Based on the daily temperature data from weather stations in China, linear trends of the seasonal mean and extreme temperatures in summer and winter are analyzed and compared for the periods of 1960–89 and 1990–2009. The results show prominent changes in those trends since the early 1990s, in particular in winter—a signal of climate shift as previously identified. The changes, however, are found to be strongly region dependent.
In summer, both seasonal mean and extreme temperatures show a considerable cooling trend in central China and a warming trend in north and south China before 1990. After 1990 all temperature indices show significant warming trends throughout China with the largest trend up to 4.47°C (10 yr)−1 in north China. In winter in north China, with the most prominent warming trend during 1960–89, there is a significant cooling trend in both the seasonal mean temperature and the cold temperature indices after 1990. The warming trends over the Tibetan Plateau are substantially enhanced since 1990. All indices for the diurnal temperature range (DTR) show consistent decreasing trends in both summer and winter throughout China before 1990 while they turn to increasing trends in northeast China in summer and over the Tibetan Plateau in winter after 1990. The annual temperature range displays a decreasing trend throughout China before 1990 while it is dominated by an increasing trend after 1990 except over the Tibetan Plateau and in a narrow band along the Yangtze River. Possible mechanisms for the observed trend changes are discussed.
Extreme temperature events have impacts on various aspects of human activities, health, agriculture, energy, and infrastructure (Pounds et al. 1999; Colombo et al. 1999; Meehl et al. 2000; Díaz et al. 2005; Ainsworth and Ort 2010; Hertel and Rosch 2010). For example, in the event of the European heat wave in 2003, about 22 000 people died across Britain, France, Italy, the Netherlands, Portugal, and Spain. Other mortality estimates are as high as 35 000 (Epstein 2005). Many rivers ran dry by the end of the summer, and the crops that had started growing 3–4 weeks earlier than normal wilted prior to harvest (Beniston 2004). Another example is the severe freezing rain and heavy snowfall that hit central and south China in the winter of 2008 and caused tremendous damage, including loss of life, property damage, transportation shutdown, and widespread power outages, causing up to $22 billion (U.S. dollars) in economic loss (Wang et al. 2008; Wen et al. 2009). As suggested by Karl and Easterling (1999), continued documentation of climate trends, particularly in terms of climate extremes, is critical for decision makers as they deal with environmental changes and their impacts.
Under the background of global warming, trends in extreme temperature events have displayed complicated distributions in intensity and frequency (Houghton et al. 2001). Previous studies have shown that in response to global warming the temperature changes are obviously asymmetric between daytime and nighttime, leading to a decrease in the diurnal temperature range (DTR) (Karl et al. 1991, 1993; Horton 1995; Cooter and Leduc 1995). In China the extreme temperature presents asymmetric trends between summer and winter and shows different features with respect to different temperature indices. In summer, central China presents a cooling trend, which contrasts sharply with the general warming trend of the surface climate in spring, autumn, and winter in the region (Zhai and Ren 1997; Hu et al. 2003; Qian and Lin 2004). The number of days with maximum temperature over 35°C showed a slightly decreasing trend for China as a whole during 1951–99, especially in eastern China (Zhai and Pan 2003). However, the frequency of warm days increased in northern and western China but decreased in most parts of central and southeastern China (Ren and Zhai 1998; Gong et al. 2004; Qian and Lin 2004).
