HTML
--> --> -->Several mechanisms have been proposed to be responsible for the tripole rainfall anomaly. Some studies have suggested that the drought/flooding conditions in northern/central China during summer are closely linked to an El Ni?o-like sea surface temperature (SST) anomaly (Weng et al., 1999; Lau and Weng, 2001; Xue and Zhao, 2017), stratosphere-troposphere interactions over East Asia (Yu and Zhou, 2007) and, on longer timescales, to the Pacific Decadal Oscillation (Yu, 2013; Yu et al., 2015) and to changes in the distribution of aerosols (Ju and Han, 2013; Wang et al., 2013). As an existing teleconnection pattern in atmospheric circulation, the tripole rainfall pattern over East Asia also shows close linkages with other remote forcings. By stimulating a wave-like pattern, variations in heating over the Tibetan Plateau are also thought to be an important contributor to the tripole summer rainfall pattern over East Asia (Hsu and Liu, 2003).
Regarding atmospheric teleconnection, the Pacific-Japan (PJ) pattern is an important mechanism to explain the existence of the tripole rainfall pattern (Nitta, 1987; Huang and Sun, 1992). Therefore, much effort has been devoted to understanding the mechanisms responsible for the formation of the PJ pattern. A general understanding is that the PJ pattern is forced by anomalous heating associated with anomalous SST in the Philippine Sea (Nitta, 1987; Nitta and Hu, 1996), as well as convective heating over the tropical western North Pacific, which could excite Rossby waves propagating northwards to midlatitudes (Kosaka and Nakamura, 2006). A recent study by (Xie et al., 2009) suggested that a positive SST anomaly in the tropical Indian Ocean associated with an El Ni?o could also trigger the PJ pattern by exciting a warm equatorial tropospheric Kelvin wave that could induce surface Ekman divergence over the tropical western North Pacific. Besides, it has also been suggested that the anomalous thermal conditions associated with Indian Ocean SST could initiate zonal wave activity that propagates from the Mediterranean eastwards to East Asia and results in an anomalous anticyclone over the Okhotsk Sea (Guan and Yamagata, 2003).
The above-cited studies mainly focused on the connection of the PJ pattern or the tripole rainfall pattern over East Asia with lower latitudes (i.e., the Indian Ocean, Philippine Sea, western North Pacific, and tropical eastern Pacific). Some studies have pointed out that there is another forcing from mid- and high latitudes for the PJ pattern and the related rainfall anomaly. Based on both observational and numerical simulation results, (Enomoto et al., 2003) suggested that the Eurasian continent diabatic heating anomaly could stimulate an eastward-propagating Rossby wave, named the Silk Road pattern (Lu et al., 2002), which propagates along the waveguide of the subtropical jet stream. The Silk Road pattern further influences the variability of the Bonin high, which results in the formation of the PJ pattern (Hsu and Lin, 2007).
These studies imply a potential teleconnection between the East Asian summer tripole rainfall pattern and upstream forcing. A much-explored upstream forcing is the variations in Arctic sea-ice cover. (Screen, 2013) focused on the influence of the entire Arctic sea ice on European summer precipitation and found an apparent annular large-scale Rossby wave at midlatitudes, suggesting a potential teleconnection between the Arctic sea ice and East Asian summer rainfall. Some early studies documented that the variability of spring (generally March, April and May) Arctic sea-ice area is significantly related to the summer rainfall anomaly over East Asia via an atmospheric pathway (Zhao et al., 2004; Wu et al., 2009). In contrast, (Guo et al., 2014) suggested that the formation of the East Asian summer tripole rainfall pattern might be attributable to anomalous spring Arctic sea ice via an oceanic pathway. Despite numerous interesting and meaningful results having been revealed in previous studies based on the summer seasonal mean (generally June, July and August), several questions remain unanswered owing to the tripole rainfall pattern showing significant sub-seasonal variability. For example, central China, South Korea and southern Japan suffered severe drought, while northern China experienced above-normal rainfall, in June 2014, and the conditions reversed in August 2014 (Wang and He, 2015). Additionally, it has been revealed that the mechanisms responsible for the rainfall anomaly in June 2014 were different from those in August 2014 (Xu et al., 2016). The fact that the atmospheric circulation and rainfall anomalies during summer differ from month to month (Wang and He, 2015; Xu et al., 2016) raises a new question: is the teleconnection of the monthly rainfall anomaly over East Asia with the Arctic sea ice different from what has been found in previous studies that focused on the summer seasonal mean? This is the main motivation of the present study.
