Processes of intraseasonal snow cover variations over the eastern China during boreal winter

This study reveals that the dominant time scale of intraseasonal snow cover variation over the eastern China is within 30 days by using the latest satellite snow cover data from the moderate resolution imaging spectroradiometer (MODIS)/Terra product. The leading empirical orthogonal function (EOF) mode of 10–30‐day snow cover variation during boreal winter from 2004 to 2018 over the eastern China has two centers: northwest part of the eastern China and north of the Yangtze River. Composite analysis based on 25 snow events identified from normalized leading principal time series (PC1) indicates that the southeastward intrusion of surface anticyclonic anomalies and accompanying low temperature anomalies provide the temperature condition for snow events. Negative Arctic Oscillation induces mid‐latitude wave train and leads to the development of surface anticyclonic anomalies and upper‐level cyclonic anomalies over East Asia. The cyclonic anomalies induce ascending motion and anomalous convergence of water vapor fluxes over the eastern China, which supplies moisture for snowfall.


| INTRODUCTION
In January 2008, a strong ice storm with snowfall hit the eastern China (Zhou et al. 2011), one of the most populated regions in East Asia. Unlike northeastern China and the Tibetan Plateau, the eastern China (20-40 N, 105-120 E) is not usually covered by snow during boreal winter (Wen et al. 2009). So, the snowfall over this region caused tremendous economic losses (Wen et al. 2009;Zhou et al. 2009). The study by Sun et al. (2010) and Zhou et al. (2018) suggested that the snowfall events over the eastern China tend to be isolated from other regions of China.
The occurrence of snowfall events over the eastern China on synoptic scale has been discussed by previous studies (Wen et al. 2009;Zhou et al. 2009;Hui 2009). Previous studies have also documented interannual to interdecadal snow variation over the eastern China (Sun et al. 2010;Zhou et al. 2017Zhou et al. , 2018. There are relatively few studies on snow variations on intraseasonal time scales. The investigation of the intraseasonal variation of snow and the factors and processes can help to improve the predictive skill of snowfall. The occurrence of snowfall requires both low temperature and large amount of moisture (Sun and Wang 2013;Zhou et al. 2018). Previous studies have indicated that the occurrence of snowfall events over the eastern China is related to the strong western Pacific subtropical high and the Indian-Burma trough that help to transfer moisture toward the eastern China (Zhou et al. 2009;Li et al. 2017), the variation of the subtropical jet stream (Wen et al. 2009), the mid-high latitude circulation systems, such as the Siberian high and the polar vortex that bring cold anomalies to the eastern China (Hui 2009;Wu 2017, 2018a). What are the factors and processes of intraseasonal variations of snow over the eastern China?
The present study addressed the following questions: (a) What is the dominant time scale of intraseasonal snow variation over the eastern China during boreal winter? (b) What are the factors and physical processes of intraseasonal snow variations over the eastern China? The rest of this study is organized as follows. Data and methodology are described in Section 2. Main results are presented in Section 3. A summary is provided in Section 4.

| DATA AND METHODOLOGY
This study uses the latest satellite daily snow cover data from the moderate resolution imaging spectroradiometer (MODIS)/Terra Snow Cover Daily L3 Global 0.05Deg Climate Modeling Grid (CMG), Version 6 (Hall and Riggs 2016). This data provide consistent and objective snow estimates through high-resolution visible light observation (Frei et al. 2012;Chen et al. 2016). The overall accuracy of the MODIS snow cover data is about 93% (Hall and Riggs 2007). This dataset is on the MODIS CMG, which consists of 7,200 columns by 3,600 rows with a horizontal resolution of 0.05 × 0.05 . The snow cover is represented by percentages of snow-covered land in 0.05 resolution CMG cells. The details for the above snow data can be found on the National Snow and Ice Data Center website (https://nsidc.org/data/ modis/data_summaries#snow). We convert the original data on 0.05 × 0.05 grid to 0.5 × 0.5 grid. A 3-day running mean is further applied to 0.5 × 0.5 grid data to reduce the missing data. The snow cover data is only available south of 60 N during winter polar night. We focus on 15 winters (December-January-February) from 2004 to 2018.
