C1041, doi: 10.3354/cr01041, In press
1
2
3
4
5
6
7
8
Rui Mao 1, 2, * , Dao-Yi Gong 1 , Jing Yang 1 , Jing-Dong Bao 2
1 State Key Laboratory of Earth Surface Processes and Resource Ecology, Beijing
9
10
11
12
13
14
Normal University, Beijing 100875, China
2 Department of Physics, Beijing Normal University, Beijing 100875, China
July 2011
Climate Research (revised) 15
16
17 *Corresponding author: Rui Mao, State Key Laboratory of Earth Surface Processes
18 and Resource Ecology, Beijing Normal University, Beijing 100875, China. e-mail:
19 mr@bnu.edu.cn
20
1

C1041, doi: 10.3354/cr01041, In press
21 ABSTRACT: In this study, we analyzed the relationship between Arctic Oscillation
22 (AO) and extreme precipitation events over China during boreal winters from 1954 to
23 2009. The extreme precipitation events are defined as those with daily precipitation
24 amount above the 80 th (or 90 th ) percentiles. The AO has a significantly positive
25 correlation with the frequency of extreme precipitation events over China during
26 January to February. Of all 287 stations in China, 238 stations have positive
27 correlation coefficients and 82 stations are positively significant above the 95%
28 confidence level. These stations with significantly positive correlation are mainly
29 located over the central-southern China. The correlation between the AO and the
30 frequency of extreme precipitation events averaging over the central-southern China
31 is 0.49, which is significant at the 99% confidence level. This relationship is still
32 significant in value even the El Niño/Southern Oscillation signal is excluded from the
33 original time series. In association with the AO-precipitation extremes linkage, the
34 Middle Asia jet stream (MEJS) and the southern branch trough (SBT) over the Bay of
35 the Bengal co-change consistently. A positive AO phase is accompanied by a
36 stronger-than-normal MEJS and a deepened SBT. The deepened SBT consequently
37 enhance weather disturbances in vertical motions in the low to middle troposphere
38 over the central-southern China. More moisture transport by the deepened SBT and
39 the active weather disturbances in vertical motions over the central-southern China
40
41 would yield more extreme precipitation events there.
KEY WORDS: Arctic Oscillation; extreme precipitation event; south branch trough;
42 Middle East jet stream; central-southern China
2

C1041, doi: 10.3354/cr01041, In press
43 1 INTRODUCTION
44 Arctic Oscillation (AO, also known as the northern annular mode) is the primary
45 mode of the internal dynamics in atmospheres over the extratropical northern
46 hemisphere with an equivalent barotropic structure from the surface to the lower
47 stratosphere. It becomes most active during cold seasons (November–April)
48 (Thompson & Wallace 1998). Fluctuations in the AO create a seesaw pattern in which
49 atmospheric pressure and mass around north polar regions and mid-latitudes change in
50 an out-of-phase way. For instance, a positive AO phase is accompanied by a low
51 pressure over the north polar regions and a high pressure at the mid-latitudes, and this
52 feature is reversed in a negative AO phase.
53 Many studies have shown that the AO has evident impacts on global climate.
54 During the winter season, the AO signal is reflected in fields like surface temperature
55 in the Northern Hemisphere (Thompson and Wallace 2001), precipitation in China
56 (Gong & Wang 2003), sea-ice over north polar and sub-polar regions (Wang & Ikeda
57 2000), and lower tropospheric circulation including East Asian winter monsoon,
58 Aleutian Low, and Siberian High (Gong et al. 2001, Wu and Wang 2002). During the
59 spring season, the AO has a good relationship with dust storm frequency in Northeast
60 Asia (Gong et al. 2006, Mao et al. 2011a, Mao et al. 2011b), and even has a lag effect
61 on the following summer monsoon rainfall in East Asia (Gong & Ho 2003, Gong et
62 al. 2011). The major of the previous studies have emphasized the AO’s effect on
63 seasonal mean climate. Few studies analyzed the possible linkage between AO and
64 extreme weather/climate events, including the cold air activity and blocking activity
3

C1041, doi: 10.3354/cr01041, In press
65 throughout the hemisphere (Thompson & Wallace 2001), synoptic temperature
66 variance over East Asia (Gong et al. 2004), extreme temperature in the Northeastern
67 United States and Canada (Wettstein & Mearns 2002), and weather disturbances in
68 East Asia (Gong et al. 2006).
69 Precipitation and its variability is an important issue in East Asia. Previous studies
70 revealed that AO is linked to monthly-seasonal precipitation over the
71 central-southwestern China during winter with a statistically positive correlation
72 (Gong & Wang 2003). However, few studies addressed the linkage of AO and
73 precipitation extremes over China. In some case studies, the AO was suggested, at
74 least partly, to relate to the extraordinarily frequent and long-lasting snowstorms over
75 central-southern China in January 2008 (Wen et al. 2009). Also, the AO was
76 accompanied by significant changes in the southern branch of the westerlies over the
77 southern flank of the Tibetan Plateau during winter, which is important for the winter
78 large-scale precipitation event over China via its control on water vapor supply and
79 weather activity (Suo & Ding 2009, Zhang et al. 2009). Therefore, we hypothesize
80 that the winter AO have had a robust influence on extreme precipitation events over
81 China during the last decades. The main goal of present study is to investigate the
82 statistical linkage between the AO and the extreme precipitation events over China
83 during the period of 1954 to 2009, and to investigate the large-scale circulation and
84 climate features observed with respect to the AO-precipitation connection.
85 The rest of the paper is organized as follows. Section 2 describes the data and
86 method used in the study. In Section 3, the statistical relationship between the AO and
4

C1041, doi: 10.3354/cr01041, In press
87 the precipitation extremes over China is analyzed, and compared that with the
88 relationship between the El Niño/Southern Oscillation (ENSO) and the precipitation
89 extremes over China. Section 4 presents atmospheric circulation anomalies observed
90 in conjunction with the AO-precipitation links. Section 5 shows the corresponding
91 weather disturbances in association with the AO-precipitation links. Finally, a
92 discussion and a summary are provided in Section 6 and Section 7, respectively.
93
94 2 DATA AND ANALYSIS METHOD
95 2.1 Data
96 The station daily precipitation dataset used in this study is obtained from China
97 Meteorological Administration. This dataset covers a time period from 1954 to 2009.
98 We firstly screened the data because there are some missing values in this dataset.
99 These missing values do not allow for constructing time series such as extreme
100 precipitation frequency, which typically require complete dataset. Thus, stations that
101 have missing values are all omitted. Among more than 700 stations, we selected 287
102 stations for which dataset is complete.
103 The monthly AO index proposed by Li & Wang (2003) is used in this study, which
104 is the difference in surface pressure between 35°N and 65°N around the hemisphere
105 based on National Centers for Environmental Prediction and National Center for
106 Atmospheric Research (NCEP–NCAR) monthly reanalysis dataset. This AO index
107 has a more symmetric correlation with surface pressure anomalies around the
108 hemisphere than does the leading empirical orthogonal function of Thompson and
5