Since this study mainly focuses on the interannual variability, linear trends have been removed from the datasets prior to analysis. To illustrate the teleconnection between the Arctic sea ice and the tripole rainfall pattern, we apply the 3D wave activity flux (WAF; Takaya and Nakamura, 2001). The WAFs are calculated in the quasi-geostrophic framework, which can identify the origin and propagation of the energy of the Rossby wave-like perturbation (Hsu and Lin, 2007). The 3D WAF is defined as follows: \begin{eqnarray} F_x&=&\dfrac{p\cos\varPhi}{2|U|}\Bigg(\dfrac{u}{a^2\cos^2\varPhi}\left[\left(\dfrac{\partial\varPsi'}{\partial\lambda}\right)^2- \varPsi'\dfrac{\partial^2\varPsi'}{\partial\lambda^2}\right]\nonumber\\[-0.5mm] &&+\dfrac{v}{a^2\cos\varPhi}\left[\dfrac{\partial\varPsi'}{\partial\lambda}\dfrac{\partial\varPsi'}{\partial\varPhi}- \varPsi'\dfrac{\partial^2\varPsi'}{\partial\lambda\partial\varPhi}\right]\Bigg) ;\ \ (1)\\[-0.5mm] F_y&=&\dfrac{p\cos\varPhi}{2|U|}\Bigg(\dfrac{u}{a^2\cos^2\varPhi}\left[\dfrac{\partial\varPsi'}{\partial\lambda} \dfrac{\partial\varPsi'}{\partial\varPhi}-\varPsi'\dfrac{\partial^2\varPsi'}{\partial\lambda\partial\varPhi}\right]\nonumber\\[-0.5mm] &&+\dfrac{v}{a^2\cos\varPhi}\left[\left(\dfrac{\partial\varPsi'}{\partial\varPhi}\right)^2- \varPsi'\dfrac{\partial^2\varPsi'}{\partial\varPhi^2}\right]\Bigg) ;\ \ (2)\\[-0.5mm] F_z&=&\dfrac{pf_0^2\cos\varPhi}{2N^2|U|}\Bigg(\dfrac{u}{a^2\cos^2\varPhi}\left[\dfrac{\partial\varPsi'}{\partial\lambda} \dfrac{\partial\varPsi'}{\partial\varPhi}-\varPsi'\dfrac{\partial^2\varPsi'}{\partial\lambda\partial\varPhi}\right]\nonumber\\[-0.5mm] &&+\dfrac{v}{a^2\cos\varPhi}\left[\left(\dfrac{\partial\varPsi'}{\partial\varPhi}\right)^2- \varPsi'\dfrac{\partial^2\varPsi'}{\partial\varPhi^2}\right]\Bigg) .\ \ (3) \end{eqnarray} Here, p is pressure, \(U(=\sqrt{u^2+v^2})\) is the climatological wind speed during 1981-2010; u and v are the zonal and meridional wind components, respectively; \(\varPhi\) and Λ are the latitude and longitude, respectively; a is the Earth's radius; \(\varPsi'\) is the quasi-geostrophic streamfunction, defined as gz/f0, where z is geopotential height, g is gravitational acceleration and f0 is the Coriolis parameter, defined as \(2\Omega\sin\varPhi\), where Ω is the speed of Earth's rotation; and N2 is the Brunt-Vaisala frequency. The Rossby wave source can be defined as -Νχ·?ζ; that is, -?·Νχ (ζ+f) (Sardeshmukh and Hoskins, 1988). Here, Νχ is the divergence wind component, ζ is the absolute vorticity, and f is the Coriolis parameter. The PJ pattern index is defined as the difference in 850-hPa geopotential height between the domains of (30°-40°N, 120°-140°E) and (15°-25°N, 110°-130°E). Based on a previous study (Enomoto et al., 2003), which suggested a potential relationship between the PJ pattern and upstream Rossby waves in August, the present study mainly focuses on the tripole rainfall pattern and atmospheric circulation in August over the period 1980-2014.