Previous studies also have revealed a dramatic change in temperature trends in early 1990s. For example, a remarkable increasing trend in the hot days after 1990 is found over the arid and semiarid regions in northern China (Ma et al. 2003; Zhai and Pan 2003; Wei and Chen 2009). Qian et al. (2011) identified a multidecadal change in amplitude of the annual cycle of daily mean surface air temperature over China. They reported a significant decrease, −0.23°C (10 yr)−1, to a significant increase, 0.29°C (10 yr)−1, in the annual temperature range (ATR) around 1993, which is imposed on a linear trend. In the meantime, the averaged global surface air temperature has increased by 0.7°C during the last century with an enhanced warming rate in the latter half of the period (Solomon et al. 2007). Liu et al. (2004) reported that the global annual mean maximum (Tmax) and minimum (Tmin) temperatures increased by 0.56° and 0.51°C, respectively, during 1990–2000. The δ18O variation of ice cores from Muztagata in the eastern Pamirs also indicated a rapid warming trend since 1990 (Tian et al. 2006). The annual mean temperature over China in 1998 was 1.38°C above normal (Wang and Gong 2000). Note also that in the high latitude the Arctic Oscillation (AO) and the North Atlantic Oscillation (NAO) indices have been decreasing significantly since 1990 (Cohen and Barlow 2005). In the tropics the anomalous conditions were characterized by persistent large-scale positive SST anomalies in the tropical Pacific and a weakening of the trade winds in the 1990s (Latif et al. 1997), as well as more frequent occurrence of the central Pacific El Niño events (e.g., Larkin and Harrison 2005; Ashok et al. 2007; Weng et al. 2007; Kao and Yu, 2009; Kug et al. 2009; Yeh et al. 2009; Zhang et al. 2011). All of these signals indicate a climatic shift in the early 1990s.
Although previous studies have reported the abrupt changes in frequency of warm days and warm nights and the amplitude of annual cycle of daily mean surface air temperature in China since 1990, a question still remains whether any dramatic changes in extreme temperature events occurred around 1990 over China. In this study, the trends in extreme surface air temperature events in summer and winter over China during the period from 1960 to 2009 are analyzed, focusing on the dramatic change around 1990.
The rest of the paper is organized as follows. Section 2 describes the dataset and definitions of extreme temperature indices used in this study. Section 3 illustrates the significant climatic change in terms of extreme temperatures in the later 1980s and early 1990s for China as a whole, as well as a quantitative test of the turning point in 1990. Individual trends in the seasonal mean and extreme temperatures in summer and winter during 1960–89 and 1990–2009 are analyzed and compared in section 4, as well as the annual range of extreme temperatures as a measure of the seasonal variation of extreme temperatures. Possible mechanisms of the observed trend changes are discussed in section 5. Conclusions are drawn in the last section.
2. Data and methodology
The daily data of 734 weather stations in mainland China from 1951 January to 2009 December used in this study were obtained from China Meteorological Administration (CMA) National Meteorological Information Center (NMIC) (see the details at http://cdc.cma.gov.cn/shuju/index3.jsp?tpcat=SURF&dsid=SURF_CLI_CHN_MUL_DAY). The dataset consists of daily maximum (Tmax), minimum (Tmin) and daily mean 2-m air temperature (T). To avoid bias in the trend analysis due to the missing data, our analysis is restricted to the period from 1960 to 2009. In the analysis, the diurnal temperature range is defined as the difference between the daily Tmax and Tmin. Summer refers to June–August (JJA) and winter refers to December–February (DJF). In total there are 548 stations that have complete data for the whole period (Fig. 1). Although it is hard to eliminate urban effects (especially the urban heat island effect) on temperature, its effect on large-scale features is very limited (Wang et al. 1990; Easterling et al. 1997). For example, Liu et al. (2004) found that temperatures at the urban stations increased slightly faster than those at the nonurban stations with a difference of 0.07 and 0.02°C (100 yr)−1 for Tmax and Tmin, respectively. In addition, the changes in stations locations and elevations, which may introduce discontinuity in trend analyses, have been corrected in the released dataset.
The thresholds for temperature extremes at each station are set at the 95th and 5th percentiles of daily temperatures (T, Tmax, Tmin, and DTR) in a given season as used in Bell et al. (2004). The 5th/95th percentiles of daily temperature for each summer (92 days) and each winter (90 days) season approximately correspond to the five hottest/coldest temperatures. Therefore, the daily temperatures are sized down in each season first. Then, the five hottest ones are selected and averaged as the warm temperature, while the five coldest ones are averaged as the cold temperature. Warm and cold temperatures are applied to T, Tmax, and Tmin, respectively. For the DTR, the mean of the top/low 5% DTR of a season are termed the high/low DTR. Moreover, the annual temperature range (ATR), an indicator of seasonal variation of the extreme temperatures, is defined as the difference between the warm Tmax in summer and the cold Tmin in winter each year. The extreme temperature indices defined in this study are listed in Table 1.
