-
The July–August precipitation anomalies in 2015 are shown in Fig. 1a. A distinct negative precipitation anomaly existed in northern China, including Hebei, Shaanxi, and most places in Mongolia Province, while a large-scale positive precipitation anomaly occurred in southern China. These anomalies were defined with a 30-yr (1981–2010) baseline as the mean climatology. The anomalous precipitation shows a pattern similar to the SF-ND pattern in China (Yatagai and Yasunari, 1994; Nitta and Hu, 1996; Zhou et al., 2009; Li et al., 2010).
Figure 1. (a) Precipitation anomaly (units: mm) during July–August 2015 relative to the climatology of 1981–2010. (b) Precipitation index 1 (PI1) and (c) precipitation index 2 (PI2), normalized time series of area-weighted mean precipitation in the subregions (31.5°–48°N, 101.5°–130°E) and (17°–28.5°N, 101.5°–130°E), respectively, during July–August 1979–2016. (d) Precipitation index (PI) during July–August 1979–2016 calculated by PI1 and PI2 as in Eq. (3). A positive value of PI indicates the occurrence of the SF-ND pattern in China. Note that these indices are not detrended.
Then, we defined a precipitation index (PI) to describe the intensity of this dipole precipitation mode, as follows:
where PI1 and PI2 are defined as the regionally weighted normalized July–August precipitation in the subregions indicated in Fig. 1a. A minimum value occurs in 2015 in the PI1 sequence (Fig. 1b), suggesting that the index has a good capacity to represent regional dry‒wet conditions. Figure 1d also displays a similar conclusion: a marked positive value often accompanies an obvious SF-ND event.
Using the PI values, we selected years greater than 0.5 standard deviations as positive-phase years (1994, 1997, 1999, 2001, 2002, 2006, 2014, 2015, 2016). Likewise, we also had negative-phase years (1981, 1982, 1983, 1985, 1987, 1990, 1993, 1998, 2003, 2010, 2011, 2012), with the positive phase showing the SF-ND precipitation mode, and vice versa. Using these selected years, composite maps of precipitation anomalies were drawn (Figs. 2a and b), both of which exhibit obvious dipole precipitation patterns.
Figure 2. The dipole precipitation anomaly (units: mm) in July–August during (a) positive-phase years and (b) negative-phase years. Areas with slashes are significant at the 99% confidence level based on Student’s t-test. (c) Regression maps of July–August 850-hPa wind anomalies with regard to PI during 1979–2016. Light (dark) shading indicates that the values are significant at the 90% (95%) confidence level based on Student’s t-test. (d) The 850-hPa wind anomaly (vectors; units: m s−1) for July–August 2015. The colour shading indicates the anomalous magnitude of wind. The letters A and C indicate the anomalous anticyclone and cyclone, respectively.
Here, we analyzed anomalies in atmospheric circulations in 2015 and their connection with rainfall anomalies in the period 1979–2016 to understand the causes of this extreme precipitation anomaly, particularly in northern China.
As shown in Figs. 3c and d, the strength of the EASM was weak, and the position of the WPSH was shifted southward. Each system made it difficult for water vapor to be transported from the ocean to northern China. Therefore, what factors caused the abnormal condition of these two systems?
Figure 3. (a) Regression maps of summer integrated water vapor flux divergence from 300 hPa to 1000 hPa with regard to simultaneous PI during 1979–2015. The dotted area passed the 90% significance test based on Student’s t-test. (b) Anomalous summer integrated water vapor flux divergence (units: g kg−1 m s−1) in 2015 (relative to the climatology of 1981–2010). (c) EASM index and (d) Northern Hemisphere Subtropical High Ridge Position Index during 1979–2015.
The wind anomalies at 850 hPa in July–August 2015 are shown in Fig. 2d. Near the Ural Mountains, a strong blocking system appeared, which was closely related to a negative EU phase. The regressed velocity at 850 hPa against the PI is shown in Fig. 2c. It is likely that a meridional circulation mode (the PJ/EAP) occurred from Japan to the northwestern Pacific, which could have had a close connection with this anomalous precipitation pattern in China. As shown in Figs. 3a and b, anomalous water vapor transport centers appeared in southern China. A similarly blocked circulation also occurred in northern China.
When the PJ/EAP mode is in the positive phase, there is a cyclonic anomaly over southern China, corresponding to ascending air masses. Therefore, substantial water vapor causes precipitation to concentrate here. Similarly, this circulation pattern can effectively prevent the northward movement of the WPSH. The abnormal anticyclone over northeastern China, corresponding to the descending air masses, made it difficult for water vapor to be transported to northern China. Ultimately, this pattern promoted the occurrence of the abnormal precipitation SF-ND mode.