Figure1. (a, b) Correlation of Precip-EOF-PC1 with (a) August precipitation and (b) June Arctic sea-ice area. (c) Detrended and normalized Precip-EOF-PC1 and SIAI during 1980-2014. (d-f) Heterogeneous correlation map of the first SVD mode for the detrended and normalized (d) August precipitation and (e) June Arctic sea-ice area during 1980-2014, in which stippled values are significant at the 90% confidence level, and the (f) corresponding SVD time series.To investigate the possible relationships between Arctic sea ice and the tripole rainfall pattern, we first calculate the correlation of Arctic sea ice in March, April and May with respect to the August Precip-EOF-PC1 (figures not shown). It is found that the relationship between the spring (March, April and May) monthly Arctic sea ice and August Precip-EOF-PC1 is very weak (barely statistically significant). There is a statistically significant correlation over the Barents Sea, but the correlation coefficients between the spring monthly area-averaged sea-ice area in the Barents Sea (70°-90°N, 0°-60°E) and August Precip-EOF-PC1 are only 0.32, 0.23 and 0.33, respectively. Since the variations in June are much stronger, and the effects of a varying sea-ice cover much larger due to the high level of incoming shortwave radiation, we expect the effects of the varying Barents Sea ice to be largest in early summer (Matsumura and Yamazaki, 2012; Matsumura et al., 2014). We therefore examine the correlation of Arctic sea ice in June with the August Precip-EOF-PC1, as shown in Fig. 1b. It is apparent that significant positive correlations can be found in the Barents Sea. Corresponding to a decrease in August Precip-EOF-PC1 of one standard deviation, the magnitude of reduced sea-ice area with a statistically significant anomaly is much larger in June (≈ 2.6× 104 km2) than that in March (≈9.5 × 103 km2), April (≈ 1.5 × 103 km2) and May (≈ 1.2× 104 km2). Thus, to represent the reduction in June sea ice, the area-averaged sea-ice area in June over the Barents Sea (70°-90°N, 0°-60°N) is referred to as the sea-ice area index (SIAI) and is depicted in Fig. 1c (solid curve). The correlation coefficient between June SIAI and August Precip-EOF-PC1 is 0.48, significant at the 99% confidence level and substantially higher than those in spring months. It is therefore suggested that there is a potential teleconnection between June Arctic sea ice and the East Asian tripole rainfall pattern in August. To find more evidence for this, we apply singular value decomposition (SVD), which can depict the covariability between two variables, to the standardized June Arctic sea-ice area and August precipitation over East Asia. The results reveal that, when the heterogeneous correlations of August rainfall over East Asia show a tripole pattern (Fig. 1d), significant correlations of June sea-ice area are located in the Barents Sea (Fig. 1e). The corresponding time coefficients of the first SVD component (explains about 29.1% variance) are correlated at 0.62 (Fig. 1f). It should be noted that the results show barely any difference when the signals of El Ni?o-Southern Oscillation (ENSO) in the previous winter are removed (figures not shown). The SVD results further confirm the speculation that the region where the Arctic sea-ice anomaly exerts its most significant influence on the August tripole rainfall pattern is the Barents Sea in June. The following analysis focuses on the June SIAI-related atmospheric anomaly, to explain how the June sea ice in the Barents Sea influences the August rainfall pattern over East Asia.
Figure2. (a, b) Climatology of the (a) 850-hPa wind and (b) vertically (surface to 300 hPa) integrated water vapor transport vector (units: kg m-1 s-1) during August 1980-2014. (c, d) Regression maps of the (c) 850-hPa wind and (d) vertically integrated water vapor transport vector (kg m-1 s-1) anomalies during August 1980-2014 with respect to the preceding June inverted SIAI. Shaded/stippled values in (c, d) are significant at the 90% confidence level based on the Student’s t-test. The black shading indicate the regions where the topographical height is higher than 3500 meters.Generally, southerly (including southeasterly or southwesterly) wind prevails over East Asia and the western North Pacific in August (Fig. 2a). Correspondingly, the vertically integrated water vapor content in this region is transported mainly from the tropical western Pacific and Indian Ocean (Fig. 2b), with larger quantities of water vapor being transported from the Indian Ocean (Wang and Chen, 2012; He, 2015). The 850-hPa wind and vertically integrated water vapor content anomalies in August regressed onto the June inverted SIAI are presented in Figs. 2c and d, respectively. The regression analysis indicates that, corresponding to a reduction in June sea ice, represented by the June inverted SIAI, an anomalous cyclone is present in the following August over the western North Pacific between 10°N and 30°N, accompanied by an anticyclonic anomaly located over South Korea (Fig. 2c), resembling the PJ pattern. It is apparent that a significant northeasterly wind anomaly appears over central eastern China and Japan. Additionally, significant anomalous southwesterly wind emerges over northern China and the western North Pacific. This means that the southerly wind prevailing in August is significantly weakened over central China and Japan, and strengthened over northern China and the western North Pacific, when the June sea-ice area in the Barents Sea is less than normal. Such anomalous atmospheric circulation suggests that the quantity of moisture transported to northern China and the western North Pacific is more than normal, while it is less than normal for central China, South Korea and Japan (Fig. 2d). It is apparent that the moisture conditions in August following a reduction in June sea ice in the Barents Sea is favorable (unfavorable) for rainfall over northern China and the western North Pacific (central China, South Korea and Japan).
