In January 2008, a series of snowstorms hit central and southern China. Persistent snow, freezing rain, and cold temperature affected the daily lives of millions of people who suffered not only from damage to agriculture, but also from disruptions in transport, energy supply, and power transmission (WMO, 2009). As shown in Figure 1(a) and (b), excessive precipitation (>60 mm month−1) was observed in the Yangtze River Basin and southern China in January 2008, which was accompanied by pronounced cold anomalies (<− 2 °C) extending from Central China to much of southern China as compared to the climatology of 1951–2007. Wen et al. (2009) also showed that the number of days that snow cover was observed in Central and southern China in January 2008 far exceeded the climatology of the past three decades.
Understanding forcings and mechanisms responsible for such extreme weather events are important for long-range weather and climate forecasting. As summarized in the WMO press release (http://www.wmo.int/pages/mediacentre/press_releases/pr_CMAWMO_en.html), the severe January 2008 snowstorm in China has been largely attributed to the moderate-to-strong La Nina event in the winter of 2007/2008 (started in the third quarter of 2007 and peaked in February 2008). Fu et al. (2008) also shows that the strong La Nina can explain the colder climate and more precipitation than usual in South China, as well as the associated circulation anomalies such as Mongolia cold high. Here, we construct the composite pattern of anomalous precipitation and surface air temperature in January for ten La Nina cases (based on the Nino3 index, http://www.esrl.noaa.gov/psd/data/climateindices) during the period 1951–2007 using China station precipitation and surface air temperature. As shown in Figure 1(c) and (d), associated with the typical La Nina cases, there is a deficit of precipitation in southern China and warm anomalies in Central and southern China. This is significantly different from what was observed in January 2008. Thus, the La Nina tends to produce opposite anomalies, and is not a necessary factor for the severe January 2008 snowstorm in China. A few recent studies have analyzed favorable background and anomalous conditions in the atmosphere that are associated with this extreme weather event, and have pointed out anomalies in high latitudes might have a substantial influence on the severe January 2008 snowstorm, i.e. the persistent Ural blocking high due to continuous negative vorticity advected from high latitudes, the strengthening of stratospheric polar vortex, and its downward propagation (Gu et al., 2008; Gao, 2009; Wang et al., 2009).
Significant decline of the Arctic sea ice has been documented in recent years (Liu et al., 2004; Comiso et al., 2008). The extreme sea ice cover anomaly observed in the summer and fall of 2007 represents a dramatic departure from historical trend line (Stroeve et al., 2008). The 2007 minimum ice extent (the cumulative area of grid boxes covering at least 15% sea ice concentration) was 4.13 million km2, 36% below the 1979–2006 average (Figure 2(a)). The loss of sea ice in summer and fall were particularly pronounced in the East Siberian, northern Chukchi, western Beaufort, Barents, and Kara Seas (Figure 2(b)). Some AGCM studies have suggested that changes of the Arctic sea ice strongly impact the overlying atmosphere, causing feedback to atmospheric circulation well beyond the Arctic's boundary, i.e. winter sea ice variability can influence the atmospheric circulation through forming stationary Rossby wave trains in the North Pacific (Honda et al., 1999), and through modulating the storm track in the North Atlantic (Alexander et al., 2004). Deser et al. (2004) and Singarayer et al. (2006) indicated that anomalies in sea ice extent might be more efficient than anomalies in sea surface temperature at exciting atmospheric response.
The question is whether the extreme sea ice cover anomaly observed in the fall of 2007 helps to drive the severe January 2008 snowstorm in China, and if so, how this contributes to the severe weather event. Speculation surrounding this based on the idea: when high-albedo sea ice is replaced by low-albedo open water, there is a significant solar heat input directly into the ocean, increasing the heat stored in the upper ocean. As demonstrated by Figure 4 of Perovich et al. (2008), the excess solar heat is sufficient to warm the upper 5 m of the ocean by 5 °C. Warming of the upper ocean retards the recovery of sea ice during the fall freeze-up period. As shown in Figure 2(a), the ice extent in October and November 2007 is also remarkably below the climatology of 1979–2000. This would extend the impact of heating into the following winter through exchanges of heat fluxes, which in turn, might have a substantial influence on atmosphere circulation. In this study, we investigate the potential contribution of extreme sea ice cover anomaly (slow recovery of sea ice) during the fall of 2007 to the severe January 2008 snowstorm in China using an atmospheric general circulation model (AGCM).