The daily output from the National Centers for Environmental Prediction (NCEP)-Department of Energy Reanalysis 2 provided by the National Oceanic and Atmospheric Administration, Oceanic and Atmospheric Research, Earth System Research Laboratory (NOAA/OAR/ESRL) Physical Science Division (Kanamitsu et al. 2002) is used in this study. The variables used include geopotential height, sea level pressure, meridional and zonal winds, vertical p-velocity, air temperature, surface air temperature and wind. The pressure level data at 17 layers from 1,000 to 10-hPa have a horizontal resolution of 2.5 × 2.5 . Surface data are available on the T62 Gaussian grid.
The daily North Atlantic Oscillation (NAO) index is obtained from the NOAA Climate Prediction Center (CPC) website (http://www.cpc.ncep.noaa.gov/products/precip/ CWlink/pna/nao.shtml). The daily Arctic Oscillation (AO) index is obtained from the NOAA (CPC) website (http://www.cpc.ncep.noaa.gov/products/precip/CWlink/ daily_ao_index/ao.shtml). The intraseasonal variation is extracted using the Butterworth band-pass filter. Regression and composite analysis are employed to obtain anomalies of different quantities. The statistical significance level is estimated based on the Student's t test. The propagation of tropospheric Rossby wave train is depicted by the Rossby wave activity fluxes (Takaya and Nakamura 2001).

| RESULTS
Snow cover over the eastern China displays obvious intraseasonal variations during boreal winter. Figure 1a presents the power spectrum of 15-year average area-mean (20-40 N, 105-120 E) snow cover during winter (December to February, DJF). Two spectrum peaks can be found at the time periods of 14 and 8 days. Therefore, the dominant frequency of the intraseasonal snow cover variation over the eastern China is within 30 days. The empirical orthogonal function (EOF) analysis is applied to the 10-30-day intraseasonal snow cover variation in DJF over the eastern China (20-40 N, 105-120 E) that is extracted using the 10-30-day band-pass filter. The corresponding distribution of snow cover anomalies is displayed in Figure 1b that is the regression map with respect to the leading principal time series (PC1). The leading mode accounts for 38.5% of the total variance, which is separated from the other modes based on the criterion of North et al. (1982). Two positive snow cover centers are observed, one over northwest part of the eastern China and the other located north of the Yangtze River. It is noted that the intraseasonal snow cover variations in the above regions are not linked to those over the Tibetan Plateau and Northeast China, in agreement with previous studies (Sun et al. 2010;Zhou et al. 2018).
Next, we perform a composite analysis based on the leading EOF mode to reveal the characteristics of 10-30-day snow cover variations over the eastern China. The normalized PC1 associated with the leading EOF is defined as an index for intraseasonal snow cover variation over the eastern China. If this index exceeds one positive SD for three consecutive days, it is defined as an intraseasonal snow event, and the peak day is marked as day 0 of the snow event over the eastern China. We identified 25 snow events in the 15 winters from 2004 to 2018, which is listed in Table 1. Composite anomalies are obtained for these 25 snow events. All the variables have been filtered by a 10-30-day bandpass filter before composite analysis.
Snow cover anomalies manifest evident spatial-temporal evolution during intraseasonal snow events over the eastern Snowfall over the eastern China needs both low temperature and high moisture (Sun and Wang 2013;Zhou et al. 2018). Here, we elaborate atmospheric circulation and surface temperature anomalies and water vapor transportation during snow events over the eastern China. Snow cover anomalies are accompanied by prominent surface temperature anomalies. On day −4, evident cold anomalies appear over western Mongolia, and warm anomalies are located over southern coast of China ( Figure 3a). The cold anomalies intrude the eastern China in the following days (Figure 3b), which are accompanied by positive snow cover tendency (Figure 2f). On day 0, the cold anomalies and snow cover anomalies over the eastern China reach the peak (Figures 2c and 3c) when the snow cover tendency is weak (Figure 2g). The cold anomalies are weakened over the eastern China on day 2 (Figure 3d), which correspond to the decrease of snow cover anomalies (Figure 2d,h). The temporal evolution of cold anomalies over the eastern China during snow events indicates a close relationship between the snowfall and low temperature anomalies.