C1041, doi: 10.3354/cr01041, In press
109 Wallace (1998) (Angell, 2006). The AO index is available at
110 http://www.lasg.ac.cn/staff/ljp/data-NAM-SAM-NAO/NAM-AO.htm. In addition, the
111 influence of ENSO signal on the precipitation extremes is considered in this study.
112 The ENSO signal is represented using monthly Niño 3.4 sea surface temperature
113
(SST). That is the mean of the SST anomalies over the area of 5°N  5°S and
114
170°W  120°W. The time series of AO index and ENSO index are constructed from
115 the monthly value by averaging the data within the analysis period.
116 In order to analyze atmospheric circulation variation associated with the AO
117 change, the NCEP/NCAR reanalysis dataset is used in the present study. The physical
118 variables analyzed include specific humidity, horizontal wind, and geopotential at
119 eight standard pressure levels, namely, 1000, 925, 850, 700, 600, 500, 400, and 300
120 hPa. All variables have a spatial resolution of 2.5° latitude × 2.5° longitude. In
121 addition, the column atmospheric water vapor flux used for measuring the water vapor
122 transport is computed by vertically integrating from surface to 300hPa same as Zhou
123 & Yu (2005). All the datasets described above are confined to the period of 1954
124 through 2009.
125
126 2.2 Analysis method
127 To delineate the linkage between AO and extreme precipitation events, we defined
128 an extreme precipitation frequency by percentiles through following procedure. For a
129 given station for a given period, 1) the 80 th percentile of daily precipitation during the
130 reference period 1971–2000 is measured and then used as the threshold value; 2)
6

C1041, doi: 10.3354/cr01041, In press
131 when a daily precipitation exceeding the threshold value, this day is classified as an
132 extreme precipitation event; 3) the number of days with daily precipitation exceeding
133 the threshold is counted year by year, and finally the time series of the frequency of
134 extreme precipitation events is constructed (hereafter referred to as P80th). To
135 highlight the lower probability events, we also analyzed the extreme precipitation
136 events using a stricter threshold of the 90 th percentile (hereafter referred to as P90th).
137 It is worth noting that on the choice of the threshold of extreme events, a stricter
138 threshold of the 90 th percentile or the 95 th percentile is usually employed. In this study,
139 however, we mainly used the 80 th percentile for analyzing AO-extreme precipitation
140 links. The choice of the 80th percentile is determined by the fact that when using the
141 80th percentile to define the extreme precipitation events, there are more stations with
142 significantly positive correlation between AO and extreme precipitation frequency, as
143 compared to using the 90th percentile and the 95th percentile. Of all 287 stations in
144 China, there are 59, 46 and 32 stations with significantly positive correlation,
145 respectively, for using the 80th percentile, the 90th percentile, and the 95th percentile
146 to define the extreme precipitation frequency (figure not shown). Moreover, there is
147 no extreme precipitation event in several years when using the 90th percentile and the
148 95th percentile. In contrast, the 80th percentile produces a moderately larger number
149 of extreme precipitation events in each year, which would help yield a more confident
150 analysis for AO-climate relationship. Therefore, the choice of the 80th percentile may
151 produce a more confident analysis for the AO-precipitation extremes relationship.
152 Accordingly, when computing the extreme event frequency of other atmospheric
7

C1041, doi: 10.3354/cr01041, In press
153 variables, the 80 th percentile or the 20 th percentile is used as the threshold in this
154 study.
155 Many previous studies have showed that winter climates over East Asia could be
156 impacted by the ENSO. For instance, Zhang & Sumi (2002) showed that there are
157 positive precipitation anomalies in winter over southern areas of China during the El
158
Niño mature phase. The ENSO often develops into a mature phase in boreal winter.
159 Its influence on winter precipitation might conceal or distort the AO-related signals. In
160 the present study we compared the precipitation anomalies solely associated with AO
161 and ENSO, respectively. In order to obtain the precipitation anomalies solely
162 associated with the AO, we regressed both time series of AO index and precipitation
163 upon the ENSO index, respectively. The two residuals of original time series minus
164 regression-estimation are regarded as the ENSO-free components, linearly speaking.
165 Then the results of correlation and regression analysis upon these two residuals are
166 considered as the precipitation anomalies solely associated with the AO. Since the
167 climate changes (such as the monthly-seasonal mean circulations) in mid-high
168 latitudes are quasi-linearly related to the ENSO, therefore this treatment of the ENSO
169 signal would work well (Hoerling et al. 1995). Similarly, in order to obtain the
170 precipitation anomalies solely associated with the ENSO, the two time series of
171 precipitation and ENSO index are regressed upon the AO index, respectively. The two
172 residuals are regarded as the AO-free components, linearly speaking. Then the results
173 of correlation and regression analysis upon these two residuals are suggested as the
174 precipitation anomalies solely associated with the ENSO. It should be pointed out that
8

C1041, doi: 10.3354/cr01041, In press
175 here we did not consider the possible influence caused by the AO-ENSO interactions
176 (Quadrelli & Wallace 2002, Nakamura et al. 2006).
177 3 RELATIONSHIP BETWEEN AO AND EXTREME PRECIPITATION
178 EVENTS
179 3.1 Correlation between AO and P80th
180 In order to identify whether there are statistical AO-extreme precipitation links, we
181 firstly investigated the correlation between AO and P80th averaged over China. Since
182 AO is most active during winter season, the AO is confined to the October to March
183 period. To take into account the possible time-lag relationship between AO and
184 precipitation, we computed their cross-correlation with a couple of months delay.
185 Results are shown in Figure 1. Among all months analyzed, significantly
186 simultaneous AO-P80th relationship can be found only in January and February. In
187 other months, the AO-P80th relationship is weak, and not significant. The
188 simultaneous correlation between AO and P80th is 0.32 during January and 0.22
189 during February, which are significant above the 90% confidence level. When
190 averaged January and February, the AO-P80th correlation becomes 0.38, significant
191 above the 99% confidence level (Table 1). We also examined the AO-P90th
192 relationship, and found that their significant relationship only occurs in January and
193 February, too (figure not shown). The simultaneous correlation between AO and
194 P90th is 0.31 during January and 0.25 during February, which are significant above
195 the 90% confidence level. Their mean time series for January to February have a
196 correlation of 0.38, significant at the 99% confidence level. These similar features in
9

C1041, doi: 10.3354/cr01041, In press
197 P90th and P80th suggest a statistical relationship of AO-precipitation extremes in
198 January–February. Thus, in the following analysis, we only investigate the
199 AO-precipitation links in January to February, and the winter mean is defined as the
200 average of January–February.
201 Besides the P80th, we analyzed the linkage between AO and precipitation amount.
202 In Figure 1, plotted together with the P80th are the correlations for precipitation
203 amount. The simultaneous correlation between AO and precipitation amount is 0.21
204 during January and 0.17 during February. Although they are not significant, the mean
205 precipitation amount of winter yields a correlation of 0.31, which is significant above
206 the 95% confidence level (Table 1). Overall, the AO-precipitation amount is
207 consistent with AO-P80th relationship.
208 Secondly, we computed the correlation for each station and then showed the spatial
209 feature of the AO-P80th relationship (Figure 2a). As seen in the figure, positive
210 correlations are covering large part of China and negative correlations only occurs in
211 the northeast of China. Of all 287 stations, 238 stations have positive correlations and
212 49 stations have negative correlations. Among all stations, 82 (104) stations are
213 positively significant at the 95% (90%) confidence level. The stations with
214 significantly positive correlation are mainly located in the central-southern China
215
(enclosed by solid lines in Figure 2a, approximately in 24ºN–35ºN and
216
101ºE–123ºE).To get more details on the temporal changes in precipitation extremes,
217 we constructed a time series of January–February P80th over the central-southern
218 China and compared with the January–February AO index. As shown in Figure 3a, the
10