Moreover, meridional circulation occurred at high latitudes and presented a significant anticyclonic anomaly on the eastern side of Scandinavia, with a cyclonic anomaly near the Okhotsk Sea. This meridional circulation pattern could also have inhibited the northwards movement of water vapor. The EU pattern is a deep quasi-positive pressure system that can be of influence from 500 hPa down to 700 hPa (Wallace and Gutzler, 1981). Since its impact can reach the height of the lower troposphere, it would have affected the northward direction of the high-pressure ridge and prevented the southern water vapor from being transported to the north.
-
To explore the circulation anomalies corresponding to the precipitation anomalies, the 500-hPa geopotential height anomalies are shown in Fig. 4a. Then, these 500-hPa geopotential heights were regressed with regard to the PI, as shown in Fig. 4b. An obvious negative EU phase appeared, with an obvious anticyclonic circulation pattern over Mongolia. As a result, abnormal northerly winds prevailed in northern China, reducing water vapor transport, which was not conducive to precipitation.
Figure 4. Anomalous (a) geopotential height (units: gpm) at 500 hPa and (c) meridional winds (units: m s−1) at 250 hPa in July–August 2015. The climate average was calculated from 1981 to 2010. Regression maps of summer (b) 500-hPa geopotential anomalies and (d) 250-hPa meridional winds with regard to concurrent PI during 1979–2016. Light (dark) shading in (b) indicates that the values are significant at the 90% (95%) confidence level, and the dotted area in (d) passes the 95% significance test based on Student’s t-test.
At 250 hPa, a significant zonal circulation mode occurred at 60°N, which agrees perfectly with the BBC pattern proposed by Xu et al. (2019a). The BBC pattern propagates along polar front jets. Defined at 250 hPa, four geographic centres are included in this mode; these are located to the west of the British Isles, the Baltic Sea, western Siberia, and Lake Baikal. Xu et al. (2019a) performed a regression between precipitation and the BBC index by using Climate Research Unit data. The result was that the BBC could obviously influence the precipitation at high latitudes, including northern China. Additionally, similar precipitation centers occurred that corresponded to the four meridional circulation centers. The northern regions of eastern China tend to be wet when the BBC is in a positive phase. Generally, the SR or circumglobal teleconnection modes propagating along the subtropical jet stream in the upper atmosphere are believed to be responsible for precipitation anomalies in China (Wang and He, 2015; Hong et al., 2018). A meridional circulation similar to SR can be seen near 30°N in Fig. 4c. Its intensity is slightly less than that of the BBC mode. In Table 1, we calculated the correlation between the different indices. As can be seen, the BBC affects precipitation in northern China and southern China equally, while the SR mainly affects precipitation in northern China. The negative correlation coefficient between SR and PI1 is weak. The correlation between the BBC and PI1 shows a positive relationship, but a negative relationship between the BBC and PI2 was obtained. Therefore, if the BBC is in a negative phase while the SR is in a positive phase, an SF-ND event is very likely to occur. The relationship between the two could form this dipole precipitation pattern.
EU pattern PJ/EAP pattern BBC pattern SR pattern EU pattern – – – – PJ/EAP pattern −0.540 – – – BBC pattern −0.780 0.562 – – SR pattern −0.021 0.264 −0.223 – EASM 0.398 −0.180 −0.112 0.090 WPSH 0.363 −0.348 −0.167 −0.163 PI1 0.576 −0.603 0.586 −0.252 PI2 −0.509 0.507 −0.423 −0.021 PI −0.646 0.662 −0.604 0.146 Table 1. Correlation coefficients between indices. Information on how these indices were calculated can be found in section 2. The 99%, 95%, and 90% confidence levels are ±0.403, ±0.312, ±0.264, respectively, according to Student’s t-test. Bold values passed the 90% confidence levels. All indices were detrended.
-
In the previous analyses, key circulation factors that possibly affected the anomalous precipitation mode were discussed. Therefore, the next question is what circulation factors caused these anomalies. Further analysis of the physical mechanisms of the circulation anomalies is not only meaningful for the typical precipitation pattern in 2015, but also plays an important role in predicting future droughts and rainfall in this region.
2015/16 was a typical super El Niño year. The wide range of strong SSTAs had a large impact on atmospheric circulations and precipitation (Zhai et al., 2016; Paek et al., 2017; Ma et al., 2018). Summer drought in northern China is likely related to El Niño. Table 2 shows the correlations between different characteristics of the WPSH and the El Niño index. It can be found that, during the year that El Niño develops, the WPSH often strengthens and increases in area, but the location is southerly. This anomaly leads to water vapor being concentrated in southern China and impedes its transport to northern China.
