More frequent summer heat waves in southwestern China linked to the recent declining of Arctic sea ice

We first introduce the unique topography of SWC. As seen in figure 1(a), the land surface height of China decreases like a three-step staircase from west to east. The first step mainly includes the Tibetan Plateau, with an average elevation above 4000 meters. The second step is primarily consisted of the Inner Mongolia Plateau and the Yunnan-Guizhou Plateau, with an average elevation of 1500–2500 meters. The third step includes the plains and basins in eastern and southern China, with elevations mostly below 500 meters. Particularly, SWC includes the provinces of Sichuan, Yunnan and Guizhou, the municipality of Chongqing, and the Tibet Autonomous Region of the Sichuan province, which lies across the three steps, with sharp elevation drops from over 4000 m in the Tibetan Plateau to less than 500 m in the Sichuan Basin (Qin et al 2010).

Zoom In Zoom Out Reset image size

The complex topography in SWC is accompanied with dramatically varying climate. Figure 1(b) presents the multi-decadal changes in drought severity, where a negative value of PDSI anomaly represents a more severe drought condition, and vice versa. We can see that the multi-decadal strengthening of drought severity has appeared mainly in SWC and northern China, while the drought severity in northwestern China has been mitigated. Correspondingly, China summer HW day (HWD) has also grown considerably in recent decades. As indicated by figure 1(c), the multi-decadal increases in China summer HWD primarily exist in eastern China, northern China, and SWC. We note that, SWC is the only region where the drought severity and HWD have both increased substantially, which could lead to increasing impacts on the ecosystem and human society in SWC.

The long-term changes in drought and HW over SWC are accompanied by significant changes in the local precipitation and surface air temperature. As seen from figure 1(d), the seasonal mean T2 m in SWC experienced a significantly upward trend for the period of 1979–2018, while the mean precipitation over the same period showed a downward trend. The significant opposing correlations between the two variables exist not only in the long-term trends, but also in the interannual time scales. The correlation coefficient between the two variables is −0.43 (−0.32 after removing linear trends), which exceeds the 99% (95%) confidence levels, suggesting that the air temperatures in SWC be intimately linked to the changes in local precipitation. As a typical case, during summer of 2006, the T2 m (precipitation) in SWC reached a record high (low), corresponding to exceptional droughts and HWs during that summer. Nevertheless, the trends in precipitation are apparently less significant than those in the air temperatures. Previous studies (Zhang et al 2017) have identified the importance of large-scale atmospheric circulation on the changes in precipitation, which suggest the critical role of remote control on the drought, and thus on the HWs in the study region.

Figures 2(a)–(b) show the climatology of HWD and the linear trend in HWD for the period of 1979–2018, respectively. As indicated by figure 2(a), the peak values of climatic HWD appear in eastern Sichuan Province and Chongqing City, with an average magnitude of 4 d. The distribution of HWD trend is generally similar to the climatology, but is more concentrated in the Sichuan Basin, with the largest trend exceeding 3 d per decade. Figure 2(c) further shows the year-to-year variations of SWC summer HWD, averaged over 26°N–33°N/102°E–108°E. It is shown that SWC HWD experienced a rapid increase over the past 40 years, especially from mid-1990 s, with the largest HWD occurring in 2006.

Zoom In Zoom Out Reset image size

Figure 2(d) displays the probability density function (PDF) of summer daily T2 m during different periods. To explore the impacts of global warming on the probability of SWC HWs, we divided the study period into two equal time slices, i.e. 1979–1998 and 1999–2018. Using 1998/1999 as the turning point enables us to obtain the equally long time slices, both of which include 20 years. As shown in figure 2(d), the average of T2 m has increased by about 2 °C from 1979–1998 to 1999–2018, when the probability of T2 m exceeding 22 °C has ascended from about 2% during 1979–1998 to 8% during 1999–2018. Nevertheless, the probability of T2 m exceeding 25 °C is nearly zero for the two periods. In other words, the shift of mean state may lead to 4 times increase of mild temperature extremes, but it alone cannot trigger the more extreme HWs. In contrast, the probability of T2 m exceeding 25 °C during peak HWD years (defined as the years when summer HWD exceeds 8 d, i.e. 2002, 2006, 2013, 2016, 2017, 2018) is more than 4%, suggesting that the occurrences of extreme HWs during specific years be linked to additional physical processes.