C1041, doi: 10.3354/cr01041, In press
219 two time series display an in-phase relationship. When AO is in a positive phase (such
220 as the early 1990s) there tends to be a larger value of P80th. Meanwhile, when AO is
221 in a negative phase during years such as 1958, 1960, 1963, 1966, 1969, and 1977,
222 there are also lower values of P80th. Their in-phase co-variations are well measured
223 by correlation analysis. The correlation between them is 0.49, significant at the 99%
224 confidence level (Table 1). Because there are strong linear trends in both time series
225 of the AO and the P80th over the central-southern China, which could produce high
226 correlation even when there is no physical linkage between them, we removed the
227 linear trends of both time series and then calculated the correlation between them. The
228 newly derived correlation coefficient is 0.45, significant at the 99% confidence level.
229 Therefore, the AO is significantly associated with the changes of P80th over the
230 central-southern China.
231 Finally, we investigated the spatial feature of the AO-precipitation amount
232 relationship (Figure 2b). The spatial feature is similar to the P80th correlation.
233 Positive correlations are spread over the major part of China, except some stations in
234 the northeast of China have negative correlations. The significantly positive
235 correlations are also centered over the central-southern China. Of all 287 stations, 230
236 stations have positive correlations, and 92 (113) stations are positively significant at
237 the 95% (90%) confidence level. We also computed a time series of mean
238 precipitation amount averaged over the central-southern China and compared that
239 with the time series of the AO index. As shown in Figure 3a, the AO and the
240 precipitation amount display a well in-phase relationship. The AO experiences
11

C1041, doi: 10.3354/cr01041, In press
241 decadal changes, being a low-value period during 1954–1986, an evident high-value
242 period during 1987–1995, and a relative high-value period during 1996–2009.
243 Accordingly, the precipitation amount changes from the low values in 1954–1986 to
244 the evident high values in 1987-1995 and to the relative high values in 1997–2009.
245 They yield a correlation of 0.45, significant at the 99% confidence level (Table 1).
246 When filtering the long-term trends of both time series, the correlation becomes 0.28,
247 significant at the 95% confidence level. The significant relationship between the AO
248 and the precipitation amount over the central-southern China provides additional
249 confidence for the significant AO-P80th relationship.
250 3.2 Comparison of AO-related precipitation changes and ENSO-related precipitation
251 changes
252 Winter climates over East Asia could be impacted by both AO and ENSO. Is the
253 AO-precipitation relationship analyzed in the previous sections distorted by the ENSO
254 signal? To answer this question, here we tried to compare the AO-related precipitation
255 anomalies and the ENSO-related precipitation anomalies, including the extreme
256 frequency and the amount.
257 To estimate the precipitation anomalies solely associated with the AO, for each of
258 stations both time series of precipitation and AO index are regressed upon the ENSO
259 index, respectively, and then the two residuals are subjected to correlation analysis
260 (for details see the Section 2.2). Figure 4a presents the spatial distribution of
261 correlation between the AO and the P80th, which is obtained by excluding the
262 ENSO-related components from their original time series. As seen in the figure,
12

C1041, doi: 10.3354/cr01041, In press
263 positive correlations are significant in spatial extent over China. Of all 287 stations,
264 226 stations have positive correlations and 94 (114) stations are positively significant
265 at the 95% (90%) confidence level. The significant stations are mainly located over
266 the central-southern China. The correlation between the AO and the P80th averaged
267 over the China (central-southern China) is 0.43 (0.45), significant at the 99%
268 confidence level (Table 1). Meanwhile, we examined the precipitation amount
269 anomalies solely associated with the AO (figure not shown). As seen in the figure,
270 positive correlations are prevailing in China and stations with significantly positive
271 correlation are centered over the central-southern China. Among 287 stations, 234
272 stations have positive correlations and 107 (122) stations are positively significant at
273 the 95% (90%) confidence level. The correlation between the regional mean
274 precipitation amount over China (central-southern China) and the AO is 0.38 (0.50),
275 significant at the 99% confidence level(Table 1). The precipitation amount anomalies
276 solely associated with the AO are consistent with the P80th anomalies solely
277 associated with the AO. When compared with the AO-precipitation correlation
278 obtained through the original time series (Figure 2a), this AO-precipitation correlation
279 derived by excluding the ENSO-related components from the original time series can
280 cause more stations with significantly positive correlation over China, especially over
281 the central-southern China, and cause more stations with higher correlation
282 coefficient.
283 To compare with the AO, the precipitation anomalies solely associated with the
284 ENSO are examined. For each station, the original time series of the precipitation and
13

C1041, doi: 10.3354/cr01041, In press
285 the ENSO index are regressed upon the AO index, respectively. Then the two
286 residuals of original time series minus the regression-estimation are regarded as the
287 AO-free components and subjected to correlation analysis. Figure 4b shows the
288 spatial feature of correlation between the ENSO and the P80th, which is obtained by
289 excluding the AO-related components from both original time series. As shown in the
290 figure, in association with positive Ni
ñ o3.4 SSTs, the P80th increases significantly
291 over the southern coast regions of China, but decreases moderately over the central
292 and northeastern China. We constructed an average time series of P80th over the
293 southern coast region (approximately in 22ºN–28ºN and 100ºE–120ºE) and compared
294 with the Niño 3.4 SST time series. After excluding the AO-related components from
295 the original time series, the correlation between the ENSO index and regional mean
296 P80th is 0.43, significant above the 99% confidence level. In addition, the
297 precipitation amount anomalies solely associated with the ENSO are also examined.
298 The results are almost identical to the P80th, significantly positive correlations
299 appearing in the southern regions and insignificantly negative correlations in the
300 central and northeastern China (figure not shown). When the AO-related components
301 are excluded from the original time series, the correlation between the ENSO index
302 and the regional mean precipitation amount over the southern coast region is 0.43,
303 significant above the 99% confidence level.
304 According to the results above, the precipitation amount/extreme difference
305 between the AO and the ENSO phases can be clearly identified. The difference mainly
306 occurs in the spatial distribution of the precipitation anomalies between the
14

C1041, doi: 10.3354/cr01041, In press
307 AO-precipitation relationship and the ENSO-precipitation relationship. The AO shows
308 much stronger connection to the precipitation amount/extreme over the
309 central-southern China, but the ENSO effect on precipitation amount/extreme is more
310 significant over the southern coast region of China. This difference might be caused
311 by different physical processes involved in the precipitation connections.
312
313 4 CHANGES IN ATMOSPHERIC CIRCULATION AND WATER VAPOR
314 TRANSPORT
315 4.1 Lower to middle troposphere
316 In order to investigate the dominant atmospheric circulation changes in association
317 with the AO-precipitation links, we performed regression analysis of simultaneous
318 anomalies in low-level to middle-level tropospheric atmosphere. The analyzed
319 variables consist of horizontal winds at 700 hPa level and geopotential height at 500
320 hPa level (H500). Note that the ENSO-precipitation correlations over our analysis
321 region (i.e. the central-southern China) are quite weak as shown in the previous
322 section. Hence, for simplicity, when investigating the AO-related atmospheric
323 circulation changes and water vapor transport, we used the original time series (not
324 considering the possible influence of the ENSO signal) in the following sections.
325 Figure 5a shows the regression coefficients of horizontal winds at the 700 hPa level
326 against the AO index. When the AO is in a positive phase, the wind anomaly pattern is
327 characterized by four distinct anomalous centers: two anticyclonic anomalies
328 appearing over the northeast Asia (35ºN–55ºN and 110ºE–140ºE) and the Arabian Sea
15

C1041, doi: 10.3354/cr01041, In press
329
(10ºN–25ºN and 40ºE–70ºE) and two cyclonic anomalies over the western Siberia
330 (40ºN–60ºN and 40ºE–80ºE) and the Bay of Bengal (BOB for short, 15ºN–25ºN and
331 80ºE–100ºE). At the same time, significantly anomalous southerly winds is prevailing
332 across the south of China, especially the central-southern China. These significant
333 southerly wind anomalies over the central-southern China might be a direct reason in
334 causing precipitation extremes as well as the precipitation amount. These anomalous
335 southerlies are likely linked to the configuration of the cyclonic anomaly over the
336 BOB and the anticyclnoic anomaly in the northeast Asia. More importantly, the
337 cycnolic circulation over the BOB is surely a sign of the frequent activities of
338 southern branch trough (SBT for short, also called India-Burma trough) over the
339 BOB. The SBT, which is mostly apparent at 700 hPa level, is the most favorable
340 synoptic system for the formation of precipitation over southern China during
341 wintertime (Suo & Ding 2009). Many previous studies indicated that a more active
342 SBT may enhance weather disturbance and moisture supply from the BOB to the
343 central-southern China (e.g. Gao & Yang 2009, Bao et al. 2010), and in return, they
344 can contribute to the increased precipitation amount and the extreme precipitation
345 events over the central-southern China.
346 We subsequently analyzed H500 anomaly in the middle troposphere associated with
347 the positive AO phase. As seen in Figure 5b, a positive AO phase is related to three
348 significant height anomalies in the middle troposphere: two positive height anomalies
349 over Northeast Asia (35ºN–55ºN and 110ºE–140ºE) and Arabian Sea (10ºN–25ºN and
350
40ºE–70ºE), respectively, and a negative anomaly over western Siberia (40ºN–60ºN
16