In addition to the horizontal moisture conditions, upward motion is essential for the formation of rainfall. Correspondingly, Fig. 3 presents the regression of the vertical wind anomaly in August related to the June inverted SIAI. Since the tripole rainfall pattern extends meridionally, we show a cross section along 115°-135°E, which represents the west-to-east conditions of the tripole rainfall pattern. It is found that significant anomalous ascending, descending and ascending motion occurs at low (south of 20°N), middle (center at 30°N) and high (center at 40°N) latitudes, respectively, along 115°E (Fig. 3a). The anomalous vertical motion in August related to the previous June inverted SIAI also exhibits a meridional tripole structure along 115°E, which is consistent with the rainfall anomaly in this region. The corresponding vertical motion anomaly along 135°E displays a meridional dipole structure, with anomalous ascending motion south of 30°N and anomalous descending motion north of 40°N (Fig. 3b). It is clear that the anomalous descending motion along 135°E is located more northwards than that along 115°E, corresponding to the northeast-southwest elongation of the dipole rainfall pattern (Fig. 1a).
To provide more detail on the anomalous vertical motion related to the June SIAI anomaly, the outgoing longwave radiation (OLR) and total cloud-cover anomalies are shown in Figs. 4a and b. Following a reduction in June sea ice in the Barents Sea, significant positive OLR anomalies extend from Japan southwestwards along the Yangtze River basin in August. Besides, significant negative OLR anomalies appear over the subtropical western North Pacific, as well as southern and northeastern China (Fig. 4a). The positive (negative) OLR anomalies correspond to an anomalous descending (ascending) air mass. Correspondingly, the spatial distribution of the total cloud-cover anomaly also shows a tripole pattern, which is highly similar to that of OLR but with opposite sign (Fig. 4b). Additionally, the 200-hPa zonal wind anomaly displays a tripole pattern, with a statistically significant negative anomaly over (30°-40°N, 60°-150°E) and positive anomaly over (40°-55°N, 110°-150°E) and (15°-25°N, 60°-130°E) (Fig. 4c; shaded). This seesaw pattern between the south and north of the Asian westerly jet axis is closely associated with the Silk Road pattern (Hong and Lu, 2016), implying that the Silk Road pattern plays a key role in connecting the Arctic sea ice with East Asian precipitation.
In summary, the reduction in June sea ice in the Barents Sea can cause an anomalous anticyclone in South Korea and an anomalous cyclone over the western North Pacific in the following August (Fig. 2c). This leads to less moisture over central China, Japan and South Korea, and more moisture over the subtropical western North Pacific, as well as northern and southern China (Fig. 2d). Additionally, significant anomalous ascending, descending and ascending motion emerges at low (around 15°N), middle (around 30°N) and high (around 40°N) latitudes over the East Asian region, respectively, exhibiting a meridional overturning wave pattern (Fig. 3). The corresponding OLR, total cloud cover and 200-hPa zonal wind anomalies show a tripole pattern (Fig. 4). It is apparent that the atmospheric circulation anomalies exhibit a meridionally oriented PJ pattern, and are favorable for the formation of a tripole pattern in East Asian rainfall. The following analysis focuses on the mechanisms bridging a reduction in June sea ice in the Barents Sea with the atmospheric anomalies and the East Asian tripole rainfall pattern in the following August.
Figure3. Regression of the vertical-horizontal cross section for August vertical wind (vectors; units: m s-1) and omega (contours, units: × 10-3 Pa s-1) anomalies along (a) 115°E and (b) 135°E during 1980-2014 onto the preceding June inverted SIAI. Stippled regions indicate omega anomalies significant at the 90% confidence level based on the Student’s t-test.