WPSH ridge position WPSH intensity WPSH area Niño3_winter −0.576 0.533 0.661 Niño3_spring −0.470 0.443 0.578 Niño3_summer −0.293 −0.082 −0.101 Table 2. Correlation coefficients between Niño3 index and the WPSH. The 99%, 95% and 90% confidence levels are ±0.403, ±0.312 and ±0.264, respectively, according to Student’s t-test. Bold values passed the 90% confidence levels.
In addition to affecting the location and intensity of the WPSH, El Niño also has a significant impact on convection in the regions influenced by the sinking branch of the Walker circulation. In particular, enhanced convection over the central and eastern Pacific induces a Rossby wave response that affects the midlatitude circulation and precipitation patterns. The convection over Indonesia and the western Pacific also alters the monsoon systems and the tropical cyclone activity in the region. The changes in the Walker circulation also affect the ocean–atmosphere coupling and the interannual variability of the climate system (Weng et al., 2007; Yuan et al., 2012, 2017; Wen et al., 2020).
Here, we calculated the July–August mean index values for the different types of El Niño in different years. The years of EP El Niño only are provided here (1965, 1972, 1976, 1997, 2009, 2014 , 2015). The summer of 2015 was a significant El Niño development period (the July–August mean index value is 1.765) that may have had a huge impact on the large-scale circulation.
Figure 5b illustrates the tropical atmospheric circulation patterns in EP El Niño years. In summer, a large-scale Walker circulation anomaly dominates the eastern Pacific Ocean, with strong upward motion over the central and eastern Pacific and significant downward motion over the western Pacific (Weng et al., 2007). The broad ascending motion from 160°W to 80°W splits into two branches at 120°W: one branch turns eastward in the upper troposphere, associated with the Walker circulation, and the other branch extends westward along the equator. Weak descending motion occurs over the tropical Indian Ocean at 60°E, which eventually joins the main subsidence over Indonesia. Although the 2015 tropical vertical circulation (Fig. 5a) was more westward than the ascending branch of a typical EP El Niño year, and its subsidence branch was stronger, it still shows similar characteristics. In June 2015 (Fig. 5c), the OLR negative anomaly center occupied central and southern parts of southern China, showing significant convective activity in these regions. However, in July and August (Fig. 5d), there was no obvious OLR negative anomaly center and descending motion prevailed in southern China. Therefore, it can be considered that the strong precipitation in the south was mainly affected by the large-scale circulation, rather than regional convection. The position of the descending motion corresponded to the intensity and position of the WPSH, which made it difficult for the precipitation in northern China to be affected because the position of the WPSH was limited to the tropical western Pacific. The OLR anomaly of the composite EP El Niño years also supports the above view (Fig. 5e), as due to the influence of the WPSH, the rain band is located in southern China.
Figure 5. (a) Vertical–horizontal cross section averaged along 5°S–5°N for vertical velocity anomalies (units: m s−1) during July–August 2015 (relative to the climatology of 1981–2010). (b) Composite Walker Circulation anomalies over the equator (5°S–5°N) (units: m s−1) in association with the summer EP El Niño. The vertical velocities are magnified 100 times. (c, d) Anomalies of OLR (units: W m−2) in (c) June and (d) July–August 2015 relative to the climatology of 1981–2010. (e) Composite maps of OLR anomalies in July–August with the summer EP El Niño.
-
The PJ/EAP pattern can link SSTAs in the western Pacific with summer precipitation in East Asia. When the PJ/EAP pattern tends to appear in the positive phase, the EASM is weak (Huang and Li, 1988; Chang et al., 2000; Xie et al., 2016). However, consensus has not been reached on the formation and maintenance mechanism of the PJ/EAP. Previous studies can be summarized into three views. The first view is that the PJ/EAP mode is a Rossby wave stimulated by convective activity near the Philippines, and the change in the PJ/EAP mode is driven by SSTAs (Nitta, 1987; Huang and Li, 1988; Xie et al., 2009, 2016; Kosaka and Xie, 2013). The second view is that the PJ/EAP mode is an intrinsic atmospheric mode inherent to the complex fundamental flow in East Asia (Kosaka and Nakamura, 2010). The third view suggests that the midlatitude PJ/EAP can be affected by latent heat released by the mei-yu precipitation. The latent heat acts as a bridge to transfer the western Pacific convective changes to the PJ/EAP (Lu and Lin, 2009). Although there is debate regarding the way that SST affects the PJ/EAP, it is not difficult to find that temperature anomalies, especially in the tropical Indian Ocean, are often an important factor in determining variations in the PJ/EAP by combining the above points. We applied the SVD method to examine the covariability between the northern tropical Indian Ocean and East Asia. The results revealed that a positive SST anomaly in the Indian Ocean is associated with a tripole “anticyclone–cyclone–anticyclone” pattern of anomalies over East Asia, which resembles the PJ/EAP mode (Figs. 6c and d).