A trend is defined as the coefficient of the linear regression. The significance of the trend is checked with the correlation coefficient method (Wei 1999). Since changes in the extreme temperature events are highly correlated with changes in the seasonal mean temperature (Colombo et al. 1999), we will also discuss trends in the seasonal mean of daily mean T, Tmax, Tmin, and DTR in our analysis. Monthly mean sea level pressure (SLP) field from the National Center for Environmental Predictions (NCEP)–National Center for Atmospheric Research (NCAR) reanalysis (Kalnay et al. 1996) is used to document the possible mechanisms responsible for the observed trend changes.
3. Detection of the turning point for China as a whole
Taking China as a whole, we conduct an 11-yr-sliding Student’s t test of the temperature indices in summer and winter (Figs. 2s and 2t), which indicates a significant climate shift in the later 1980s and early 1990s in both seasonal mean and extreme temperature events. The turning point varies with index and season. The indices for Tmin turn first, followed by those for the daily mean T, and finally those for Tmax. In general, the temperature indices turn a little earlier in winter than in summer. However, it is in 1990 when most of the temperature indices present a significant trend change. So 1990 can be considered as the turning point to examine the trend changes over China.
Also shown in Fig. 2 are the different trends of temperature indices in the two periods for China as a whole. Neutral trends are found in summer during 1960–89 except for the cold Tmax, which shows a cooling trend of −0.11°C (10 yr)−1 and the cold Tmin, which shows a warming trend of 0.23°C (10 yr)−1 (Figs. 2o and 2q, Table 2). In sharp contrast, significant warming trends are detected after 1990 for all temperature indices, with the largest change from −0.09°C (10 yr)−1 to 0.51°C (10 yr)−1 for the trend in the warm Tmax (Table 2). In winter all temperature indices show increasing trends before 1990, which are considerably larger than those in summer. But, in the latter period the warming trends are decreasing in the seasonal mean and the cold temperature indices, while increasing in the warm temperature indices (Table 2). The cold Tmin (Fig. 2r) shows the largest warming trend among all temperature indices and reaches 0.57°C (10 yr)−1 before 1990 (Table 2). However, the warming trend decreases to only 0.08°C (10 yr)−1 after 1990 (Table 2), which is not statistically significant at 95% confidence level.
The above results suggest that a noticeable trend change occurs in both seasonal mean and extreme temperature indices around 1990. Therefore, in this study we choose 1990 as the turning point, although some disagreement exists among different seasons and different temperature indices. Note that Fig. 2 only shows the trend changes for China as a whole. The region-dependent features will be discussed in the next section.
4. Trend changes
a. Summer (JJA)
Figure 3 shows trends in the seasonal mean and extreme temperature indices in summer during 1960–89. Different from the neutral trend for China as a whole (Fig. 2), both the seasonal mean temperature and extreme temperature indices in T, Tmax, and Tmin show significant cooling trends in central China and warming trends in north and south China (Fig. 3). The warming trends in Tmin (Figs. 3g, 3h, and 3i) are greater than those in Tmax (Figs. 3d, 3e, and 3f), while the cooling trends in Tmin are weaker than that in Tmax, leading to decreasing trends in the DTR (Figs. 3j, 3k, and 3l), consistent with the DTR trend due to global warming (Karl et al. 1991, 1993; Horton 1995; Cooter and Leduc 1995). The largest warming trends are found in the cold Tmin with maximum values over 1.71°C (10 yr)−1 in north China (Fig. 3i). These trends are significantly higher than those of 0.13°C (10 yr)−1 in the global mean temperature during the recent 50 years documented in the Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report (AR4) (Solomon et al. 2007). In contrast, remarkable cooling trends appear in central China, particularly in Tmax (Figs. 3d, 3e, and 3f), which reaches −0.6°C (10 yr)−1 at some stations, in agreement with earlier studies (Zhai and Ren 1997; Hu et al. 2003; Qian and Lin 2004).