C1041, doi: 10.3354/cr01041, In press
351 and 40ºE–80ºE). The positive height anomaly over the Northeast Asia indicates a
352 weak East Asian trough, meanwhile the negative height anomaly over the western
353 Siberia shows a weak Ural High. The changes in the East Asian trough and the Ural
354 High implies a weak East Asian winter monsoon, which is consistent with the
355 southeasterly anomalies across the northern China to the Mongolia in the lower
356 troposphere, as documented in Figure 5a. The weaker winter monsoon in the East
357 Asia provides favorable conditions for anomalous northward transport of warm and
358 moist air from the BOB and the southern China to the central China. Therefore, both
359 the anomalous southerly wind across the central-southern China and the weaker
360 winter monsoon in China are helpful for causing more precipitation extreme events
361 and amount over the central-southern China.
362
363 4.2 Upper troposphere
364 Here the zonal wind at 200 hPa level (U200) is regressed onto the AO index to
365 represent the anomaly in the upper troposphere associated with the AO change (Figure
366 5c). As shown in the figure, during a positive AO phase there are two bands of
367 positive anomaly: to the north of northeastern China (50ºN–60ºN and 100ºE–160ºE)
368 and over the West Asia (25ºN–40ºN and 40ºE–80ºE). Meanwhile, there is a long
369 distance belt of negative anomaly, stretching from the Arabian Sea (10ºN–25ºN and
370
40ºE–70ºE) to the region to the south of Japan, and with both centers over the Arabian
371 Sea and the region to the south of Japan, respectively. These anomalies are significant
372 above the 95% confidence level. The negative anomaly to the south of Japan indicates
17

C1041, doi: 10.3354/cr01041, In press
373 a weak East Asian jet stream, and the positive anomaly over the West Asia
374 (25ºN-40ºN and 40ºE-80ºE) implies a strong Middle East jet stream (MEJS). Many
375 previous studies (e.g. Gao & Yang 2009, Wen et al. 2009) emphasized the influences
376 of upper tropospheric jet stream, such as the MEJS, on precipitation anomalies over
377 the central-southern China. The intensified MEJS strengthens the trough embedded in
378 the southern branch of the subtropical westerlies over the southern flank of Tibetan
379 Plateau (Wen et al. 2009), thus bringing more water vapor from the BOB to eastern
380 China (see Figure 5d). Following Yang et al (2004), we defined an index of MEJS as
381
U200 in (20ºN–30ºN, 40ºE–70ºE) minus U200 in (30º–40ºN, 15º–45ºE). The
382 correlation between the MEJS and the P80th over the central-southern China is 0.43,
383 significant at the 99% confidence level. Meanwhile, the AO-MEJS correlation is 0.36,
384 significant above the 99% confidence level. The significant correlations among the
385 AO, the MEJS, and the P80th over the central-southern China provide additional
386 evidence that the MEJS might be an important factor in connecting AO and
387 precipitation over the central-southern China.
388 4.3 Water vapor transport
389 We computed the regression of water vapor transport on the AO index. As seen in
390 Figure 5d, following a positive AO phase, the anomalies in water vapor flux are
391 characterized by a cyclonic anomaly over the BOB, southwesterly anomaly over the
392 south of China, large area of easterly anomaly to the south of Japan (15ºN–30ºN and
393 120ºE–160ºE), and an anticyclonic anomaly over the Arabian Sea (10ºN–25ºN and
394
40ºE–70ºE). All these anomalies are significant at the 90% confidence level. The
18

C1041, doi: 10.3354/cr01041, In press
395 southwesterly anomalies of water vapor flux over the south of China, especially over
396 the central-southern China, which help induce larger precipitation amount and more
397 precipitation extremes, are consistent with the southerly wind anomalies in the lower
398 troposphere (see figure 5a). Note that the southwesterly anomalies of water vapor flux
399 over the central-southern China are likely caused by the cyclonic water vapor flux
400 anomaly over the BOB and the large area of easterly water vapor flux anomaly over
401 the northeast Asia. The cyclonic anomaly of water vapor flux over the BOB is
402 associated with frequent SBT activities there. In associated with a positive AO phase,
403 when the SBT activity over the BOB strengthens, it initiates the cyclonic anomaly of
404 water vapor flux and conducts much water vapor to the south of China and transports
405 water vapor northward farther than normal. Meanwhile, the large area of easterly
406 water vapor flux anomaly to the south of Japan (15ºN–30ºN and 120ºE–160ºE) may
407 be caused by the geopotential height anomaly in the middle troposphere shown in Fig.
408 5b over the northeast Asia. The easterly water vapor flux anomaly to the south of
409 Japan would enhance moisture transport from the Pacific to the central-southern
410 China and thus help induce more precipitation over the central-southern China.
411 In addition, we analyzed variation of vertical air motions (represented by the omega
412 averaged between 700 hPa and 500 hPa) associated with AO change (Figure 5d).
413 Negative (positive) values of omega indicate ascending (descending) motions, which
414 are denoted by dashed (solid) lines in the figure 5d. As seen in the figure, when AO is
415 in a positive phase, a significant enhancement of ascent motions appear from the east
416 of the BOB to the central-southern China. The changes over these regions are
19

C1041, doi: 10.3354/cr01041, In press
417 significant at the 95% confidence level. This evident enhancement of ascending
418 motions over the south of China may be caused by increased SBT activities over the
419 BOB. The enhancement of SBT activity not only provides excess water vapor from
420 the BOB to the central-southern China, but also initiates ascending motions over the
421 south of China (Suo and Ding, 2009). Therefore, the SBT over the BOB can be an
422 important factor for the occurrence of extreme precipitation over the central-southern
423 China.
424 Overall, the configuration of arising motion and excess water vapor transport over
425 the central-southern China in association with a positive AO phase enhance the
426 possibility of occurrence of extreme precipitation events over the central-southern
427 China. During this process, the increased SBT activity plays a role in initiating
428 ascending motions, meanwhile, the easterly water vapor flux anomaly to the south of
429 Japan produces the excess water vapor transports.
430
431 5 CHANGES IN WEATHER DISTURBANCES
432 5.1 Weather disturbances in meridional wind
433 Changes in monthly-seasonal circulation, as depicted above, support the AO-P80th
434 relationship over the central-southern China, representing the climate background for
435 the extreme events. But weather activity on the synoptic scale is the direct cause for
436 extreme precipitation events. Therefore, in order to further explain the AO-P80th
437 relationship, it is needed to also examine the AO-related changes in weather activity
438 on the synoptic scale. Here we used weather disturbances (referred to as
20

