthors, 2010: How are seasonal prediction skills related to models’ performance on mean state and annual cycle? Climate Dyn., 35, 267283, doi:10.1007/s00382-010-0857-4.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Li, Q. P., Y. H. Ding, W. J. Dong, and G. H. Yang, 2007: A numerical study on the winter monsoon and cold surge over East Asia. Adv. Atmos. Sci., 24, 664678, doi:10.1007/s00376-007-0664-y.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lu, F. C., H. M. H. Juang, and C. C. Liao, 2007: A numerical case study of the passage of a cold surge across Taiwan. Meteor. Atmos. Phys., 95, 2752, doi:10.1007/s00703-006-0192-9.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Luo, X., and B. Wang, 2017: How predictable is the winter extremely cold days over temperate East Asia? Climate Dyn., 48, 25572568, doi:10.1007/s00382-016-3222-4.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Moorthi, S., H.-L. Pan, and P. Caplan, 2001: Changes to the 2001 NCEP operational MRF/AVN global analysis/forecast system. NWS Tech. Procedures Bull. 484, 14 pp., http://www.nws.noaa.gov/om/tpb/484.pdf.

  • Nakamura, H., and T. Murakami, 1983: Orographic effects on cold surges and lee-cyclogenesis as revealed by a numerical experiment, Part I: Time mean aspect. J. Meteor. Soc. Japan, 61, 524546, doi:10.2151/jmsj1965.61.4_524.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Park, T. W., J. H. Jeong, C. H. Ho, and S. J. Kim, 2008: Characteristics of atmospheric circulation associated with cold surge occurrences in East Asia: A case study during 2005/06 winter. Adv. Atmos. Sci., 25, 791805, doi:10.1007/s00376-008-0791-0.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Park, T. W., C. H. Ho, S. Yang, and J. H. Jeong, 2010: Influences of Arctic Oscillation and Madden-Julian Oscillation on cold surges and heavy snowfalls over Korea: A case study for the winter of 2009–2010. J. Geophys. Res., 115, D23122, doi:10.1029/2010JD014794.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Park, T. W., C. H. Ho, and S. Yang, 2011a: Relationship between the Arctic Oscillation and cold surges over East Asia. J. Climate, 24, 6883, doi:10.1175/2010JCLI3529.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Park, T. W., C. H. Ho, S. J. Jeong, Y. S. Choi, S. K. Park, and C. K. Song, 2011b: Different characteristics of cold day and cold surge frequency over East Asia in a global warming situation. J. Geophys. Res., 116, D12118, doi:10.1029/2010JD015369.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Park, T. W., C. H. Ho, and Y. Deng, 2014: A synoptic and dynamical characterization of wave-train and blocking cold surge over East Asia. Climate Dyn., 43, 753770, doi:10.1007/s00382-013-1817-6.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Saha, S., and Coauthors, 2006: The NCEP Climate Forecast System. J. Climate, 19, 34833517, doi:10.1175/JCLI3812.1.

  • Saha, S., and Coauthors, 2010: The NCEP Climate Forecast System Reanalysis. Bull. Amer. Meteor. Soc., 91, 10151057, doi:10.1175/2010BAMS3001.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Saha, S., and Coauthors, 2014: The NCEP Climate Forecast System version 2. J. Climate, 27, 21852208, doi:10.1175/JCLI-D-12-00823.1.

  • Simmons, A., S. Uppala, D. Dee, and S. Kobayashi, 2006: ERA-Interim: New ECMWF reanalysis products from 1989 onwards. ECMWF Newsletter, No. 110, ECMWF, Reading, United Kingdom, 25–35, https://www.ecmwf.int/sites/default/files/elibrary/2006/14615-newsletter-no110-winter-200607.pdf.

  • Takaya, K., and H. Nakamura, 2005: Mechanism of intraseasonal amplification of the cold Siberian high. J. Atmos. Sci., 62, 44234440, doi:10.1175/JAS3629.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Vitart, F., A. W. Roberston and S2S Steering Group, 2015: Sub-seasonal to seasonal prediction: Linking weather and climate. Seamless Prediction of the Earth System: From Minutes to Months, G. Brunet, S. Jones, and P. M. Ruti, Eds., WMO-1156, World Meteorological Organization, 385–401, http://library.wmo.int/pmb_ged/wmo_1156_en.pdf.

  • Vitart, F., and Coauthors, 2017: The Subseasonal to Seasonal (S2S) Prediction project database. Bull. Amer. Meteor. Soc., 98, 163173, doi:10.1175/BAMS-D-16-0017.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wang, B., and Q. Ding, 2008: Global monsoon: Dominant mode of annual variation in the tropics. Dyn. Atmos. Oceans, 44, 165183, doi:10.1016/j.dynatmoce.2007.05.002.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Woo, S. H., B. M. Kim, J. H. Jeong, S. J. Kim, and G. H. Lim, 2012: Decadal changes in surface air temperature variability and cold surge characteristics over northeast Asia and their relation with the Arctic Oscillation for the past three decades (1979–2011). J. Geophys. Res., 117, D18117, doi:10.1029/2011JD016929.

    • Search Google Scholar
    • Export Citation
  • Wu, B., and J. Wang, 2002: Winter Arctic oscillation, Siberian high and East Asian winter monsoon. Geophys. Res. Lett., 29, 1897, doi:10.1029/2002GL015373.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wu, M. C., and J. C. L. Chan, 1997: Upper-level features associated with winter monsoon surges over South China. Mon. Wea. Rev., 125, 317340, doi:10.1175/1520-0493(1997)125<0317:ULFAWW>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhang, Y., and W. C. Wang, 1997: Model-simulated northern winter cyclone and anticyclone activity under a greenhouse warming scenario. J. Climate, 10, 16161634, doi:10.1175/1520-0442(1997)010<1616:MSNWCA>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhu, J., and J. Shukla, 2013: The role of air–sea coupling in seasonal prediction of Asia–Pacific summer monsoon rainfall. J. Climate, 26, 56895697, doi:10.1175/JCLI-D-13-00190.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhu, J., B. Huang, Z. Z. Hu, J. L. Kinter III, and L. Marx, 2013: Predicting U.S. summer precipitation using NCEP Climate Forecast System version 2 initialized by multiple ocean analyses. Climate Dyn., 41, 19411954, doi:10.1007/s00382-013-1785-x.

    • Crossref
    • Search Google Scholar
    • Export Citation
Save