In the latest 20 years (1990–2009), the seasonal mean and the warm temperatures of T, Tmax, and Tmin all show considerably larger warming trends than those in 1960–89 (Fig. 4), suggesting an accelerated warming since 1990 over most regions of mainland China, especially in the arid and semiarid regions (north of 35°N) and over the Tibetan Plateau (west of 100°E). Moreover, the cooling region discussed above during 1960–90 appears to have shrunk with few stations at which the trends are statistically significant at 95% confidence level. Meanwhile, trends in the cold T (Fig. 4c) and the cold Tmax remain weakly negative in the Yellow River basin (35°–45°N, 105°–120°E) and south of the Yangtze River (25°–30°N, 105°–120°E) (Fig. 4f). Because of the different trend change in Tmax and Tmin during 1990–2009, the DTR increases over northeast China and in both seasonal mean and the warm temperatures over south China (Fig. 4).
Therefore, the summer warming trend was enhanced during 1990–2009 not only in the seasonal mean temperature but also in the extreme temperature events, especially for the arid and semiarid regions (north of 35°N). The greatest increase in the warming trend occurs in Shanxi Province (35°–40°N, 110°–115°E) from 0.008°C (10 yr)−1 during 1960–89 to 4.47°C (10 yr)−1 during 1990–2009. However, the warming trends of the cold T (Fig. 4c) and the cold Tmin (Fig. 4i) decrease in Heilongjiang Province (north of 45°N, east of 115°E). Therefore, as the trends in extreme temperature events are concerned, the regional dependence needs to be considered.
b. Winter (DJF)
In winter all temperature indices in the period before 1990 (1960/1961–1989/1990) show overall increasing trends over China (Fig. 5). Though cooling trends appear in some stations, not one is statistically significant at 95% confidence level. Consistent with the trend for China as a whole (Fig. 2 and Table 2), the warming trend is weaker in Tmax, especially in the warm Tmax, than in the seasonal mean temperatures and the cold temperature indices (Fig. 5). The greatest warming trend occurs in the cold Tmin reaching up to 3.0°C (10 yr)−1 (Fig. 5i). Moreover, the trends in winter are considerably larger than those in summer, indicating a seasonal asymmetry in the trends between summer and winter (Figs. 3 and 5).
As displayed in Fig. 2 and Table 2 for China as a whole, the warming trends in the warm temperature indices are accelerated, while the warming trends in the seasonal mean and the cold temperature indices decrease after 1990. Qian et al. (2011) also reported a significant decrease in the warming trend for China as a whole from 0.43°C (10 yr)−1 during 1961–93 to 0.04°C (10 yr)−1 during 1993–2007. However, we show that the changes are region dependent over China (Fig. 6).
For the seasonal mean temperature indices (Figs. 6a, 6d, and 6g), the warming trends significantly increased over the Tibetan Plateau and its surrounding areas with the maximum trend up to 4.65°C (10 yr)−1. The accelerated warming over the Tibetan Plateau can also be detected in all extreme temperature indices (Fig. 6). Earlier studies have suggested that the increasing daily mean temperature and the decreasing DTR are attributed to the increase in greenhouse gases (Karl et al. 1991; Horton 1995; Duan et al. 2006). However, the DTR tends to increase over the Tibetan Plateau after 1990 (Figs. 6j, 6k, and 6l), which is in sharp contrast to that before 1990 shown in Fig. 5. Liu et al. (2004) suggested that the DTR in winter is closely related to solar irradiance. Before 1990 the coexistence of growing greenhouse gases and the decrease in solar irradiance leads to the decreasing trends in Tmax and increasing trends in Tmin, thus the decreasing trend in the DTR (Liu et al. 2004). However, the increase in solar irradiance since 1990 results in an enhanced increase in Tmax and, thus, the increase in the DTR for China as a whole (Liu et al. 2004; Qian et al. 2011).
Warming trends turn to cooling trends at some stations in north and central China after 1990, in particular for the cold temperature indices (Figs. 6c, 6f, and 6i). Though the cooling trends are statistically significant at few stations, they result in much reduced warming trends in the seasonal mean and the cold temperature indices over China as a whole after 1990 (Table 2). More dramatic than the seasonal mean temperature indices, the cold T, Tmax, and Tmin show significantly decreasing trends after 1990 in north and south China with the largest cooling trend of −5.31°C (10 yr)−1 in the cold Tmin in north China (Fig. 6i). This cooling trend in the cold Tmin over north and south China contributes to the remarkable change from a significant warming trend of 0.57°C (10 yr)−1 during 1960–89 to a nearly neutral trend of 0.08°C (10 yr)−1 during 1990–2008 for China as a whole (Table 2).