The temperature anomalies are related to surface wind anomalies. On day −4, surface cyclonic anomalies are observed around the Lake Baikal, accompanied by surface southerly wind anomalies over northern China (Figure 3e). In the following days, the anticyclonic anomalies move southeastward with northerly wind anomalies dominating the eastern China (Figure 3f), which explain the southward  (h) day 2 −3 − 2.5 −2 − 1.5 −1 − 0.5 0 0.5 1 1.5 2 2.5 3 F I G U R E 2 Composite snow cover anomalies (%) on day −4 (a), day −2 (b), day 0 (c), day 2 (d), and snow cover tendency (%/day) on day −4 (e), day −2 (f), day 0 (g), day 2 (h) of snow events. The black dots denote anomalies significant at the 95% confidence level The surface anticyclonic anomalies causing cold anomalies over the eastern China are associated with mid-latitude uppertropospheric wave trains ( (d)) on day −6 (a), day −4 (b), day −2(c), day 0 (d) of snow events. Composite water vapor flux integral from 1,000 to 100-hPa (vector, scale at right bottom of subplot (h)) and its divergence (shading, *10 −5 kg/m 2 *s, scale in the lower bar) on day −6 (e), day −4 (f), day −2 (g), day 0 (h) of snow events pattern over Eurasia toward the eastern China (Figure 4a). Another branch of wave activity fluxes emits from anticyclonic anomalies over the Mediterranean Sea and travels eastward along the subtropical jet (Hoskins and Ambrizzi 1993). The two braches merge on day −4 (Figure 4b) and contribute to the development of cyclonic anomalies over Central Asia and anticyclonic anomalies over East Asia. On day −2, the subtropical wave fluxes weaken over East Asia and another branch of wave activity fluxes appear over the Mediterranean Sea ( Figure 4c). Meanwhile, the mid-latitude wave train moves eastward. On day 0, both the mid-latitude and subtropical wave activity fluxes contribute to the enhancement of cyclonic anomalies over East Asia (Figure 4d), which is followed by surface anticyclonic anomalies and the southward intrusion of cold air (Figure 3c,g) (Song and Wu 2017). It should be noted that the subtropical wave train is weaker than the mid-latitude wave train, which suggests a dominant role of the mid-highlatitude atmospheric circulation in the development of intraseasonal snow cover events over the eastern China. The moisture for snowfall over the eastern China is related to the northward water vapor transportation. Northward transportation of water vapor appears over the southwestern flank of the anticyclonic anomalies over the eastern China on day −6 (Figure 4a,e). This leads to convergence of water vapor fluxes over southeastern China. On day −4, the above water vapor fluxes increase ( Figure 4f). Correspondingly, water vapor convergence becomes larger. The water vapor fluxes turn westward over the eastern China and bring water vapor from the East China Sea, which leads to the positive snow cover tendency over the eastern China (Figure 2e). The easterly wind anomalies on the southern flank of the anticyclonic anomalies over East Asia (Figure 4b) contribute to the westward water vapor fluxes. Divergence of water vapor fluxes develops over the eastern China after day −4 (Figure 4g,h), which corresponds to the southward intrusion of surface anticyclonic anomalies (Figure 3f,g) that bring cold and dry air from the high-latitudes (Figure 3b,c). The northward water vapor fluxes are weakened on day −2 (Figure 4g). Divergence of water vapor fluxes dominates the southeast coast of China on day 0, which corresponds to surface anticyclonic anomalies and weakened snow cover tendency (Figures 2g and 3g). F I G U R E 5 (a) Time evolution of composite NAO index (pink curve), AO index (blue curve), regional mean surface air temperature anomalies ( C) (black curve) and snow cover anomalies (%) (red curve) in the region of 20-40 N, 105-120 E. (b) Time evolution of composite anomalies of regional mean snow cover tendency (%/day) (black curve), vertical velocity (Pa/s) (blue curve), and divergence of water vapor flux integral from 1,000 to 100-hPa (*10 −6 kg/(m 2 *s)) (pink curve) in the region of 20-40 N, 105-120 E. Dots on the curves indicate anomalies significant at the 95% confidence level The temporal evolutions of regional mean surface temperature and snow cover anomalies over the eastern [105][106][107][108][109][110][111][112][113][114][115][116][117][118][119][120] are further illustrated in Figure 5a. Snow cover varies nearly out of phase with surface temperature during the snow events. Both the snow cover and surface temperature anomalies reach the peak on day 0. This indicates that the cold anomalies provide condition for snow accumulation. The NAO index is positive on day −7 and −6. This suggests that the NAO triggers the subtropical wave train, consistent with previous study (Watanabe 2004). The AO index is negative from day −6 to day 1 and it reaches the peak on day −3, during which most part of the polar region is covered by anomalous anticyclone and cyclonic anomalies situate over the North Atlantic and North Pacific at lower level, corresponding to negative AO (figure not shown). The negative AO appears to contribute to the development of the mid-latitude Rossby wave train from day −6 to day 0 (Figure 4a-d) (Park et al. 2011;Song and Wu 2018b). In comparison, the AO signal is much stronger than the NAO, and the subtropical wave train is not as evident as the mid-latitude wave train (Figure 4a-d). Therefore, the development of snow events over the eastern China is mainly related to negative AO.