Figure 6. (a, b) Heterogeneous correlation map of the first mode of the SVD for the detrended and normalized (a) 850-hPa geopotential height and (b) SST during July–August 1979–2016. (c) The corresponding time series. (d) SSTAs in summer 2015 (units: K), with a climate baseline of 1981–2010.
In previous studies (Hirota and Takahashi, 2012; Xu et al., 2019b), the years in which the PJ/EAP phase was positive were associated with significantly high precipitation in southern China and corresponded to the state of El Niño the following year. In a year when EP El Niño develops, the tropical Pacific Ocean becomes a strong heat source that affects the global climate. It has been widely documented that the SST in the tropical Indian Ocean exhibits a warming tendency in the post-El Niño season (Weare, 1979; Nigam and Shen, 1993; Liu and Alexander, 2007; Schott et al., 2009; Xie et al., 2016). The lagged correlation coefficient between the El Niño signal in the preceding winter and the heat source anomaly in the northern tropical Indian Ocean in summer reaches 0.7 (figure not shown; Xie et al., 2016). Therefore, Indian Ocean warming is often considered as a response to El Niño (Fig. 6a). This view is also consistent with Figs. 5c and d, which show that atmospheric convection is suppressed rather than enhanced over the warming tropical Indian Ocean during the developing and mature phases of El Niño. El Niño influences the Indian Ocean temperature through the capacitor charging and discharging effect (Yang et al., 2007; Xie et al., 2009; Chowdary et al., 2019; Na and Lu, 2023), and the SST anomaly in the northern Indian Ocean acts as a wave source that triggers the PJ/EAP. It can excite Kelvin waves that propagate eastward along the equator, and an anomalous anticyclone forms near the Philippine Sea. Furthermore, a tripole “anticyclone–cyclone–anticyclone” pattern of anomalies over East Asia occurs (Xie et al., 2016).
-
Li and Leung (2013) suggested that the sea-ice changes in spring may have a stronger influence on China’s summer precipitation than the sea-ice changes in summer, as Arctic warming is more pronounced in spring. They also indicated that changes in spring and summer sea ice can affect the atmospheric teleconnection patterns in the mid and high latitudes of Eurasia in summer. Previous studies have shown that the EU pattern over Eurasia in summer modulates the distribution of the East Asian subtropical jet and the westerly belt (Gao et al., 2017), and the changes in these jets influence the northward propagation of the jet stream. Arctic sea ice has undergone a significant decline since the 1980s. Therefore, the sea-ice change in the Arctic in early spring plays an important role in modulating the teleconnection patterns of the mid and high latitudes in Eurasia and the East Asian climate in late summer.
As mentioned in the Introduction, the sea-ice index of the Barents Sea in spring was denoted by the SIC in the region (70°–82°N, 0°–55°E), which we hereafter refer to simply as the SIC index. The distribution and correlation coefficient of the normalized SIC and EU indexes are displayed in Fig. 7a. The correlation coefficient is 0.46, which suggests that a negative phase of the EU pattern that influences the Chinese region is likely to occur when the sea ice in the Barents Sea is relatively low. The heat flux anomaly of the underlying surface may be affected by substantial changes in Arctic sea ice. The upward turbulent heat flux in regions with less sea ice in the Barents Sea in spring is significantly stronger, and a large amount of heat is transferred from the ocean to the atmosphere. Persistent anomalous heat flux over the underlying surface in spring and summer can provide energy for the development of the summer EU pattern.
Figure 7. (a) Averaged EU index in July–August, SIC in spring, and snow depth (SD) within (55°–70°N, 30°–60°E) in April and May 1979–2015. (b) Snow cover with regard to SIC values during 1979–2016. The dotted area passes the 95% significance test based on Student’s t-test. (c, d) Heterogeneous correlation map of the first mode of the SVD for the detrended and normalized (c) 500-hPa geopotential height during July–August and (d) ICEC (sea-ice concentration) anomalies during spring 1979–2016. (e) The corresponding time series. (f) ICEC in MAM of 2015, with a climate baseline of 1981–2010.