In contrast to the trend change in the seasonal mean and cold temperature indices, the warm T, Tmax, and Tmin show warming trends in the recent 20 years throughout China except at some stations north of 40°N. Because of the different trends in the warm and cold temperature indices, the subseasonal variability of the extreme temperatures in winter becomes greater at most stations throughout China after 1990 except for the region over the Tibetan Plateau. This indicates more frequent extreme temperature events in winter since 1990.
Therefore, the trends in seasonal mean and extreme temperature indices present more significant changes around 1990 in winter than in summer. The changes vary from region to region, however. The Tibetan Plateau becomes warmer and warmer, while north China experiences more severe extreme cold temperature events in the winters since 1990. This regional dependence could not be revealed for China as a whole, as shown in Fig. 2 and Table 2. Therefore, it is inadequate to consider trends in the extreme temperature events for China as a whole because averaging over the whole China would obscure the spatial distributions of trends in the observed extreme temperature events.
It is well known that changes in the variability would have a larger effect on the frequency of extreme events than changes in the mean (Katz and Brown 1992). The analysis above shows seasonal asymmetric changes in the extreme temperature trends between summer and winter, which indicates a robust seasonal variation of extreme temperatures. For this reason, we define the annual temperature range (ATR) as the difference between the warm Tmax in summer and the cold Tmin in winter each year. The trends in ATR during the two periods are discussed in this subsection. In the period from 1960 to 1989 (Fig. 7a), the ATR shows a decreasing trend throughout China due to the greater warming trend in winter than in summer. In contrast, in the recent 20 years the ATR tends to increase over both north and south China (Fig. 7b), while it remains negative over the Tibetan Plateau and central China, displaying a sandwichlike distribution from south to north over China.
Based on the spatial pattern of the ATR trend after 1990 (Fig. 7b), four regions are defined: north China, the Tibetan Plateau, central China, and south China, as shown in Fig. 1. Figure 8 plots the time series of the warm Tmax in summer and cold Tmin in winter over the four regions, superimposed by the corresponding linear trends for the period before and after 1990, respectively.
In north China (Fig. 8a), the cold Tmin in winter exhibits a significant warming trend of 0.63°C (10 yr)−1, while a cooling trend of −0.11°C (10 yr)−1 is detected in the warm Tmax in summer before 1990. During the period after 1990 the cooling trend in the warm Tmax in summer turns to a significant warming trend of 0.68°C (10 yr)−1. In sharp contrast, the pronounced warming trend in the cold Tmin in winter turns to a cooling trend of −0.18°C (10 yr)−1. The cooling trend is not statistically significant at the 95% confidence level because of the regional average. In fact, the cold Tmin at many stations in north China presents a dramatic cooling trend after 1990. For example, the cold Tmin at station Hailar (49.13°N, 119.45°E) decreases with a trend of −1.4°C (10 yr)−1 since 1990. A similar trend change is found in central (Fig. 8c) and south (Fig. 8d) China, where the warm Tmax in summer shows a cooling trend before 1990 that turns to a warming trend after 1990, while the warming in the cold Tmin in winter slows down significantly since 1990. Over the Tibetan Plateau both the warm Tmax in summer and the cold Tmin in winter show an enhanced warming trend (Fig. 8b). However, the different trends of 0.23°C (10 yr)−1 in the warm Tmax in summer versus 0.64°C (10 yr)−1 in the cold Tmin in winter lead to a decreasing trend in the ATR after 1990 (Fig. 7b).
