We focused on hot extremes in summer (June–August). We obtained the daily surface air temperatures for 1961–2022 from a dataset of 2481 stations provided by the National Meteorological Information Centre of the China Meteorological Administration. This dataset has been recommended for evaluating temperature extremes in China (Li and Dong, 2009). After strict quality control measures, 1240 stations were retained for this study. Stations were excluded if the rate of missing values in the annual data series was > 5%.
To analyze the effects of city size on hot extremes, we classified the selected cities into three categories according to population and the method of classifying city populations in China (Jones and O’Neill, 2016). Cities with a population of > 10 million were defined as megacities, cities with a population of 5–10 million were defined as large cities, and cities with a population < 5 million were defined as medium- and small-sized cities.
China has undergone rapid urbanization during the last four decades. We considered five city clusters in the Beijing–Tianjin–Hebei (BTH) region, the Yangtze River Delta (YZRD), the Pearl River Delta (PRD), the Sichuan–Chongqing region (SCR), and the Wuhan region (WHR). Each city cluster included one to two megacities, large cities, and medium- and small-sized cities.
We focused on the high-impact features of hot extremes, including their intensity, frequency, and duration. Table 1 lists several relevant hot extreme indices and their definitions. Apart from the terms “compound hot day” and “heatwave,” the definitions of hot extremes referred to the Expert Team on Climate Change Detection and Indices (ETCCDI; Zhang et al., 2011). Trends in the frequency and intensity of hot extremes were detected by using Kendall’s tau-based slope estimator (Sen, 1968). Any possible autocorrelation in the time series was removed by using an iterative procedure, which has been demonstrated to be suitable for trend detection by the Monte Carlo simulation (Zhang and Zwiers, 2004).
Hot extremes Definition TXx Summer maximum value of the daily maximum temperature (Tmax; °C) TNx Summer maximum value of the daily minimum temperature (Tmin; °C) Hot day A day on which Tmax exceeds the 95th percentile threshold (day) Hot night A night on which Tmin exceeds the 95th percentile threshold (day) Compound hot day A day on which a hot day is followed in sequence by a hot night (day) Heatwave Prolonged number of days with Tmax > 35°C (day)
Table 1. Definitions of hot extremes
Extreme high temperatures were recorded over much of China in summer 2022 (Witze, 2022). Extreme high temperatures were observed in many parts of central and eastern China, especially in large cities. We did not include the western regions of China in this study because the cities and populations are smaller in these regions.
Figure 1 shows our in situ observations of the hot extremes, including the summer maximum value of the daily maximum temperature (TXx), the summer maximum value of the daily minimum temperature (TNx), the number of heatwave days, and the number of compound hot days. Daytime extreme high temperatures > 35°C covered most of central and eastern China and temperatures > 40°C occurred in the YZRD, WHR, and parts of southern BTH. The maximum temperature in the SCR was > 42°C (Fig. 1a). The PRD experienced extreme high temperatures > 38°C.
Figure 1. Spatial distributions of (a) TXx, (b) TNx, (c) the number of heatwave days, and (d) the number of compound hot days over China in 2022. The black dashed ellipse areas indicate the five city clusters in this study. We used the Cressman interpolation technique to interpolate the station-based data and indices on 1° × 1° grids.
In addition to the high values of TXx normally perceived in the daytime, TNx at night cannot be ignored. The distribution of TNx shown in Fig. 1b suggests that the maximum nighttime temperatures > 30°C were recorded in the YZRD, SCR, and WHR, and > 28°C in the PRD and BTH region. There were > 40 heatwave days in the SCR, WHR, and YZRD in summer 2022 (Fig. 1c).
Compound hot days with a sequence of daytime/nighttime hot extremes have severe impacts on human health (Chen and Zhai, 2017; He et al., 2021; Wang et al., 2021). In summer 2022, the YZRD was the region most frequently influenced by compound hot extremes. There were > 32, > 28, and > 20 compound hot days in the SCR, YZRD, and WHR, respectively.
We focused on nine megacities. Figure S1 and Table S1 show the distribution of the selected stations and the corresponding city information. We compared the record-breaking hot extremes of each megacity between the historical time period 1961–2021 and the year 2022 (Table 2). Our results show that Shanghai, Hangzhou, Chongqing, Chengdu, and Wuhan all broke historical temperature records in 2022, including TXx in the daytime, TNx in the nighttime, the number of heatwave days or compound hot days. The hot extremes in Beijing, Tianjin, Guangzhou, and Shenzhen ranked high in the historical values, although historical records were not broken. Ren et al. (2016) reported that cities release a large amount of heat at night when the heat island effect is much stronger than in the day, leading to higher temperatures at night and more compound hot extremes in cities. Wang et al. (2021) suggested that this increase in compound hot days in urban areas of eastern China is attributable to the increase in greenhouse gas emissions and urbanization.