C1041, doi: 10.3354/cr01041, In press
439 high-frequency variations in meridional wind and omega
 with typical synoptic
440 timescale in this manuscript) to represent weather activity.
441 We firstly checked the AO-related changes in the frequency of strong meridional
442 wind. The frequency of strong meridional wind is set by percentiles using daily
443 meridional wind at 700 hPa level (V700) through following procedure. 1) A
444 climatology of daily V700 during January to February is obtained on the reference
445 period of 1971-2000. 2) The climatological V700 determines the threshold value; for
446 a given grid point, when its climatological V700 is southerly (northerly) wind, its
447 threshold value is the 80th (20th) percentiles of daily V700 during January to
448 February in 1971–2000. 3) For the grid point with southerly (northerly) climatological
449 V700, the number of days with daily V700 exceeding (below) the threshold during
450 January to February is counted year by year, and finally yields a time series of the
451 frequency of strong meridional wind.
452 We applied a regression analysis of strong meridional wind frequency against AO
453 index. Results are shown in Figure 6. During a positive AO phase, there are higher
454 frequencies of strong southerly wind from the east of BOB to southern China and
455 higher frequencies of strong northerly wind over the west of BOB. This implies that
456 the positive AO winters tend to be accompanied by frequent strong SBT activities.
457 The enhanced strong SBT activity is in favor of precipitation over the central-southern
458 China. At the same time, when the AO is in a positive phase, there is a decreased
459 frequency of strong northerly wind across Mongolia and northern China. It indicates
460 that East Asian trough shifts eastward and causes cold monsoonal flow moving
21

C1041, doi: 10.3354/cr01041, In press
461 eastward off the continent. Therefore, moist air can be transported much further north
462 into the central-southern China, which is helpful for bringing more precipitation in
463 amount as well as in the extreme events, as depicted in Section 4.1.
464 As analyzed above, the SBT is an important circulation system for initiating
465 precipitation over the central-southern China. On synoptic time scales, the SBT
466 appears as a transient trough, or cyclone, or low pressure system. To delineate the
467 synoptic SBT activity, we defined a daily SBT index. An active SBT generally has
468 persistent strong meridional shear in V700 (Suo & Ding 2009), hence we divided the
469 domain of climatological SBT into two regions, namely, region 1 (80ºE–90ºE and
470
15ºN–25ºN) and region 2 (90ºE–100ºE and 15ºN–25ºN), and used their difference of
471 daily V700 averaged in both regions (region 2 minus region 1) as the daily SBT index.
472 A high index measures a larger difference, and corresponds to a stronger SBT activity.
473 Because extreme precipitation over the central-southern China during winter is
474 normally caused by strong SBT activities, we constructed a strong SBT frequency and
475 checked the relationships among the strong SBT frequency, the AO and the P80th
476 over the central-southern China. The strong SBT frequency is defined by percentiles
477 based on the daily SBT index. Threshold value for computing strong SBT frequency
478 is set as the 80th percentile of daily SBT index during January to February in
479 reference period of 1971–2000. The number of days with daily SBT index exceeding
480 the threshold value during January to February is counted year by year and finally
481 yields the time series of strong SBT frequency during 1954–2009. The time series of
482 the AO index is obtained by averaging AO index in January to February during
22

C1041, doi: 10.3354/cr01041, In press
483 1954–2009.The time series of strong SBT frequency and AO index are plotted
484 together for comparison in the figure 3b. As seen in the figure, both time series
485 co-change consistently with an in-phase relationship. The correlation between them is
486 0.55, significant above the 99% confidence level. After excluding the ENSO-related
487 components from their original time series, the correlation remains 0.55, still
488 significant at the 99% confidence level (Table 1). Meanwhile, the correlation between
489 the strong SBT frequency and the P80th averaged over the central-southern China is
490 0.51, significant above the 99% confidence level. After excluding ENSO-related
491 components from the original time series, the correlation becomes 0.58, significant at
492 the 99% confidence level. These high correlations suggest that a positive AO phase is
493 accompanied by frequent strong SBT activity, which is related to the extreme
494 precipitation events.
495 Finally, we checked the frequency distribution of daily SBT in detail. We selected 5
496 years with the highest AO indices (1989, 1990, 1993, 2002, and 2008), and compared
497 that to the 5 years of the lowest AO index (1956, 1963, 1965, 1968, and 1969).
498 Results show that the frequency distribution of daily SBT in these years is
499 quasi-normal distribution. During the positive AO years, the mean and variance of
500 daily SBT index are 3.8 m s
–1
and 11.66 m 2 s
–2
, however, during the negative AO
501 years they are 1.9 m s
–1 and 7.76 m 2 s
–2 for mean and variance, respectively. It is
502 evident that compared to the negative AO years, the positive AO years is accompanied
503 by decreased frequency of small SBT index and increased frequency of high SBT
504 index (Figure 7). Based on the frequency distribution, we estimated the percentage of
23

C1041, doi: 10.3354/cr01041, In press
505 the SBT extremes in daily indices, and then compared the percentage changes in
506 association with the high-AO years and low-AO years. During the positive AO years,
507 the extreme SBT accounts for nearly 35%. But the percentage remarkably drops to
508 only 10% during the negative AO years. This comparison is consistent with the
509 significant positive correlation between the AO and the strong SBT frequency. Thus,
510 the positive AO phase tend to be accompanied by an evident increasing of strong SBT
511 events. That consequently induces the increasing of precipitation amount and extreme
512 frequency over the central-southern China.
513
514 5.2 Weather disturbances in vertical motion
515 In order to better understand the AO-related changes in large scale arising motions,
516 which plays important roles during extreme precipitation, we analyzed the AO-related
517 changes in the weather disturbances of omega. The weather disturbance of omega are
518 represented in the forms of , where the prime indicates the anomalies resulting
519 from high-pass filtering and the bar indicates the time mean during January to
520 February. The typical timescale of weather activity is about 7 days; therefore, we
521 applied a high-pass filter to remove the components longer than 7 days. Then the
522 synoptic scale variance is calculated locally for January to February. Here the
523 omega
 variance between 700-500 hPa levels are averaged. Figure 8a shows the
524 correlation between weather disturbance of omega and AO. Positive correlation spans
525 through a band along the southern flank of the westerly flow with centers in West Asia
24

C1041, doi: 10.3354/cr01041, In press
526 (25ºN–40ºN and 40ºE–70ºE), northern Indian (20ºN–30ºN and 70ºE–85ºE), and
527 Myanmar (15ºN–25ºN and 90ºE–100ºE), respectively. Meanwhile, there occurs
528 positive correlation over northeast Asia (30ºN–55ºN and 100ºE–140ºE). This belt of
529 positive correlation from the West Asia (25ºN–40ºN and 40ºE–70ºE) to Myanmar
530
(15ºN–25ºN and 90ºE–100ºE) is apparently the track of the MEJS along the southern
531 Tibetan Plateau. It implies that the MEJS may be an important factor for the
532 enhancement of weather activity over the BOB during a positive AO phase. The
533 enhanced SBT and synoptic activity over the BOB and southern China is not a local
534 phenomenon but with large scale in conjunction with the AO-related westerly jet
535 stream variations.
536 In addition, we investigated the correlation between weather disturbances of the
537 omega and the P80th over the central-southern China (Figure 8b). In association with
538 the increased P80th over the central-southern China, the changes in weather
539 disturbances increase over the West Asia (25ºN–40ºN and 40ºE–70ºE), Myanmar
540
(15ºN–25ºN and 90ºE–100ºE), southern China (10ºN–30ºN and 100ºE–120ºE), and
541 northeast Asia (30ºN–55ºN and 100ºE–140ºE). Clearly, the regional precipitation
542 extremes in the central-southern China may be associated with the weather
543 disturbances in vertical motion. Interestingly, the west-east oriented correlation from
544 the BOB to the south of China, and the simultaneous high correlation over the West
545 Asia, again suggest the importance of the transient variations along the MEJS and the
546 southern branch of the westerlies over the southern flank of the Tibetan Plateau.
547 Therefore, it can be concluded that a positive AO phase may be accompanied by an
25