Figure4. Regression maps of August (a) OLR (shaded; units: W m-2; data available to December 2013 only), (b) total cloud cover (shaded; units: %), and (c) 200-hPa zonal wind (shaded; units: m s-1) onto the June inverted SIAI during 1980-2014. Gridded or dotted regions indicate statistical significance at the 90% confidence level based on the Student’s t-test. Contours in (c) indicate the climatological 200-hPa zonal wind during August 1980-2014.Concurrent with the statistically significant reduction in sea-ice area over the Barents Sea in June (Fig. 5a), there are significant negative (maximum of about -5 W m-2) turbulent heat flux anomalies in situ (Fig. 5c), meaning the ocean obtains net heat fluxes in June that might be caused by a lower albedo and more open water associated with the reduction in sea-ice area. As the sea ice continues to melt, the sea-ice area in the following months becomes even smaller, with a large area of open water (Fig. 5b) and greater absorption of solar radiation. In August, the warmer open water induced by the reduction in sea-ice area starts to influence the atmosphere, which is supported by the inverse sign of turbulent heat flux anomalies, from negative to positive, especially in northern regions (Fig. 5d). This means that the net total turbulent heat flux is upwards, implying that the ocean releases heat to the atmosphere in August. Because of the change in thermal conditions, anomalous upward motion emerges over the Barents Sea, which further triggers meridionally anomalous downward, upward and downward motion, which might be the effect of meridional circulation (e.g., the Ferrel cell and Polar cell) (Fig. 5e). At midlatitudes, significant anomalous upward and downward motion appears alternately along the subtropical jet stream (along 40°N; green line in Fig. 5e). As anomalous vertical motion is generally associated with anomalous divergence, the reduction in June sea-ice area over the Barents Sea may dynamically induce a Rossby wave, because the advection of vorticity by divergence wind can be regarded as a wave source (Sardeshmukh and Hoskins, 1988). As shown in Fig. 5f (shaded), statistically significant anomalous Rossby wave sources are centered near the Mediterranean and Caspian seas, where anomalous upward motion is located (Fig. 5e). Consequently, an apparent Rossby wave, which is evident by inspecting the WAF related to the June SIAI (Fig. 5f; vectors), propagates eastwards along the subtropical jet stream to East Asia (30°-55°N; green frame in Fig. 5f).
Figure5. (a, b) Regression maps of sea-ice area (units: km2) in (a) June and (b) August onto the June inverted SIAI during 1980-2014. (c, d) As in (a, b) but for the total turbulent heat flux (sensible plus latent heat flux; units: W m-2). (e, f) Regression maps of 200-hPa (e) vertical-component wind (units: × 10-3 Pa s-1) and (f) Rossby wave source (shaded; units: × 10-11 s-2) and WAF (vectors; units: m2 s-2) in August onto June inverted SIAI during 1980-2014. Stippled regions indicate values significant at the 90% confidence level based on a two-tailed Student’s t-test. The blue frames in (a-d) indicate where the Barents Sea is located. The blue frame in (e) indicate pathway of the meridional overturning wave pattern related to the SIAI. The blue and green frames in (f) indicate the zonally wave patterns related to the SIAI.
Figure6. (a, d, e) Regression maps of (a) 200-hPa meridional wind (shaded; units: m s-1), (d) vertical-horizontal cross section (outer is 40°N) meridional wind anomalies (shaded; units: m s-1) and WAF (vectors; units: m2 s-2), and (e) 500-hPa geopotential height (contours; units: gpm) and WAF (vectors; units: m2 s-2) in August associated with the preceding June inverted SIAI during 1980-2014. (b, c) Regression maps of August 200-hPa meridional wind (shaded; units: m s-1) onto the time series of the (b) first and (c) second leading EOF mode for the August meridional wind in the domain (30°-60°N, 30°-130°E). (f) 500-hPa geopotential height (contours; units: gpm) and WAF (vectors; units: m2 s-2) in August regressed on the simultaneous PJ index. Regions enclosed by contours in (a-d) and shaded in (e, f) denote anomalies significant at the 90% confidence level based on a two-tailed Student’s t-test.
