Changes in Temperature‐Precipitation Compound Extreme Events in China During the Past 119 Years

This study analyzes the change characteristics of compound extreme events (CEEs) of temperature and precipitation (including warm‐wet, warm‐dry, cold‐wet and cold‐dry) in China on interannual and interdecadal scales between 1901 and 2019. The results demonstrate a long‐term increasing trend and interdecadal oscillations in CEEs total frequency. However, the frequency of each type of CEEs changes in a different manner compared with total CEEs frequency. There are fewer CEEs but increasing warm‐dry during 1901–1950. The period 1951–1995 are characterized by frequent cold CEEs (cold‐wet and cold‐dry), cold‐wet are largely distributed in most areas except for northeast and coastal areas of China, while cold‐dry are distributed in most areas except for the northwest regions of China. There are frequent warm CEEs (warm‐wet and warm‐dry) and fewer cold CEEs during 1996–2019. Warm‐wet frequently occurs in the Tibetan Plateau and northwest China, and warm‐dry mainly concentrates in southwest and northern China during this period. The frequency of warm‐dry and cold‐wet were higher than that of warm‐wet and cold‐dry over the past 119 years, whereas warm‐wet increased fastest in the northwest region after 1996, consistent with the warming and wetting characteristics in the northwest region of China. Further study show that long‐term change and low frequency oscillations have the greatest impact on CEEs among different time scale factors. Furthermore, the temperature rise caused by climate change affects the interdecadal characteristics of CEEs in China through the changes of circulation fields such as East Asian trough and subtropical high and the configuration between them.

PENG ET AL. 10.1029/2022EA002777 2 of 15 event is usually higher than the simple superposition effect (Geirinhas et al., 2021;Haqiqi et al., 2021;Wang et al., 2021;Zhang et al., 2018;Zscheischler et al., 2014;Zscheischler, Van Den Hurk, et al., 2020). Accordingly, an isolated analysis of univariate extreme events may seriously underestimate the losses caused by multivariable compound extreme events (CEEs). Besides, CEEs are known as relevant or complex extreme events. The CEEs had been reported in a special report from the Intergovernmental Panel on Climate Change in 2012 (Field et al., 2012). Subsequently, more general definitions have been proposed, that is, CEEs can be defined as the combination of multiple statistical variables or extreme events that lead to social and environmental risks (Leonard et al., 2014;Zscheischler et al., 2018). The typological classification has presented a coherent framework for the analysis of CEEs (Zscheischler, Martius, et al., 2020). The recent IPCC AR6 report has summarized new insights and revelations about CEEs (Masson-Delmotte et al., 2021;Yu & Zhai, 2021).
Numerous CEEs are correlated with the compound effects of temperature and precipitation. For instance, when high temperature stress is combined with water stress, the average yield loss of maize in the United States can be as high as four times (Haqiqi et al., 2021); the increase of heat wave-drought CEEs in India in the future can seriously threaten people's life and health (Das et al., 2022). The combined effects between the joint distribution of precipitation and surface runoff in Texas and the temperature conditions caused by the El Niño-Southern Oscillation (ENSO) and rising global temperatures often cause more disastrous compound flood events . Temperature-precipitation CEEs can be basically divided into four combinations in accordance with different combinations of temperature and precipitation (warm-wet, warm-dry, cold-wet, and cold-dry) (Hao et al., 2013(Hao et al., , 2018a(Hao et al., , 2018bMartin & Germain, 2017;. Of course, there are many other types of CEEs and some new types of CEEs worthy of attention (Zscheischler, Martius, et al., 2020).
For the temperature-precipitation CEEs in China, the frequency and spatial range changes since the 1960s have been discussed (e.g., X. Wu et al., 2021;X. Wu, Hao, Hao, Li, & Zhang, 2019;Xiao et al., 2020). The result suggests that the warm-dry in summer in eastern China with the high urbanization level have increased significantly since the late 1990s (Yu & Zhai, 2020). The probability of flood-heat wave CEEs in south, northwest and northeast China increased significantly after 2000 . Southeast and southwest China are expected to likely experience historically unprecedented heat-flood CEEs in the future .
Although increasing studies focus on the temporal and spatial characteristics of temperature-precipitation CEEs in China (Lin et al., 2015;Qian et al., 2014;Yuan et al., 2016), only studying CEEs since 1960s is insufficient to fully understand the characteristics of long-term changes. On the other hand, there are few studies on the factors that affect the change of CEEs. Although existing studies simply explain the change characteristics of CEEs after 1960s through the correlation and coupling between temperature and precipitation (X. Yu & Zhai, 2020;Zhou & Liu, 2018), they do not separate and compare the respective influence of different time scale factors on CEEs, which makes the understanding of factors affecting CEEs still insufficient. In this paper, we attempt to investigate the interdecadal changes and long-term characteristics of four types of CEEs in China from 1901 to 2019, and further isolate and compare the effects of different factors on the changes of CEEs. This is conducive to more comprehensively revealing the changing rules of CEEs in China and improving the mechanistic research on CEEs.