C1041, doi: 10.3354/cr01041, In press
548 intensified MEJS and SBT, which could enhance synoptic activities over the north of
549 BOB and central-southern China. The enhanced weather disturbances in vertical air
550 motion induce more precipitation amount and extremes over the central-southern
551 China.
552 6 DISCUSSIONS
553 The results above indicate that the AO has a statistical relationship with the
554 frequency of extreme precipitation events over the central-southern China during
555 January to February. However, some issues that should be discussed are: 1) whether
556 this relationship is stable across different periods; 2) whether this relationship depends
557 on the phases of solar activity; 3) whether this relationship is caused by the fact that
558 the AO may be driving precipitation amount as a whole and not specifically extreme
559 precipitation; 4) what mechanism exists for the AO to link MEJS and SBT.
560 Firstly, we calculated correlation between the AO and the P80th for each of stations
561 twice during two different periods, respectively, i.e., 1954 to 1981 and 1982 to 2009
562 (figure not shown). Results show that during different periods, the major of stations
563 show an in-phase relationship between the AO and the P80th, and stations with
564 significantly positive correlation mainly occur over the central-southern China. This
565 pattern is similar to that obtained through the data during the whole period
566 (1954-2009, see figure 2a), although the number of stations with significantly positive
567 correlation is a bit less than that in the latter. Thus, the AO-related changes in the
568 frequency of extreme precipitation is stable.
26

C1041, doi: 10.3354/cr01041, In press
569 Secondly, we examined whether the AO-P80th relationship depends on the phases
570 of solar activity during winter. Recently some studies have indicated that solar cycle
571 phase may modify the influence of AO on the wintertime climate (e.g., Gimeno et al.,
572 2003). For solar maximum phases NAO and Northern Hemisphere temperature are
573 positively correlated, but for solar minimum phases correlations are not significant or
574 even negative. We used monthly mean values of the 10.7cm solar ratio flux during
575 January to February to represent the wintertime solar cycle phase. The solar ratio flux
576 data is downloaded from National Geophysical Data Center, NOAA
577 (www.ngdc.noaa.gov/stp/solar/solardataservices.html). Based on this dataset, we
578 selected 25 high solar activity winters and 31 low solar activity winters, depending on
579 whether the solar ratio flux was higher or lower than the average. Then we computed
580 the correlations between the AO and the P80th averaged over the central-southern
581 China during high and low solar activity years, respectively, and compared them to
582 the correlation derived by whole data during 1954-2009. Results show that during the
583 high solar activity winters, the correlation is increased with a correlation coefficient of
584 0.56. On the contrary, during the low solar activity winters, the correlation is
585 decreased with a correlation coefficient of 0.32. These correlations are significant at
586 the 99% confidence level. Therefore, the AO-related anomalies in the frequency of
587 extreme precipitation over the central-southern China may be enhanced during the
588 solar maximum phases.
589 Due to that the spatial distribution of correlation between P80th and AO is not
590 different from that of the correlation between precipitation amount and AO, this
27

C1041, doi: 10.3354/cr01041, In press
591 would indicate that the AO may be driving the distribution of precipitation amount as
592 a whole and not specifically extreme precipitation events. For this claim to be justified,
593 we thirdly showed the probability distribution of daily precipitation amount in January
594 to February during AO extreme years; the selection of the AO extreme years are the
595 same in Section 5.1 (figure not shown). We found that there are two different types of
596 probability distribution of daily precipitation amount during the AO extreme years.
597 When all stations in China are considered, the probability distribution curve (pdv)
598 during the positive AO phase is shifted to the right as a whole compared with the pdv
599 during the negative AO phase; it means that the AO may be driving the precipitation
600 distribution as a whole and consequently increases the frequency of extreme
601 precipitation events. However, when stations in the central-southern China are solely
602 considered, the pdv is skewed; the probability of precipitation amount between 3mm
603 and 10 mm is high in the positive AO phase than that in the negative AO phase, and
604 the probability of precipitation amount less than 3 mm is not changed during the AO
605 extreme years. The above analysis implies that in accordance to the central-southern
606 China, the AO may be evidently influencing the extreme precipitation events over
607 there.
608 Finally, a causal mechanism has been proposed linking AO to MEJS and SBT
609 through a wave train bridge. Figure 9 shows a composite difference of wave activity
610 flux (Takaya and Nakamura, 2001) and geostrophic streamfuction between positive
611 AO phase and climate means. To highlight the pure impacts of the AO rather than the
612 ENSO on MEJS and SBT, 3 years are selected, i.e., 1990, 1993, and 2002, in which
28

C1041, doi: 10.3354/cr01041, In press
613 AO index is larger than 1 standard value and absolute value of Niño 3.4 SST is below
614 0.5 standard value. On the map of composite difference, when AO is in a positive
615 phase, a wave train-like pattern occurs emanating downstream from the Europe to
616 South Asia, with both positive anomalies of geostrophic streamfuction locating over
617 the Europe (40°N-60°N, 0-30°E) and the northwest India (10°N-30°N, 60°E-90°E),
618 respectively, and a negative anomaly of geostrophic streamfuction over the West Asia
619 (30°N-40°N, 30°E-40°E). Horizontal wave activity flux is superimposed over the
620 wave train. Horizontal wave activity flux clearly illustrates that the wave train seems
621 to originate from the Europe across the northern Africa to the northwest India. There
622 are some studies dealing with the wave train response which originates from the
623 Europe across the northern Africa to the northwest India. Suo and Ding (2010)
624 constructed this wave train in geopotential height field at 500 hPa level using one
625 point correlation map. Their study suggested that when the positive height anomaly
626 over the Europe strengthens, the positive height anomaly over the West Asia becomes
627 strong and consequently increases an intensified SBT. All the processes are associated
628 with enhanced MEJS that works as waveguide to induce the wave propagation. In
629 return, the intensified SBT over the BOB may transport more water vapor transport
630 from the BOB to the central-southern China, and provide favorable weather condition
631 for the occurrence of precipitation extreme event. It is worth noting that, although
632 figure 9 skillfully implies the possible mechanism that links AO to MEJS and SBT,
633 the knowledge of mechanisms needs the verification of numerical modeling and it is
634 currently beyond the scope of this study.
29

C1041, doi: 10.3354/cr01041, In press
635
636 7 CONCLUSIONS
637 In this study, we have examined the relationship between the AO and the extreme
638 precipitation frequency over China during the 1954 to 2009 period. Results show that
639 the AO has a positive correlation with the P80th over China during January to
640 February. Of all 287 stations, 238 stations have positive correlations and 90 stations
641 are positively significant above the 95% confidence. The stations with high positive
642 correlation are mainly located over the central-southern China. The correlation
643 coefficient between the AO and the regional mean P80th over the central-southern
644 China is 0.55, significant at the 99% confidence level.
645 In addition, we compared the AO-related P80th anomalies and the ENSO-related
646 P80th anomalies. The large-scale spatial inconsistency exists between the AO-P80th
647 relationship and the ENSO-P80th relationship. The AO might play a more important
648 role in influencing the P80th over the central-southern China. However, the ENSO’s
649 possible effect on the P80th is confined to the southern coast region of China. The
650 AO-P80th relationship over the central-southern China is similar even when the
651 ENSO-related component is excluded from the original time series.
652 The AO-P80th relationship over the central-southern China is likely related to two
653 factors: MEJS and strong SBT activity. A positive AO phase is accompanied by
654 intensified MEJS and SBT. They enhance the weather disturbances from the West
655 Asia to the BOB. Consequently, a stronger SBT activity may enhance both moisture
656 supply and vertical air motions over the central-southern China. As a result,
30