Figure7. Vertical-horizontal cross section for vertical wind (vectors; units: m s-1) and omega (contours; units: × 10-3 Pa s-1) anomalies along (a) 50°E and (b) 40°N uring August 1980-2014 regressed onto the preceding June inverted SIAI. Dotted regions indicate omega anomalies significant at the 90% confidence level based on the Student’s t-test.We further examine the structure of WAF in August related to the reduction in June sea ice in the Barents Sea. Figure 6a shows the meridional wind anomaly in August related to the June inverted SIAI. An apparent east-west wave disturbance is observed at midlatitudes. Significant positive, negative, positive and negative meridional wind anomalies are found over the Caspian Sea, Balkhash, eastern China and Okhotsk Sea, respectively (Fig. 6a). This anomalous wave-like pattern closely resembles the first leading mode of EOF (EOF1) for the August 200-hPa meridional wind (Fig. 6b; explains about 32.4% of the total variance), which is used to describe the Silk Road pattern (Lu et al., 2002; Ding and Wang, 2005). Additionally, the August wave-like pattern related to the June SIAI shows many similarities with the second leading mode of EOF (EOF2) for the August 200-hPa meridional wind (Fig. 6c; explains about 21.9% of the total variance), especially over East Asia. The corresponding time series of EOF1/EOF2 is correlated with the inverted June SIAI at 0.33/0.33, statistically significant at the 95% confidence level. This means that the zonal Rossby wave at midlatitudes in August is closely related to the June sea-ice variability over the Barents Sea. To better depict the propagation of this east-west wave-like pattern, a vertical cross section of the meridional wind anomaly and associated WAF in August along 40°N is further examined (Fig. 6d). The vertical cross section of meridional wind anomaly clearly exhibits a wave-like pattern, with significant positive (centered at 50°E), negative (90°E), positive (105°E) and negative (150°E) anomalies along the entire meridional domain (Fig. 6d; shaded). The WAF indicates that the propagation of this wave-like pattern is mainly confined to the middle and upper troposphere (Fig. 6d; vectors). Close examination of Fig. 6d reveals that there is a wave source around 50°E (judging from the divergence of WAF), consistent with the results in Fig. 5f. Meanwhile, a significant negative center emerges to the north of the Caspian Sea (Fig. 6a). Such a north-south dipole (along 50°E) supports the speculation that the east-west wave-like pattern at midlatitudes (along 40°N) might originate from high latitudes (e.g., the Arctic). The anomalous vertical motion in August further supports this view (Fig. 7). Corresponding to a significant reduction in sea ice in the Barents Sea in June, significant anomalous upward motion, possibly due to strong near-surface thermal forcing, appears over the Arctic region (70°-80°N; Fig. 7a) in August, where the anomalous sea ice is observed. The anomalous motion ascends upwards into the upper troposphere (at around 200 hPa), then bends equatorwards to midlatitudes (40°-50°N), and finally descends downwards. Such atmospheric perturbation further propagates eastwards along the midlatitudes. As shown in Fig. 7b, apparent anomalous descending motion appears at around 45°E, consistent with Fig. 7a. At the same time, anomalous ascending and descending motion appear alternately at 60°E, 90°E, 120°E, 140°E and 160°E. The results revealed by Fig. 7 are consistent with that in Fig. 5e. It is therefore suggested that anomalous thermal conditions related to the reduction in sea ice over the Barents Sea could induce anomalous vertical motion in situ from June to August, driving the meridional overturning and resulting in anomalous vertical motion at midlatitudes in its southern branch, which further induces an eastward propagation of a Rossby wave along the subtropical westerly jet. Consequently, the east-west wave-like atmospheric perturbation at midlatitudes, which resembles the Silk Road pattern, is established. The propagation of the Silk Road pattern is favorable for the variation in the Bonin high (Fig. 2a), which is a subtropical anticyclone observed over Japan (Enomoto, 2004). As suggested earlier, the Bonin high might be related to the descending branch of the Hadley cell. Here, we show that the anomalous August Bonin high associated with the reduction in June sea ice in the Barents Sea could trigger a meridional overturning, leading to the formation of the PJ pattern (Enomoto et al., 2003; Hsu and Lin, 2007). As shown in Fig. 6e, the spatial distribution of 500-hPa geopotential height anomalies and WAF in August related to the June inverted SIAI closely resembles its counterpart related to the PJ pattern index (Fig. 6f). The correlation coefficient between the June inverted SIAI and August PJ pattern index is 0.55, statistically significant at the 99% confidence level. Such an interpretation is illustrated schematically in Fig. 8.