2. Data, model, and experiment design
The datasets used in this study include (1) precipitation and surface air temperature obtained from 160 stations in China, which is archived at the National Climate Center, the China Meteorological Administration (http://ncc.cma.gov.cn/cn/member.php?Login = 1), (2) Arctic sea ice concentrations retrieved from the Scanning Multichannel Microwave Radiometer and the Special Sensor Microwave/Imager based on a team algorithm (Cavalieri et al., 2003), and (3) the 250, 500, and 700 mb geopotential height and winds from the National Centers for Environmental Prediction (NCEP) reanalysis (Kalnay et al., 1996).
The ECHAM5.4 AGCM, developed by the Max-Planck Institute for Meteorology, is the baseline choice of model for this study. Some studies have suggested that the ECHAM performed best in simulating high-latitude winter climatology and variability, i.e. the model captures the observed dominant mode of atmospheric circulation and essential characteristics of the vertical wave propagation from the troposphere to the stratosphere in the Northern hemisphere (Petoukhov and Semenov, 2010; Stefan et al, 2010). Also, the ECHAM has been used in a number of climate-related studies for East Asia (Cherchi and Navarra, 2007). The ECHAM5 has evolved from the spectral operational weather forecast model used at the European Center for Medium-Range Weather Forecasts, incorporating physical parameterizations and revised numerical methods appropriate for climate simulations (Roeckner et al., 2003). The horizontal resolution employed here is T106 (approximately 1.12°), and 19 vertical layers with the top level at 100 mb. The sea ice cover and sea surface temperature are prescribed using the AMIP II boundary conditions (Hurrell et al., 2008).
Two numerical experiments were conducted: (1) a baseline simulation in which the model is forced with observed (time varying) sea ice cover and sea surface temperature from January 1989 to January 2008 (hereafter referred to as SI-RT) and (2) a sensitivity simulation, in which sea ice cover in the Arctic during the fall freeze-up period (September–October–November) of 2007 is replaced with the climatological mean of 1979–2000 (hereafter referred to as SI-FC). For each experiment, the model is integrated from 1 January 2007 to 31 January 2008, and the mode integration is repeated five times with different initial atmospheric conditions to reduce intrinsic uncertainties.
The simulated precipitation in January 2008 for SI-RT is shown in Figure 1(e). Compared to the observed precipitation, there is a close match in the overall pattern, although the simulated maximum precipitation is more toward southeastern China. Moreover, the spatial distribution of simulated and observed precipitation anomalies in January 2008 (with respect to the model's and observed climatology for the period 1989–2007) resembles each other (not shown). The consistency between the model simulations and observations gives us more confidence about the ability of the ECHAM5 to simulate the essential characteristics associated with the severe January 2008 snowstorm in China.
Figure 1(f) shows the simulated precipitation in January 2008 for SI-FC. Interestingly, compared to SI-RT (Figure 1(e)), the excessive precipitation in central and southern China is gone in SI-FC, which is retreated to the narrow coastal region of southeastern China (<30 mm month−1). The differences of the simulated precipitation between SI-RT and SI-FC are characterized by increased precipitation in Central and southern China (Figure 1(h)). This is generally consistent with the observational precipitation anomaly in much of Central and southern China (Figure 1(g)), suggesting that sea ice anomalies in the fall of 2007 do play a critical role in reproducing the observed precipitation anomalies in January 2008.
Figure 3 shows the differences of the simulated atmospheric circulation in January 2008 between SI-RT and SI-FC. Induced by the record low Arctic sea ice in September 2007 and slow recovery of Arctic sea ice in October and November 2007, in the Middle East, there is a weakening of 250 mb westerly wind in the Middle East between 35° and 60°N and a strengthening of 250 mb westerly wind to the south at 25–40°N (Figure 3(a)). This is generally in agreement with the observed 250 mb zonal wind anomalies, although the magnitude of the anomalies is weaker, and the location of the anomalies is more toward west relative to the observations (Figure 3(b)). Following Yang et al. (2004), we calculate the Middle East Jet Stream (MEJS, defined as the averaged zonal wind difference within 20–30°N, 40–70°E, and 30–40°N, 15–45°E) index at the level of 250 mb in January 2008 for SI-RT minus ST-FC, which is 1.13 m s−1. Thus, the extreme Arctic sea ice anomalies in the fall of 2007 tend to result in a strengthening and southward shift of the MEJS. Associated with the change of the MEJS, there is a large-scale decrease of 500 mb geopotential height and a broad anomalous cyclonic circulation at 700 mb in Central Siberia and subtropical Asia (Figure 3(c)), which deepens the trough over India and southern Tibetan Plateau. This anomalous atmospheric circulation pattern favors advection of warm and moist air from the Bay of Bengal to southern China, and frequent southward intrusions of cold air from far eastern Europe and central Siberia to central Asia. This provides favorable conditions for increasing precipitation in central and southern China.