The mid–high latitude circulation may be affected by the lack of Arctic sea ice not only through direct dynamic processes but also through indirect thermodynamic processes involving the snow cover in the Eurasian mid–high latitude region. Previous studies (Liu and Yanni, 2002; Bader et al., 2011; Li et al., 2018) have indicated that sea-ice changes are closely related to snow cover changes in Eurasia, and snow cover can further affect the Eurasian climate by altering the albedo and hydrological conditions. Figure 7b shows the composite snow cover in April and May and SIC index in spring. As can be seen, there is a significant positive correlation between sea ice and snow in Siberia; that is, in years with less sea ice, there is less snow here. The anomalous signal of sea-ice reduction in the Barents Sea in spring can persist until summer, leading to higher air temperatures over the Barents Sea. This favours upward motion and forms an anomalous low pressure in the upper troposphere. This low pressure further stimulates the southward wave train and generates the EU pattern. A low-pressure anomaly emerges over the North Pole because of the reduction in snow cover over the Barents Sea and Eurasia in the previous spring, which stimulates a wave train to propagate from the North Pole to the northeast region, and a high-pressure anomaly occurs in northern China. This anticyclonic circulation anomaly is unfavorable for the accumulation of water vapor in northern China.
In addition to the contributions of the EU and PJ/EAP discussed above, the BBC and SR teleconnection modes also have an important impact on the precipitation modes in China. Among these, the negative phase of the BBC (shown in Fig. 4d) has a good correlation with the precipitation anomalies in the southern and northern parts of China, while the SR type mainly promotes or weakens the precipitation in northern China (shown in Table 1). The BBC propagates within the polar front jet (Xu et al., 2019a). The SR mode originates near the Caspian Sea and the Mediterranean Sea, and then propagates along the subtropical westerly jet (Lu et al., 2002; Enomoto et al., 2003).
Here, we used the Plumb wave flux method to analyze the wave activity flux anomaly in 2015. A remarkable heat source emerged on the continent near the Caspian Sea in 2015 (Fig. 8b). This heat source induced an anomalous anticyclonic circulation in the upper troposphere near the Caspian Sea. This region is also situated at the entrance of the Asian westerly jet stream, so the strong warming and the associated low-level convergence and high-level divergence can stimulate Rossby waves (Enomoto et al., 2003; Sun et al., 2008) that propagate along the jet stream (i.e., the SR pattern). Figures 8a and b illustrate the impact of this heat source and the wave activity flux at 60°E. Consequently, a positive geopotential height anomaly occurred over high-latitude Asia and a negative anomaly over midlatitude East Asia.
Figure 8. (a) Geopotential height (shading; units: gpm) and wave activity flux (vectors; units: m2 s−2) anomalies at 250 hPa in July–August 2015. (b) Anomalous air temperature at 1000 hPa (units: K) and winds (units: m s−1) in July–August 2015. (c) SSTAs (units: K) in July–August 2015 relative to the 1981–2010 climatology.
The initial perturbation source over the North Atlantic influences the ducting of the polar jet and the development and propagation of teleconnection patterns (Hall et al., 2015; Lee et al., 2019). A previous study proposed that barotropic instability and the interaction between circulation anomalies in the outlet region of the Atlantic jet and high-frequency synoptic-scale transients may stimulate the BBC (Xu et al., 2019a). The BBC and EU patterns have overlapping positions in northern China during their westward propagation, so they can modulate the precipitation here by enhancing or reducing each other’s intensity. For instance, when the BBC is a cyclonic anomaly and the EU is a west-low east-high anomaly, they will increase the precipitation in northern China; on the contrary, when the BBC is an anticyclonic anomaly and the EU mode is a west-high east-low anomaly, they will decrease the precipitation in northern China. Therefore, this system can affect the precipitation in northern China by strengthening/weakening the intensity of the EU.
EU pattern | PJ/EAP pattern | BBC pattern | SR pattern | |
EU pattern | – | – | – | – |
PJ/EAP pattern | −0.540 | – | – | – |
BBC pattern | −0.780 | 0.562 | – | – |
SR pattern | −0.021 | 0.264 | −0.223 | – |
EASM | 0.398 | −0.180 | −0.112 | 0.090 |
WPSH | 0.363 | −0.348 | −0.167 | −0.163 |
PI1 | 0.576 | −0.603 | 0.586 | −0.252 |
PI2 | −0.509 | 0.507 | −0.423 | −0.021 |
PI | −0.646 | 0.662 | −0.604 | 0.146 |