Data
Monthly mean precipitation and near-surface temperature data were derived from a gridded data set (CRU TS v. 4.04) compiled by the Climatic Research Unit (Harris et al., 2020). The data set span from January 1901 to December 2019 with a spatial resolution of 0.5° × 0.5°, which was generated by interpolating station-derived meteorological observations. Monthly 500-hPa geopotential height (Z500) data set with resolution of 1.0° × 1.0° during the period January 1836 to December 2015 was obtained from the NOAA-CIRES-DOE twentieth Reanalysis (V3) (20CR data set; Slivinski et al., 2019).

Definition of CEEs
Based on monthly data, the 25th and 75th percentiles of precipitation and temperature for each calendar month are used as relative thresholds in this study according to the existing research and definitions of CEEs (Beniston, 2009;Hao et al., 2013;. For January, the selected samples are all January of 1901-2019 (total 119 months), and so on in other months. Compound extremes represent the simultaneous extremes of temperature and precipitation in the same month. As a result, the temperature-precipitation CEEs can be divided into four types (warm-wet, warm-dry, cold-wet, and cold-dry).

Definition of Frequency of CEEs
In this study, the frequency of CEEs on a certain grid point is defined as the total number of months, in which both temperature and precipitation are extreme in a certain period (cf. X. . For the whole China region, the total frequency of CEEs is the sum of the months of CEEs occurring at all grid points in China. To eliminate the effect of the poleward convergence of longitude, a weighting by the cosine of the corresponding latitude is applied to each grid point (Hannachi et al., 2007) to obtain effective frequency. The CEEs on 3,845 grid points in China are obtained.

Identification of Interdecadal Transition
Cumulative anomalies were calculated to identify the interdecadal transition (Y. Ren et al., 2022;Sun et al., 2020). The increase (decrease) of the cumulative anomaly reflects that the CEEs are in the frequent (less frequent) period, and the slope of the cumulative anomaly reflects the rate of its positive or negative anomaly accumulation. The formula is as:̂= where x i (1, 2, 3, …, n), , n, and are the original time series, the average over a period, the length of time, and the cumulative anomaly at time t, respectively.

Separation of CEEs Components on Different Time Scales
Although existing literature have partially explained the changes of CEEs through the correlation and coupling between temperature and precipitation (X. Yu & Zhai, 2020;Zhou & Liu, 2018), it is worth noting that the CEEs defined by relative thresholds of monthly values introduces climate change signals into CEEs changes. Therefore, it is necessary to further separate components of CEEs caused by different time scale factors when studying the causes affecting the changes of CEEs.
For convenience, we refer to the previously defined CEEs as Case A, including components of CEEs on all time scales. Sections 3.1-3.3 discusses the spatio-temporal and interdecadal characteristics of Case A.
Then we define extreme temperature or precipitation events based on the 25/75th percentile of each quantity across a 31-year moving window centered on each target year and recalculate CEEs changes. By doing so, we remove the effects of long-term secular trends due to climate change and low-frequency oscillation. The resultant CEEs changes are denoted as Case B. Therefore, the difference between Case A and Case B represents the component of CEEs caused by long-term change and low frequency oscillations.
Case C inherits the definition of extreme temperature and precipitation events individually from Case B, but assumes that the two events are independent of each other (Equation 2). Case C represents the component of CEEs caused by changes in the variability of higher frequencies.
(Case C) = (extreme temperature) × (extreme precipitation) The difference between Case B and Case C represents the component of CEEs caused by changes in temperature-precipitation coupling.
Section 3.4 separates the components of CEEs on different time scales and the relative contributions of different components can be compared by their variance contributions rates.