Meanwhile, there is a weakening of 250 mb westerly wind in eastern Asia between 20° and 40°N, which weakens the East Asian Jet Stream (EAJS, Figure 3(a)). This is generally consistent with the observed 250 mb zonal wind anomalies, although the location of the anomalies is more toward north relative to the observations (Figure 3(b)). Also, the EAJS index (defined as the averaged zonal wind within 30–35°N, 130–160°E) for SI-RT minus ST-FC in January 2008 is − 9.9 m s−1. As the EAJS weakens, the east Asian trough, which is climatologically anchored along the coast of east Asia and extending from northeast China, through Japan to the Sea of Okhotsk (not shown), becomes shallower, as shown by an increase in 500 mb geopotential height and a broad zonally oriented anomalous anti-cyclonic circulation over much of the North Pacific (Figure 3(c)). The anomalous easterlies in the southern flank of the anomalous anticyclone impinge on eastern China across Japan and Korea, which advect warm and moist air from the mid-latitude of the north Pacific to eastern China. This also provides favorable conditions for increasing precipitation in central and southern China.
4. Discussion and conclusion
Our model experiments clearly demonstrate that the extreme Arctic sea ice anomaly during the fall of 2007 leads to conditions conducive to the severe January 2008 snowstorm in China, since the differences of precipitation and atmospheric circulation between SI-RT and SI-FC reproduce major favorable conditions responsible for the extreme weather event. The record minimum of the Arctic sea ice in September 2007, and follow-up slow recovery of the Arctic sea ice in October and November 2007 due to increased heat storage in the upper ocean layer tends to excite two stationary wave trains through local thermal response. As shown in Figure 4(a), differences of the net heat flux resulting from the reduced sea ice are 20–50 W/m2 from the ocean to the atmosphere. The latent and sensible heat fluxes are the dominant form of heating contributing ∼10–40 W/m2, followed by longwave at 5–15 W/m2.
As shown in Figure 4(b), the first wave train propagates southeastward from the Barents/Kara Seas to central Asia, forming anti-cyclonic anomalies over eastern Europe, and cyclonic anomalies extending from central Siberia and subtropical Asia. Associated with this wave train, the MEJS is intensified, which results in wet and cold conditions in central and southern China by increasing both stationary wave activity flux and higher frequency eddy activities (Yang et al., 2004). The second wave train propagates southward from central Arctic Ocean/eastern Siberia Sea to the mid-latitude and subtropics of the north Pacific. Associated with this wave train, the EAJS weakens, which is accompanied with wet but warm conditions in eastern China. The anomalous westerly associated with the strengthening of the MEJS and the anomalous easterlies associated with the weakening of the EAJS form a strong convergence zone over central and southern China, which favors increased precipitation there.
In summary, our model simulations provide evidence that the dramatic reduction of sea ice in September 2007 and follow-up low recovery of the Arctic sea ice in October and November 2007 produced an atmospheric response that contributed to the severe January 2008 snowstorm in China. The results support an important hypothesis that has been studied, namely that lower late summer Arctic sea ice cover can induce anomalous heat fluxes over the Arctic Ocean and impact weather and climate at lower latitudes the following season (Jennifer et al.2009). As shown in Figure 5, we further give the composite differences of precipitation anomalies in the following January between the standardized autumn Arctic sea ice extent anomalies larger than one positive standard deviation and smaller than one negative standard deviation from 1989 to 2008. First, the composite pattern for both the observation and simulation is similar. In recent years, corresponding to reduced Arctic sea ice, there is a tendency of increased precipitation in southern China. Second, we notice that the increased precipitation in the composite analysis is mainly confined to southern China, which is not as broad as the observed anomalies in January 2008. This suggests that the coverage of the Arctic sea ice might be reduced to a particularly low threshold, i.e. 2007, to trigger responses observed in January 2008. Relationship between the extreme Arctic sea ice anomaly and climate anomalies during the following months merits further investigation, given that the Arctic sea ice has been continuing on its downward trajectory, which will likely continue as climate warms.
This research is supported by the ‘Hundred Talent Program’ of the Chinese Academy of Sciences, National Natural Science Foundation of China (40930848, 40876099), National Key Technology R&D Program (2008AA121 704, 2011BAC03B02), 2011 China-Canada Polar Cooperation Project of State Oceanic Administration (QT06-13), Physical Oceanography Research of Chinese Forth Arctic Expedition, National Basic Research Program of China (2011CB309704, 2010CB950301) and FIO Special Fund Basic Research and Operating Expenses (2008T27, 2010T01).