5. Discussion of possible mechanisms
In section 4, the differences in trend changes around 1990 across China are discussed. In particular, under the background of global warming two cooling trends are identified: one in central China in summer before 1990 and the other in north China in winter after 1990. In this section, the possible mechanisms that could be responsible for the observed trend changes will be discussed based on some well-known decadal climate variability, including changes in aerosols and solar irradiance, tropical SST condition, the Arctic Oscillation, and stratospheric cooling in East Asia as revealed in previous studies.
a. Aerosols and solar irradiance
Atmospheric aerosols comprise a mixture of mainly sulfates, nitrates, carbonaceous (organic and black carbon) particles, sea salt, and mineral dust. Black carbon is of special interest because it absorbs solar radiation, heats the air, and contributes to global warming (Hansen et al. 2000; Jacobson 2001). It is unlike other aerosols that reflect sunlight to space and have a global cooling effect (Penner et al. 2004). Therefore, the collective effect of aerosols is complicated (Menon et al. 2002). However, according to Liu et al. (2004) and Qian et al. (2011), the solar irradiaance increases since 1990 from dimming to brightening, which is partly due to a highly significant increase of nitrogen dioxide by about 50% (Richter et al. 2005). Therefore, the combined effect of the global dimming to brightening transition and a gradual increase in the warming due to greenhouse gases (Liu et al. 2004; Qian et al. 2011) is responsible for the accelerated warming trend since early 1990 in summer throughout China as well as in winter over the Tibetan Plateau (Figs. 2, 4, and 6).
b. Tropical SST conditions
Moreover, the central Pacific El Niño events—also known as the date line El Niño, El Niño Modoki, or the warm pool El Niño—occur more frequently since 1990 compared to those in 1960–89 (e.g., Larkin and Harrison 2005; Ashok et al. 2007; Weng et al. 2007; Kao and Yu, 2009; Kug et al. 2009; Yeh et al. 2009; Zhang et al. 2011). The central Pacific SST warming results in an anomalous two-cell Walker circulation (Ashok et al. 2007; Weng et al. 2007) and weakens the East Asian summer monsoon (Kwon et al. 2007; Li et al. 2010). Therefore, the southwesterly summer monsoon extends less far northward, leading to reduced rainfall in north China and increased rainfall in south China (Zhou et al. 2009), which accelerates the warming trend in north China and slows down it in south China in summer (Hu et al. 2003).1 This is consistent with the larger warming trend in the warm Tmax in north China, 0.78°C (10 yr)−1, than in south China, 0.38°C (10 yr)−1, in summer after 1990 (Figs. 8a and 8d).
Surface air temperature during winter in China is closely related with the Arctic Oscillation (Gong et al. 2001). For China as a whole, the seasonal mean T and Tmin are significantly correlated with the winter AO index from NOAA/Climate Prediction Center (CPC), which is constructed by projecting the daily 1000-mb height anomalies poleward of 20°N onto the leading pattern of the AO, at 0.38 and 0.44 (Table 3). However, the correlation coefficients between the seasonal mean temperature indices and the winter AO index are not statistically significant at the 95% confidence level over the Tibetan Plateau, central and south China except for the Tmin over central China. The high correlation between the AO index and the winter temperature indices appears in north China with a correlation coefficient up to 0.52 (Table 3). The change of the AO seems to significantly affect the cold temperatures of T, Tmax, and Tmin over north and central China with the correlation coefficient reaching 0.56 (Table 3). As proposed by Cohen and Barlow (2005), the winter AO index exhibits an increasing trend in the period before 1990, while a statistically significant decreasing trend during the recent 20 years. The AO turned to the high pressure phase in the polar region since early 1990s (Fig. 9b), leading to more severe cold air intrusions into north China in winter.
In sharp contrast, both seasonal mean and extreme temperature indices in winter in south China are not significantly correlated with the winter AO index (Table 3), indicating that the weak winter cooling trend in the cold temperature indices in south China after 1990 might not be directly related to the decreasing trend in the winter AO during the recent 20 years.
d. Stratospheric cooling
Before 1990 a significant cooling trend is detected in summer in central China (Fig. 8c). Yu et al. (2004) and Yu and Zhou (2007) indicated that this cooling trend may be connected to stratospheric temperature changes and involves interaction between the troposphere and stratosphere. Model simulations suggest a significant contribution by sulfate aerosols to this cooling, which induces a positive gradient of air temperature in the middle–upper troposphere and an enhanced East Asian summer monsoon, leading to more clouds over central China (Li et al. 2007). After 1990 the strength of zonal winds near the subtropical jet has undergone a decadal decrease (Kwon et al. 2007), leading to weakening of the cooling trend in central China. The counteraction of the regional cooling and global warming causes the trend in the warm temperature indices to increase after 1990 (Fig. 8c).