C1041, doi: 10.3354/cr01041, In press
657 precipitation amount and extreme precipitation events increase over the
658 central-southern China.
659 Acknowledgements .
This research was supported by projects 2007BAC29B02 and
660 2008AA121704. Mao R was supported by the China Postdoctoral Science Foundation
661 under Grant 20090460222. This work is State Key Laboratory of Earth Surface
662 Processes and Resource Ecology contribution No. 2010-ZY-01.
663
664
31

C1041, doi: 10.3354/cr01041, In press
665 LITERATURE CITED
666 Angell JK (2006) Changes in the 300-mb North circumpolar vortex, 1963-2001. J
667 Climate 19: 2984–2994
668 Bao Q, Yang J, Liu Y, Wu G, Wang B (2010) Roles of Anomalous Tibetan Plateau
669
670
Warming on the Severe 2008 Winter Storm in Central-Southern China. Mon Wea
Rev 138: 2375–2384
671 Gao H, Yang S (2009) A severe drought event in northern China in winter 2008-2009
672 and the possible influences of La Nina and Tibetan Plateau. J Geophys Res doi:
10.1029/2009jd012430 673
674 Gimeno L, Torre LDL, Nieto R, García R, Hernández E, Ribera P (2003) Changes in
675
676 the relationship NAO-Northern hemisphere temperature due to solar activity.
Earth and Planetary Sci Lett 206: 15–20
677 Gong DY, Ho CH (2003) Arctic Oscillation signals in East Asian summer monsoon. J
678 Geophys Res doi:10.1029/2002JD002193
679
680
Gong DY, Mao R, Fan YD (2006) East Asian dust storm and weather disturbance:
Possible links to the Arctic Oscillation. Int J Climatol 26: 1379

1396
681 Gong DY, Wang SW (2003) Influence of Arctic Oscillation on winter climate over
682 China. J Geographical Sci 13: 208–216
683 Gong DY, Wang SW, Zhu JH (2001) East Asian winter monsoon and Arctic
684 Oscillation. Geophys Res Lett 28: 2073–2076
685 Gong DY, Wang SW, Zhu JH (2004) Arctic Oscillation influence on daily temperature
686 variance in winter over China. Chinese Sci Bull 49: 637–642
687 Gong DY, Yang J, Kim SJ, Gao YQ, Guo D, Zhou T, Hu M (2011) Spring Arctic
688
689
690
Oscillation-East Asian summer monsoon connection through circulation changes over the western North Pacific, Climate Dyn doi
10.1007:s00382-011-1041-1
32

C1041, doi: 10.3354/cr01041, In press
691 Hoerling MP, Ting M, Kumar A (1995) Zonally flow-stationary wave relationship
692 during El Nino: Implications for seasonal forecasting. J Climate 8: 1838–1852
693 Li J, Wang J (2003) A modified zonal index and its physical sense. Geophys Res Lett
694 doi:10.1029/2003GL017441
695 Mao R, Gong DY, Bao JD, Fan YD (2011a) Possible influence of Arctic Oscillation
696 on dust storm frequency in North China. J Geogr Sci 21: 207–218
697 Mao R, Ho CH, Shao Y, Gong DY, Kim J (2011b) Influence of Arctic Oscillation on
698 Dust Activity over Northeast Asia. Atmos Environ 45: 326–337
699 Nakamura T, Tachibana Y, Honda M, Yamane S (2006) Influence of the northern
700
701 hemisphere annular mode on ENSO by modulating westerly wind bursts.
Geophys Res Lett doi:101029/2005GL025432
702 Suo MQ, Ding Y (2009) The Structures and Evolutions of the Wintertime Southern
703
704
Branch Trough in the Subtropical Westerlies. Chinese J Atmos Sci 33: 425–442
Suo MQ, Ding Y, Wang ZY (2010) Relationship Between Rossby Wave Propagation
705 in Southern Branch of Westerlies and the Formation of the Southern Branch
Trough in Wintertime. J Appl Meterol Sci 19: 731–740 706
707 Quadrelli R, Wallace JM (2002) Dependence of the structure of the northern
708
709 hemisphere annular mode on the polarity of ENSO. Geophys Res Lett doi:101029/2002GL015807
710 Takaya K, Nakamura H (2001) A formulation of a phase-independent wave-activity
711
712 flux for stationary quasigeostrophic eddies on a zonally varying basic flow. J
Atmos Sci 58: 608

627
713 Thompson DWJ, Wallace JM (1998) The Arctic Oscillation signature in the
714
715 wintertime geopotential height and temperature fields. Geophys Res Lett 25:
1297

1300
716 Thompson DWJ, Wallace JM (2001) Regional climate impacts of the Northern
33

C1041, doi: 10.3354/cr01041, In press
717
719
Hemisphere annular mode. Science 293: 85–89
718 Wang J, Ikeda M (2000) Arctic Oscillation and Arctic sea-ice oscillation. Geophys
Res Lett 27: 1287–1290
720 Wen M, Yang S, Kumar A, Zhang PQ (2009) An Analysis of the Large-Scale Climate
721
722
Anomalies Associated with the Snowstorms Affecting China in January 2008.
Mon Wea Rev 137: 1111-1131
723 Wettstein JJ, Mearns LO (2002) The Influence of the North Atlantic–Arctic
724 Oscillation on Mean Variance and Extremes of Temperature in the Northeastern
United States and Canada. J Climate 15: 3586–3600 725
726 Wu BY, Wang J (2002) Winter Arctic Oscillation Siberian High and East Asian winter
727 monsoon. Geophys Res Lett doi: 18101029/ 2002GL015373
728 Yang S, Lau KM, Yoo SH, Kinter JL, Miyakoda K, Ho CH (2004) Upstream
729
730
731
732 subtropical signals preceding the Asian summer monsoon circulation. J Climate
17: 4213–4229
Zhang R, Sumi A (2002) Moisture circulation over East Asia during the El Niño episode in northern winter spring and autumn. J Meteor Soc Japan 80: 213–227
733 Zhang ZY, Gong DY, Hu M, Guo D, He XZ, Lei YN (2009) Anomalous winter
734
735 temperature and precipitation events in southern China. J Geographical Sci 19:
471–488
736 Zhou TJ, Yu RC (2005) Atmospheric water vapor transport associated with typical
737
738 anomalous summer rainfall patterns in China. J Geophys Res doi:101029/2004JD005413
34

C1041, doi: 10.3354/cr01041, In press
743
744
745
746
739
740
LIST OF FIGURES
Figure 1 Time lag correlation of P80th (line with circle) and precipitation amount (line with
741
742 triangle) with AO. The dashed lines are for confidence level.
Figure 2 Correlation between AO and precipitation in January to February. (a) P80th, (b) precipitation amount. The correlation in excess of the 95% confidence level is filled.
Triangles represent negative values and circles represent positive values. In (a), the domain enclosed by solid lines is denoted as the central-southern China in this study.
Figure 3 (a) The time series of AO, P80th , and precipitation amount over the central-southern
747
748
749
750
China during January to February, (b) the time series of the frequency of strong southern branch trough and AO during January to February. All the time series are standardized with respect to the reference period of 1971–2000.
Figure 4 P80th correlations with (a) AO and (b) Niño 3.4 SST. The correlation in
751
752
753 excess of the 95% confidence level is filled. Triangles represent negative values and circles represent positive values. In (a) the correlation is obtained by excluding the ENSO-related components from the time series, and in (b) the correlation is obtained by excluding the AO-related components. 754
755 Figure 5 Regression of variables against one standard deviation of the AO index: (a)
756
757 horizontal winds at 700 hPa level, (b) geopotential height at 500 hPa level, (c) zonal wind at 200 hPa level, (d) water vapor flux from the surface to 300 hPa
758
759
760
761
762
(vectors) and vertical velocity averaged between 700 hPa and 500 hPa (contour lines). In (a) and (d), the largest length of vectors is equivalent to 1.90 m s -1 in
(a) and 27.7 kg m -1 s -1 in (d). In (b, c, d), solid lines are for positive values and dashed lines are for negative values, and value in excess the 95% confidence level is shaded. The unit is m and m s -1 in (b) and (c), respectively, and 0.001 pa s -1 for vertical motion in (d). 763
764 Figure 6 Regression of frequencies of strong meridional wind against one standard
765 deviation of AO index. The climatology of January–February horizontal winds
35