Megacity (station ID) Historical record (1961–2021) Year 2022 TXx (°C) TNx (°C) No. of heatwave days No. of compound hot days TXx (°C) TNx (°C) No. of heatwave days No. of compound hot days Beijing (54511) 41.9 (1999) 29.2 (2010) 26 (2000) 14 (2018) 39.2 28.4 15 7 Tianjin (54527) 40.5 (2000) 29.4 (2018) 25 (2018) 15 (2018) 40.1 28.4 14 5 Shanghai (58367) 40.9 (2017) 32.1 (2010) 47 (2013) 34 (2013) 40.9* 31.6 49* 36* Hangzhou (58457) 41.6 (2013) 30.7 (2010) 51 (2013) 28 (2013) 41.8* 31.3* 56* 32* Chengdu (56295) 40.3 (2006) 29.4 (2016) 35 (2006) 21 (2006) 43.4* 32.4* 51* 41* Chongqing (57511) 44.3 (2006) 32.8 (1964) 58 (2006) 25 (2006) 45* 34.5* 59* 30* Wuhan (57494) 39.7 (2017) 32.3 (2003) 44 (1961) 13 (2013) 39.7* 31.6 48* 22* Guangzhou (59287) 39.1 (2004) 30.4 (2015) 35 (2006) 23 (2020) 38.1 29.5 26 18 Shenzhen (59493) 38.7 (1980) 30.3 (2005) 10 (1998) 20 (2000) 36.2 28.9 9 11
Table 2. Comparison of hot extremes between historical records during the time period 1961–2021 and the year 2022 in the selected nine megacities in China. Bold values marked with an asterisk indicate that the values are greater than or equal to the corresponding historical record. Year of historical record is given in parentheses
China’s cities are facing increasingly frequent and intense heatwaves under global warming. Cities have had an additional warming influence on hot extremes as a result of the superimposed urbanization effect. To understand the effect of the background of urbanization on changes in hot extremes, we analyzed the historical changes in the frequency and intensity of hot extremes and mean temperatures in different sized cities from five city clusters and investigated possible linkages.
Figure 2 shows the changes in the frequency of heatwaves in five of China’s megacities (chosen from the corresponding city cluster). Increasing trends in the number of heatwave days over the last six decades are evident in Beijing, Guangzhou, and Shanghai; the intensity of heatwaves has clearly amplified since the 1990s. A remarkably increased frequency and greater intensity of heatwaves have been observed in Chongqing and Wuhan since the 1980s, although there has been an interdecadal variation during the last six decades. More frequent and stronger heatwaves are projected to affect China’s cities in the near future (Liu et al., 2018; Yu et al., 2018).
Figure 2. Changes in the number of heatwave days in China’s five megacities from 1961 to 2022. The size of the colored circles represents the intensity: the larger the size, the stronger the intensity.
The urbanization effects were also non-negligible for local warming and the exacerbation of hot extremes. To reflect the effects of urbanization and background warming, we calculated the changes in the summer mean temperature for different city scales in the five city clusters from 1961 to 2022 (Fig. 3). Figure S1 shows the distribution of selected stations (including the station number and classification) and Table S1 provides information on the population of the cities. Figure 3 shows that the changes in the mean summer temperatures in the five major city clusters were similar, with a clear increasing trend. Megacities and large cities have experienced higher temperatures and greater warming trends than medium- and small-sized cities due to the intensifying urbanization effect. The enhanced heat island effect has an important role in amplifying warming in megacities and large cities. The warming trend in large cities has become increasingly obvious since the 1990s, especially in the BTH region, YZRD, PRD, and WHR. These warming amplification characteristics coincide with China’s urban development (Zhai et al., 2019).
Figure 3. Changes in the summer mean surface air temperature in megacities (red lines), large cities (orange lines), and medium- and small-sized cities (light blue lines) in China’s five city clusters from 1961 to 2022.
During 1961–2022, the mean trends in temperature changes of megacities in the BTH region, YZRD, PRD, SCR, and WHR were 0.28, 0.33, 0.3, 0.14, and 0.28°C decade−1, respectively. For large cities, the mean trends in temperature changes in these regions were 0.21, 0.24, 0.17, 0.11, and 0.24°C decade−1, respectively. For medium- and small-sized cities, the mean trends in temperature changes in these regions were 0.16, 0.18, 0.14, 0.09, and 0.12°C decade−1, respectively. The differences in the warming trends between large and small-sized cities were from 0.05 to 0.16°C decade−1. This quantitative estimation can largely be attributed to the intensification of the urbanization effect of China’s cities. The warming trend in medium- and small-sized cities is more likely to be affected by background warming, whereas megacities and large cities are affected by both background warming and intensified urbanization effects. The phenomenon of trend dependence on city size reflects the intensified heat island effect.
Table 3 shows the average number of days with hot extreme events for megacities, large cities, and medium- and small-sized cities in recent decades (1991–2022), the earlier reference period (1961–1990), and their differences. In general, the cities along the Yangtze River (e.g., in the YZRD, WHR, and SCR) experienced more heatwaves and compound hot days than those in the BTH region and PRD due to the influence of the background climate. During 1961–1990, the average number of hot extreme events in the five major city clusters was relatively small. However, the number of heatwave days and compound hot days increased significantly in the five city clusters during 1991–2022 when there was rapid urbanization. The megacities and large cities in all of the city clusters experienced a larger increase in heatwave days and compound hot days than the medium- and small-sized cities. These results suggest that the rapid urbanization during the recent decades has had an important role in the increase in local hot extremes.