Extreme temperature events have received increasing attention in recent years, mainly due to their frequently induced loss of human lives and economic costs. An abrupt change in climate regime occurred in the early 1990s. The trends in the seasonal mean and extreme temperature indices over China have been examined for the period from 1960 to 1989 and the period from 1990 to 2009, based on the high-quality daily temperature dataset obtained from weather stations over mainland China. The results show that the trends changed prominently since 1990, especially in winter. The change varies for different regions and in different indices, however.
In the early period (1960–89), for China as a whole, neutral trends are presented in both seasonal mean and extreme temperature indices for T, Tmax, and Tmin in summer. The neutral trends result mainly from an offset of pronounced cooling trends in central China and weak warming trends in north and south China. All temperature indices for T, Tmax, and Tmin show significantly warming trends in winter with the warming trends greater than those in summer, leading to a consistent decrease in the annual temperature range (ATR).
Since the early 1990s, for China as a whole, all temperature indices for T, Tmax, and Tmin display significant warming trends in summer with the greatest warming trend reaching 4.47°C (10 yr)−1 in north China. Nonetheless, the cooling trends in central China were weakened greatly and only appeared in the cold temperature indices. The trends in the DTR became positive in northeast China and in the seasonal mean and the high DTR indices in south China since 1990. This is mainly a result of the rapid warming trend during daytime in these regions for the particular indices.
The change in all temperature trends after 1990 is more significant and complicated in winter than in summer. The most prominent and greatest warming region in north China during 1960–89 experienced a cooling trend in both the seasonal mean and the cold temperature indices in winter since 1990. Such a cooling trend is most likely induced by the winter AO, whose index has decreased rapidly since early 1990s. At the same time, the warming trends of all temperature indices over the Tibetan Plateau were enhanced, reaching up to 4.65°C (10 yr)−1. The ATR showed a decreasing trend in the early period because of the greater warming trend in winter than in summer. However, it turned to an increasing trend in both north and south China because of the different trend changes in summer and winter. The trends remain negative over the Tibetan Plateau and in central China along the Yangtze River basin since 1990.
Our analysis reveals two unusual cooling trends under the background of global warming. One is the significant cooling trend in summer in central China before 1990 and the other is the pronounced cooling trend over north China after 1990. Furthermore, because of the different trends in the warm and cold temperature indices in different seasons and within a season, the seasonal and subseasonal variability in extreme temperatures become greater at most stations, particularly in winter, throughout China since 1990 except for the region over the Tibetan Plateau. This indicates more frequent extreme temperature events since 1990.
Although our main objective of this study was to document changes in the observed trends in extreme temperature over China around 1990, the possible mechanisms responsible for the observed changes have also been discussed. It is suggested that the increase in solar irradiance since 1990 could be partly responsible for the accelerated warming trend in summer throughout China as well as in winter over the Tibetan Plateau. The central Pacific SST warming due to more frequent El Niño Modoki events since 1990 plays a role in weakening the East Asian summer monsoon, limiting its northward penetration and leading to reduced rainfall in north China (e.g., Kwon et al. 2007; Li et al. 2010). This might also contribute to the accelerated warming trend in north China in summer after 1990. The significant cooling trend in summer in central China before 1990 was likely to be partly connected to the stratospheric temperature changes, which might play a role in strengthening the East Asian summer monsoon and thus more clouds in summer over central China. The prominent cooling trend in the seasonal mean and cold temperature indices in winter over north and central China is most likely a result of the change in the Arctic Oscillation, which turned to a high pressure phase in the polar region since early 1990s, leading to more severe cold air intrusions into north China in winter after 1990. Future studies are required to quantify the individual contributions of each possible mechanism to the changes of the observed trends.
This study has been supported by the National Nature Science Foundation of China (40905044, 41075068), Public Sector (Meteorology) Special Research Foundation (GYHY200906014). Additional support has been provided by the Japan Agency for Marine-Earth Science and Technology (JAMSTEC), NASA, and NOAA through their sponsorships to the International Pacific Research Center (IPRC) in the School of Ocean and Earth Science and Technology (SOEST) at the University of Hawaii at Manoa.