Figure8. Schematic diagrams summarizing the dynamical linkage between the reduction in June sea ice in the Barents Sea with the Silk Road pattern as well as the Pacific-Japan pattern. The cross sections for (a) and (b) are taken from Figs. 7a and b, respectively. Blue and red arrows denote zonal and meridional overturning, respectively.Finally, to provide more robust evidence for the interpretation of the teleconnection between the reduction in June sea ice in the Barents Sea and the downstream atmospheric perturbation in August, we conduct two numerical experiments——a control experiment and a sensitivity experiment——to examine the atmospheric response to the June sea-ice variability in the Barents Sea. We use version 5 of the Community Atmosphere Model (CAM5), with a 1.9° × 2.5° finite volume grid, and with 26 hybrid sigma pressure levels. The sea-ice concentration and SST are prescribed as boundary conditions in the model; all other external variables are fixed. First, we run a control experiment for 25 years forced by seasonal-varying climatology (1979-2000) of sea ice. Then, we perform a pair of 12-month sensitivity experiments from January to December. The sea ice is reduced in June over the Barents Sea (70°-90°N, 0°-60°E) (as shown in Fig. 5a), while other months are prescribed by climatological sea ice. The experiment is repeated 20 times, with different initial conditions on 1 January, taken from the 6th to 25th model year. The SSTs in both experiments are set to their climatological monthly values. We focus on the difference between the sensitivity and control experiments in August to reveal any lagged response of the atmosphere to the reduction in June sea ice.
Model-simulated differences in 850-hPa wind and 200-hPa zonal wind between the sensitivity and control experiments are displayed in Fig. 9. The 850-hPa wind anomaly is clearly characterized by a significant anomalous anticyclone over South Korea and a dominant anomalous cyclone over the subtropical western North Pacific (Fig. 9a), which is highly consistent with the observational results (Fig. 2c). The response of August 200-hPa zonal wind shows apparent change in the meridional shear of zonal wind, with significant positive and negative anomalies over (45°-50°N, 110°-150°E) and (25°-35°N, 110°-150°E), respectively (Fig. 9b), resembling its observational counterpart (Fig. 4c). Such a distribution of 200-hPa zonal wind over East Asia will lead to a south-north tripole rainfall anomaly (Guo et al., 2014), as shown in Fig. 1a.
Based on the observational results, we suggest that the reduction in June sea ice in the Barents Sea may first trigger an east-west wave-like pattern at midlatitudes (i.e., the Silk Road pattern), and then induce a meridional overturning wave-like pattern over the East Asia-Pacific region (i.e., PJ pattern). Figure 10a shows the model-simulated difference in 200-hPa meridional winds in August. Clear negative and positive anomaly centers appear alternately at 60°E, 90°E, 120°E and 150°E, roughly along 40°N, which displays an apparent wave-like pattern (Figs. 10a and b). Note that there is a dipole pattern over (30°-60°N, 30°-90°E; Fig. 10a), supporting the existence of the southward-propagating wave pattern described by observations (Figs. 6a and 7a). When the east-west wave pattern is established, a south-north wave pattern emerges over East Asia and the western North Pacific, as estimated from the 500-hPa geopotential difference (Fig. 10c; contours). The WAF clearly suggests that this wave-like pattern propagates from the subtropical western North Pacific northwards to East Asia (Fig. 10c; vectors), resembling its observational counterpart well (Fig. 6e).
Figure9. The difference in (a) 850-hPa wind (units: m s-1) and (b) 200-hPa zonal wind (m s-1) anomalies in August between the sensitivity and control experiment in CAM5. Values shaded or enclosed by grids are significant at the 90% confidence level based on the Student’s t-test.
Figure10. The difference in the (a) 200-hPa meridional wind (units: m s-1), (b) vertical-horizontal cross section (outer is 40°N) of meridional wind anomalies (shaded; units: m s-1) and WAF (vectors; units: m2 s-2), and (c) 500-hPa geopotential height (contours; units: gpm) and WAF (vectors; units: m2 s-2) in August between the sensitivity and control experiment in AM5. Regions enclosed by white contours in (a, b) and shaded in (c) denote anomalies significant at the 90% confidence level based on a two-tailed Student’s t-test.It is found that a reduction in June sea ice over the Barents Sea is accompanied by an anomalous anticyclone over South Korea and anomalous cyclone over the subtropical western North Pacific in the following August, which results in a decrease in moisture transport to central China, South Korea and Japan, and an increase in moisture to northern and southern China as well as the subtropical western North Pacific. The anomalous moisture conditions favorable for the formation of the tripole rainfall pattern are therefore provided. Additionally, corresponding to a reduction in June sea ice in the Barents Sea, the vertical convection over East Asia and western North Pacific in August also shows significant anomalies. Along the section of 115°E, significant anomalous ascending, descending and ascending motion is observed around 20°N, 30°N and 40°N, respectively. Such anomalous vertical motion is further supported by the tripole pattern of the OLR (total cloud cover) anomaly, with s significant negative (positive), positive (negative) and negative (positive) anomaly in northern China, central China to Japan, and southern China to the western North Pacific, respectively. It is thus suggested that the anomalous dynamical conditions related to the reduction in sea ice in the Barents Sea are favorable for the East Asian tripole rainfall pattern.