City cluster City classification Average number of days > 35°C during 1961–1990 (P1) Average number of days > 35°C during 1991–2022 (P2) Difference in the average number of days > 35°C (P2 − P1) Average number of compound hot days during 1961–1990 (P1) Average number of compound hot days during 1991–2022 (P2) Difference in the number of compound hot days (P2 − P1) BTH MC 5 10 5* 1.2 3.8 2.6* LC 7 9 2 1 3.2 2.2* M&SC 3 5 2 0.6 2.1 1.5 YZRD MC 13 24 11* 2.6 11.5 8.9* LC 10 16 6* 2.4 8.7 6.3* M&SC 9 15 6* 1.7 7.3 5.6* PRD MC 4 12 8* 1.4 12.3 10.9* LC 6 10 4 1.4 11.5 10.1* M&SC 3 5 2 0.9 9.5 8.6* SCR MC 15 22 7* 1.9 5.2 3.3* LC 10 15 5* 1.6 4.7 3.1* M&SC 6 9 3 1.5 4.3 2.7* WHR MC 17 22 5* 1.6 3.5 1.9* LC 15 16 1 1.5 3.1 1.6 M&SC 11 13 2 1.3 2.8 1.5
Table 3. Average number of heatwave days and compound hot days in megacities, large cities, and medium- and small-sized cities from the five city clusters during the historical time period 1961–1990 and the time period 1991–2022. Bold values marked with an asterisk indicate that the difference is statistically significant at the 0.05 level (MC: megacities; LC: large cities; M&SC: medium- and small-sized cities)
Background warming and increasingly intensified heat island effects have triggered more hot extremes in China’s cities in recent decades (Fig. 4). Figures 4a–c show the trends in the frequency of hot days, hot nights, and compound hot events associated with enhanced urbanization effects (as reflected by the trends in the mean temperatures) for different sized cities from the five city clusters. The YZRD and PRD regions had larger and more obviously increasing trends in hot extreme events than the BTH region, SCR, and WHR. This regional difference is probably because the YZRD and PRD are more affected by an intensified Northwest Pacific subtropical high (Luo and Lau, 2017), water vapor feedback related to the South China Sea summer monsoon (Deng et al., 2020), and land–atmosphere interactions (Chen and Zhou, 2018; Wang et al., 2020).
Figure 4. Linkage between the trends in the summer mean temperatures and numbers of (a) hot days, (b) hot nights, (c) compound hot days, (d) the daytime TXx, and (e) the nighttime TNx in different sized cities. The time period for the trend estimation is 1961–2022 (MC: megacities; LC: large cities; M&SC: medium- and small-sized cities).
All the megacities and large cities in the five city clusters showed greater increasing trends of hot days, hot nights, and compound hot days than the medium- and small-sized cities. For hot days, for each 0.1°C decade−1 increase in the trend of the mean temperature, the trend in the numbers of hot days increased by 0.4, 1.24, 1.67, 1.21, and 0.46 days decade−1 in the megacities located in the BTH region, YZRD, PRD, SCR, and WHR, respectively. For hot nights, for each 0.1°C decade−1 increase in the trend in the mean temperature, the trend in the increase in frequency of hot nights was 0.82, 1.62, 1.81, 1.7, and 0.75 days decade−1 in megacities in the BTH region, YZRD, PRD, SCR, and WHR, respectively. For compound hot days, the increasing trends in megacities were 0.31, 1.14, 1.41, 1.2, and 0.5 days decade−1 in the BTH region, YZRD, PRD, SCR, and WHR, respectively. The increasing trends of hot extremes in the five city clusters during 1991–2022 were similar to the trends during 1961–2022, but showed more obvious trends as a result of rapid urbanization since the 1990s.
Figures 4d and 4e show similar results for the trends in the intensity of TXx and TNx. For all five city clusters, rapid urbanization was reflected by the rapid increase in the mean temperature. The increasing trends in the intensity of TXx and TNx in megacities and large cities were clearly greater than the trends in medium- and small-sized cities. It should be emphasized that nighttime hot extremes are more affected by urbanization effects (Fig. 4e). Increasingly frequent and stronger hot extremes are therefore associated with faster warming in larger cities.
|TXx||Summer maximum value of the daily maximum temperature (Tmax; °C)|
|TNx||Summer maximum value of the daily minimum temperature (Tmin; °C)|
|Hot day||A day on which Tmax exceeds the 95th percentile threshold (day)|
|Hot night||A night on which Tmin exceeds the 95th percentile threshold (day)|
|Compound hot day||A day on which a hot day is followed in sequence by a hot night (day)|
|Heatwave||Prolonged number of days with Tmax > 35°C (day)|