The temporal relationship between the snow cover variation and the atmospheric condition for snowfall are further demonstrated in Figure 5b. The snow cover tendency (calculated using central differentiation) turns to positive after day −7, indicating the snow accumulation over the eastern China (Figure 2a-c and e-g). The convergence of anomalous water vapor fluxes and anomalous ascending motion at 700-hPa are evident after day −8. They reach the peak on day −4. The time-lag relationship indicates that snowfall is preceded by the accumulation of water vapor in the atmosphere. This validates that the water vapor transportation provides the moisture condition for snow events over the eastern China. The snow accumulation from day −4 to day −2 is accompanied by the decrease of anomalous water vapor flux convergence and the weakening of anomalous ascending motion. From day −2 to day 0, the snow continues to accumulate. At the same time, anomalous descending motion develops accompanying anomalous water vapor flux divergence. After day 0, snow melting begins, accompanying strong anomalous descending motion and water vapor divergence.
The snow cover tendency lags the moisture convergence and ascending motion. This is related to time lags between the moisture convergence and snowfall and between snowfall and snow accumulation on ground. The moisture convergence provides the moisture condition, and meanwhile, the ascending motion provides the dynamical condition for cloud/snow formation. The snowfall may lag the cloud formation by some time as it also depends upon the temperature condition. Whether the snow can accumulate on the ground when it falls also depends upon the surface temperature condition. As such, there is a time lag between the snow cover tendency and the moisture convergence and ascending motion.

| SUMMARY
By using the latest satellite daily snow cover date from the MODIS/Terra product, the present study documents the intraseasonal snow cover variations over the eastern China during boreal winter and the associated factors and the physical processes. It is found that the dominant time scale of intraseasonal snow cover variation over the eastern China is within 30 days. The present study focused on the 10-30-day intraseasonal variation extracted using a band-pass filter. The leading EOF mode of the 10-30-day snow cover variation over the eastern China has large loading over the northwest part of the eastern China and the north of the Yangtze River. Intraseasonal snow events over the eastern China are identified based on the normalized PC1. A composite analysis is conducted for 25 snow events identified in 15 winters (December to February) from 2004 to 2018.
The temporal evolution of composite anomalies reveals the following processes involved in intraseasonal snow cover variations over the eastern China. Negative AO contributes to the enhancement of a mid-latitude wave train, which is accompanied by a weak subtropical wave train. The two wave trains lead to the development and eastward move of cyclonic anomalies over Asia. This induces convergence of anomalous water vapor fluxes and anomalous ascending motion over the eastern China, which provides the moisture condition for snowfall. Surface anticyclonic anomalies and accompanying surface northerly wind anomalies bring cold air toward the eastern China, which provides the temperature condition for snow accumulation. The snowfall leads to the switch of snow cover anomalies over the eastern China from negative anomalies to positive anomalies in about 2 days. Following the development of anomalous descending motion and water vapor divergence, snow starts to melt, leading to the decrease in snow cover.