Further analyses indicate that the near-surface anomalous thermal conditions associated with a reduction in sea ice tends to excite anomalous upward motion due to strong near-surface thermal forcing over the Barents Sea region, which triggers a meridional overturning extending to midlatitudes in the form of a wave train. The wave train propagates southwards and forms anomalous downward motion over the Caspian Sea, which further induces a zonally oriented overturning circulation, exhibiting an east-west wave train (i.e., Silk Road pattern). The mechanism of southward propagation, which might be related to the effect of meridional circulation (e.g., the Ferrel cell and Polar cell), is still unclear. The Silk Road pattern propagates eastwards along the subtropical jet stream to East Asia and causes an anomalous anticyclone over South Korea (i.e., the Bonin high). As the Bonin high might be related to the descending branch of the Hadley cell (Enomoto et al., 2003), the anomalous Bonin high ultimately triggers a meridional overturning and a south-north wave-like pattern is formed——the so-called PJ pattern. Consequently, the tripole rainfall pattern emerges over East Asia.
To further explore the observational findings, numerical experiments using CAM5 are conducted. The results support the conclusion that the atmospheric anomalies over East Asia are related to the reduction in June sea ice in the Barents Sea. The perturbed experiment integrated with prescribed sea-ice loss in June over the Barents Sea successfully reproduces a respective pair of anticyclonic and cyclonic anomalies over East Asia. Meanwhile, significant meridional shear of 200-hPa zonal wind that favors the tripole rainfall pattern over East Asia, with significant positive and negative anomalies over (45°-50°N, 110°-150°E) and (25°-35°N, 110°-150°E), respectively, is also produced by the model. More importantly, the Silk Road pattern and PJ pattern are both reproduced in our experiment. Therefore, analysis based on both observations and numerical simulations suggests that the reduction in June sea ice in the Barents Sea tends to induce a tripole rainfall pattern over East Asia in August by modulating the atmospheric anomaly via a triggering of the Silk Road pattern and PJ pattern.
The variability of Arctic sea ice could explain 25%-36% of the total variance in East Asian precipitation (Figs. 1c and f), showing that there are also other types of forcing for the climate variability in this region. Most previous studies have linked the tripole rainfall pattern over East Asia to tropical forcing, such as ENSO. The 2015/16 winter featured a super El Ni?o event in the eastern tropical Pacific of almost the same strength as the one that occurred in winter 1997/98. However, the rainfall anomaly in August 2016 was nearly opposite to that in August 1998, suggesting a complex set of climatic systems and the need to comprehensively consider the effects of different forcings. Our results might shed more light on understanding the diversity of East Asian precipitation. The effect of sea ice over the Barents Sea should be taken into account in ENSO-based predictions of East Asian summer rainfall anomalies. Additionally, statistically significant correlations between the East Asian tripole rainfall pattern and Arctic sea-ice area are found over the Chukchi Sea (Figs. 1c and f), which might be a result of the teleconnection related to East Asian monsoonal rainfall (Grunseich and Wang, 2016). Besides, although the model results in the present study are largely consistent with the observations, they are obtained from only one climate model; further results from multiple models are therefore needed. This is especially so given that, despite the model's simulation supporting the chain of events indicated by the observations, several differences are still apparent. For example, the observed PJ pattern related to the reduction in June sea-ice area over the Barents Sea originates from the subtropical western North Pacific, propagates northwards, and turns eastwards over Japan (Fig. 6e). By contrast, the simulated PJ pattern is dominated by northward propagation (Fig. 10c). This might be induced by the stronger northward shift of the Asian jet stream simulated by the model (Fig. 9b versus Fig. 4c). Meanwhile, the observational 200-hPa zonal wind anomaly shows a comparable dipole pattern in the west (60°-90°E) and east (120°-150°E) of the Eurasian continent (Fig. 4c; shaded), which is closely related to the Silk Road pattern (Hong and Lu, 2016). However, only one dipole anomaly of 200-hPa zonal wind appears over East Asia in the model simulation (Fig. 9b). The inconsistency between the model and reanalysis data might be attributable to topographical effects, as shown in Fig. 9a. Unfortunately, a deeper dynamical and physical explanation regarding the model's performance remains unclear.