C1041, doi: 10.3354/cr01041, In press
766
767
768
769 at 700 hPa level during 1954 to 2009 are displayed in vectors. For clarity, the vector with its meridional component exceeding 1 m s -1 is shown. The largest vector is equivalent to 20.5 m s -1 . The red and blue shadings represent positive and negative regression coefficient, respectively, and the light, moderate, and
770
771
772 heavy shadings indicate absolute values exceeding 1, 2, and 3, respectively. For instance, at a specific grid, when its climatological meridional wind is southern
(northern) wind, the regression coefficient indicates the anomaly in the frequency of strong southern (northern) wind against one standard deviation of 773
774
776
777
AO index.
775 Figure 7 Composite of daily southern branch trough index during positive AO phase and negative AO phase. The vertical solid line indicates the threshold of strong southern branch trough. The threshold is determined as the value of the 80 th
778
779 percentile of daily southern branch trough index for the reference period of
1971–2000. Note that the curve of probability density function of daily southern branch trough index is smoothed. 780
781 Figure 8 Correlation between AO and synoptic variance in vertical velocity between
782
783
784
785
700 hPa and 500 hPa (a). (b) is the same but for the P80th over the central-southern China. Solid lines are for positive values and dashed lines are for negative values, and values in excess 95% confidence level are shaded. The heavy solid line is to enclose the Tibetan Plateau and the values in this region is
786 removed.
787 Figure 9 Composite difference of wave activity flux and geostrophic streamfunction at
788 200 hPa level between positive AO phase and climate means. Presented are the
789
790
791 horizontal component of wave activity flux with arrows based on geostrophic streamfuction anomalies shaded (positive minus climate means). The selected years for the composite difference are 1990, 1993, and 2002, in which AO
36

C1041, doi: 10.3354/cr01041, In press
792
793
794
795 index are larger than 1 standard value and the absolute value of Nino 3.4 SST are below 0.5 standard value. Unit: m 2 s -1 for geostrophic streamfunction and m 2 s -2 for wave activity flux.
37

C1041, doi: 10.3354/cr01041, In press
796 Table 1 Correlations of precipitation amount, frequencies of extreme precipitation and
797 strong SBT with AO during January to February
P80th
†
P90th
†
Precipitation Amount
†
Strong SBT Frequency
ENSO-excluded series
0.54
a (0.43
a )
0.54
a (0.44
a )
0.50
a (0.38
a )
0.55
a
Original series
0.49
a (0.38
a )
0.47
a (0.38
a )
0.45
a (0.31
a )
0.55
a
798 a: significant at the 99% confidence level
799
†: The correlation out of parentheses is for value averaged over the central-southern
800 China, and in parentheses it is for value averaged over the China
38

C1041, doi: 10.3354/cr01041, In press
801
802 Figure 1 Time lag correlation of P80th (line with circle) and precipitation amount (line
803 with triangle) with AO. The dashed lines are for confidence level.
39

C1041, doi: 10.3354/cr01041, In press
804
805
40

C1041, doi: 10.3354/cr01041, In press
806 Figure 2 Correlation between AO and precipitation in January to February. (a) P80th,
807 (b) precipitation amount. The correlation in excess of the 95% confidence level is
808 filled. Triangles represent negative values and circles represent positive values. In (a),
809 the domain enclosed by solid lines is denoted as the central-southern China in this
810 study.
41

C1041, doi: 10.3354/cr01041, In press
811
812 Figure 3 (a) The time series of AO, P80th, and precipitation amount over the
813 central-southern China during January to February, (b) the time series of the
814 frequency of strong southern branch trough and AO during January to February. All
815 the time series are standardized with respect to the reference period of 1971–2000.
42

C1041, doi: 10.3354/cr01041, In press
816
817
43

C1041, doi: 10.3354/cr01041, In press
818 Figure 4 P80th correlations with (a) AO and (b) Niño 3.4 SST. The correlation in
819 excess of the 95% confidence level is filled. Triangles represent negative values and
820 circles represent positive values. In (a) the correlation is obtained by excluding the
821 ENSO-related components from the time series, and in (b) the correlation is obtained
822 by excluding the AO-related components.
44

C1041, doi: 10.3354/cr01041, In press
823
824
45

C1041, doi: 10.3354/cr01041, In press
825
826
827 Figure 5 Regression of variables against one standard deviation of the AO index: (a)
828 horizontal winds at 700 hPa level, (b) geopotential height at 500 hPa level, (c) zonal
829 wind at 200 hPa level, (d) water vapor flux from the surface to 300 hPa (vectors) and
830 vertical velocity averaged between 700 hPa and 500 hPa (contour lines). In (a) and
831 (d), the largest length of vectors is equivalent to 1.90 m s -1 in (a) and 27.7 kg m -1 s -1 in
832 (d). In (b, c, d), solid lines are for positive values and dashed lines are for negative
46

C1041, doi: 10.3354/cr01041, In press
833 values, and value in excess the 95% confidence level is shaded. The unit is m and m
834 s -1 in (b) and (c), respectively, and 0.001 pa s -1 for vertical motion in (d).
47

C1041, doi: 10.3354/cr01041, In press
835
836 Figure 6 Regression of frequencies of strong meridional wind against one standard
837 deviation of AO index. The climatology of January–February horizontal winds at 700
838 hPa level during 1954 to 2009 are displayed in vectors. For clarity, the vector with its
839 meridional component exceeding 1 m s -1 is shown. The largest vector is equivalent to
840 20.5 m s -1 . The red and blue shadings represent positive and negative regression
841 coefficient, respectively, and the light, moderate, and heavy shadings indicate absolute
842 values exceeding 1, 2, and 3, respectively. For instance, at a specific grid, when its
843 climatological meridional wind is southern (northern) wind, the regression coefficient
844 indicates the anomaly in the frequency of strong southern (northern) wind against one
845 standard deviation of AO index.
48

C1041, doi: 10.3354/cr01041, In press
846
847 Figure 7 Composite of daily southern branch trough index during positive AO phase
848 and negative AO phase. The vertical solid line indicates the threshold of strong
849 southern branch trough. The threshold is determined as the value of the 80 th percentile
850 of daily southern branch trough index for the reference period of 1971–2000. Note
851 that the curve of probability density function of daily southern branch trough index is
852 smoothed.
49

C1041, doi: 10.3354/cr01041, In press
853
854
855 Figure 8 Correlation between AO and synoptic variance in vertical velocity between
856 700 hPa and 500 hPa (a). (b) is the same but for the P80th over the central-southern
857 China. Solid lines are for positive values and dashed lines are for negative values, and
858 values in excess 95% confidence level are shaded. The heavy solid line is to enclose
859 the Tibetan Plateau and the values in this region is removed.
860
50

C1041, doi: 10.3354/cr01041, In press
861
862 Figure 9 Composite difference of wave activity flux and geostrophic streamfunction at
863 200 hPa level between positive AO phase and climate means. Presented are the
864 horizontal component of wave activity flux with arrows based on geostrophic
865 streamfuction anomalies shaded (positive minus climate means). The selected years
866 for the composite difference are 1990, 1993, and 2002, in which AO index are larger
867 than 1 standard value and the absolute value of Nino 3.4 SST are below 0.5 standard
868 value. Unit: m 2 s -1 for geostrophic streamfunction and m 2 s -2 for wave activity flux.
51