Motivated by previous studies that Arctic warming amplification may trigger more HWs in lower latitudes (Francis and Vavrus 2012, Deng et al 2019b, 2020), we explore the possible linkages between SWC summer HWs and the Arctic sea ice. As displayed in figures 3(a) and (c), the SWC summer HWD is significantly correlated with the reduction of sea ice in the Barents Sea and Kara Sea during antecedent winter and spring. During summer, the reductions in sea ice extend northward, and appear in the entire Arctic regions (figure 3(e)). Right panels of figure 3 further show the time series of area-averaged sea ice and SWC summer HWD. The fluctuations of Arctic sea ice were small until mid-1990 s, when the sea ice started to decrease sharply. Meanwhile, SWC summer HWD was small before mid-1990 s, but went up rapidly after late-1990 s. The opposite relationships between SWC summer HW and the Arctic sea ice are observed not only in the trends but also in the year-to-year variations. For example, in 2006, 2013 and 2016, the Arctic sea ice was at extremely low levels, while SWC summer HWD during the same year showed peak magnitudes. Moreover, the opposite correlation between the two variables seemed to be more prominent after late 1990 s. One possible explanation for the lack of such a strong relationship between the two during the initial years could be that the thickness of the sea ice has become critically low after late 1990 s, which led to increased heat fluxes at the sea-air interface, making it easier to trigger atmospheric circulation anomaly in the Arctic regions.

Zoom In Zoom Out Reset image size

The correlation between SWC HWD and Arctic sea ice during summer is as high as −0.67, which exceeds the 99% confidence level, implying that there could be physical bridges between them. Figure 4 depicts the lag regression maps of 200-hPa geopotential height and wave activity flux onto SWC HWD. As seen from figures 4(a) and (c), the occurrences of SWC summer HWs are significantly correlated with high pressure anomalies over the polar region and low pressure anomalies over the Barents Sea during the preceding winter and spring. In summer, significantly low (high) pressure anomalies appear over the Arctic (subarctic) regions (figure 4(e)). That is, anomalous losses of sea ice in Barents Sea and Kara Sea enables increased heat fluxes released into the atmosphere from the beneath ocean, resulting in higher air temperature and lower surface pressures. The establishment of abnormal low (high) pressures in Arctic (subarctic) is reminiscent of an intensified Arctic Oscillation (AO), whose correlation coefficient with SWC HWD during summer is 0.45 (trends removed), passing the 99% confidence level. (Yang et al 2016) have proposed that the Arctic lower-tropospheric warming due to the sea ice loss could lead to climate anomaly in China via atmospheric teleconnection. As seen from figures 4(b), (d) and (f), the SWC summer HWs are significantly correlated with anomalous wave trains in the troposphere over Eurasia, which propagate from northern Europe to East Asia, leading to anomalous high pressures over SWC.

Zoom In Zoom Out Reset image size

To substantiate the teleconnection mechanisms proposed, we further examine the model simulations forced by the present-day and pre-industrial sea ice. As seen from figure S1 (stacks.iop.org/ERL/15/074011/mmedia), compared to the pre-industrial period, the sea ice in the present-day period has reduced considerably in winter, spring, and summer. The reductions in sea ice mainly concentrate in the Barents Sea and Kara Sea during the winter and spring, which extend to the entire Artic regions during summer. As a result, due to the declining of Arctic sea ice, the atmospheric circulation has experienced remarkable changes. Under the forcing of decreased Arctic sea ice, anomalous Rossby wave trains were triggered during spring and summer, which seem to originate from the polar region and propagate toward East Asia, leading to anomalous high pressures over SWC. The model simulations further demonstrate that the SWC summer HWs are likely to be affected by the changes in Arctic sea ice.

Finally, we investigate local physical processes responsible for SWC summer HWs. As seen from figure 5(a), correlated with SWC summer HWD, significantly high pressure anomalies appear over northern China, which is believed to be forced by the Arctic sea ice reductions, as mentioned above. The abnormally high pressures over northern China lead to the westward extension of the western North Pacific subtropical high, which significantly reduces the total cloud cover in SWC (figure 5(b)). Due to the decrease in total cloud cover, there would be more solar short-wave radiation reaching surface, leading to anomalous high T2 m (figure 5(c)). In addition, under the control of the western North Pacific subtropical high, anomalously descending air and divergent winds tend to exist over SWC, which suppress the local precipitation (figure 5(d)), leading to reduced soil water, weakened evaporative cooling effect, and increased sensible heat, which are conducive to the occurrence of HWs.

Zoom In Zoom Out Reset image size

Moreover, as shown in figure 5(e), the atmospheric circulation anomalies over northern China also cause a northward shift of the East Asian summer westerly jet stream, which is consistent with previous studies (e.g. Lin and Lu 2008). Lin (2013) has further noted that the northward jumps of the East Asian upper-tropospheric jet stream could lead to increased precipitation in North China and decreased precipitation in lower latitudes, which is due to the enhanced (reduced) moisture convergence in North China (lower latitudes). Indeed, as shown in figure 5(f), the poleward shift of the East Asian westerly jet stream is featured by the significant reductions of rainfall in SWC. The decreased precipitation as well as the longer insolation could ultimately increase the probability of SWC HWs.