Temporal Characteristics During the Past Hundred Years
The temporal characteristics of original CEEs (i.e., Case A as defined in the method) in China over the past 119 years are presented first. For the annual change (Figure 1a), the sum sequence of four types of CEEs shows generally continuous growth in most of time, nearly four times higher in the recent decade than that 100 years ago. In the periods of 1900s-1920s and 1950s-1980s, cold CEEs (cold-wet and cold-dry) contributed significantly to the total numbers of CEEs. However, the contribution of warm CEEs (warm-dry and warm-wet) is significantly greater than that of cold CEEs in the last three decades, which has also appeared in 1930s-1940s.
Warm-dry has oscillated and risen over the past 120 years, especially the last three decades, which is basically consistent with the existing studies (e.g., X. Wu et al., 2021;X. Wu, Hao, Hao, Li, & Zhang, 2019;Xiao et al., 2020), whereas increasing trend before 1950s and decreasing trend in 1950s are found.
For warm-wet, the frequency was relatively less before the 1930s, increased since 1940s and increased significantly after the 1990s. Notable is, warm-wet has maintained a rapid growth rate over the past 20 years, different from the warm-dry. The frequency of warm-wet has actually exceeded the frequency of warm-dry over the last few years, especially in summer ( Figure 1c).
Two types of cold CEEs have occurred frequently in the period of 1950s-1990s, and cold-dry has been less than cold-wet in most of the time. Both of them oscillated during 1900s-1940s, increased rapidly after the 1950s but oscillated in the following 30 years and rapidly decreased in the recent 30 years. The cold-dry has rarely occurred recently.
For CEEs in winter and summer (Figures 1b and 1c), notably, the dominance of cold CEEs in summer lasts longer and warm CEEs do not exceed the cold CEEs until the 2000s.
In general, CEEs in China have been generally increasing over the past one hundred years, and there are two increasing periods: one is the period of 1900s-1950s, warm CEEs increased steadily and cold CEEs first remained stable and then increased sharply around 1950. The other is the last three decades, contributed by warm-dry and warm-wet. For the dominant CEEs types, warm CEEs have exceeded cold CEEs since 2000 and warm-wet has exceeded warm-dry over the last few years. Table S1 in Supporting Information S2 shows that the total frequencies of annual CEEs show a significant increasing trend, which is 0.31 months/decade (P < 0.001). For four different types of CEEs, the linear trend of warm-dry is the largest, reaching 0.16 months/decade, followed by warm-wet and cold-wet, respectively, and cold-dry does not show a significant linear trend.

Spatial Characteristics During the Past One Hundred Years
The spatial distribution of the frequency of CEEs between 1901 and 2019 shows the highest frequency of total CEEs in the eastern part of northwest China and lowest frequency in the western part of northwest China ( Figure 2). From the four types of CEEs, there were relatively more warm-dry and cold-wet in history and the two have a similar spatial distribution, in north, southwest and southeast China. Moreover, in summer, the frequency and range of warm-dry and cold-wet in China are significantly higher and greater than warm-wet and cold-dry ( Figure S1 in Supporting Information S1). Warm-wet and cold-dry also have a similar distribution situation and frequently occur in west China, especially for annual and summer.
From the linear trend field ( Figure S2 in Supporting Information S1), the frequency of total CEEs in China shows a general and significant positive trend over the whole period, and the maximum positive trends primarily concentrate in northwest China and Tibetan Plateau, mainly contributed from warm-wet and warm-dry. Coldwet and cold-dry show a negative linear trend in the eastern and western marginal areas of China while showing positive linear trends in other inland areas ( Figure S2 in Supporting Information S1).

Change of Interdecadal Characteristics
Cumulative anomalies were calculated to identify the interdecadal transition (Equation 1) (Y. Ren et al., 2022;Sun et al., 2020). The change of cumulative anomaly of CEEs in China can reflect its interdecadal variation and transitional year ( Figure S3 in Supporting Information S1). Based on the analysis of the cumulative anomaly of CEEs, 1950 and 1995 may be selected as the transitional year of the frequency of CEEs, so the whole period is divided into three periods : 1901-1950, 1951-1995 and 1996-2019. Figure 3 shows the spatial distribution of the frequency of annual CEEs in China in three periods. Between 1901 and 1950, the frequency of warm-wet was less than that of the other three events, warm-dry mainly occurred in southwest China and northeast of the Tibetan Plateau, cold-wet and cold-dry occurred in the northeast and eastern China. During the period between 1951 and 1995, the range and frequency of the four types of CEEs were significantly expanded, especially cold-wet, and warm-wet was still less than the other three types of CEEs. During the period between 1996 and 2019, the rapidly decreasing cold CEEs were no longer dominant and warm CEEs were dominant in this period. Warm-wet increased rapidly in the northwest China, which indicates the warming and wetting characteristics in the northwest region of China, consistent with the existing studies (P. Yang et al., 2021). Warm-dry increased significantly in the north and southwest China. For summer ( Figure S4 in Supporting Information S1), notably, from 1950-1995 to 1996-2019, the main occurrence area of warm-dry experienced a large interdecadal northward jump. The frequency of winter warm-dry increased in the southwest China ( Figure S5 in Supporting Information S1). In addition, we also analyzed the spatial distribution of the trend of four types of CEEs in three periods ( Figure S6 in Supporting Information S1). Before 1950, warm CEEs basically showed a positive trend in the whole country, while cold CEEs showed a positive trend in the north and a negative trend in the southeast China. During the second period, the spatial trend of warm CEEs was relatively insignificant compared with the previous period, there was a significant negative trend of cold-wet in northeast China but a positive trend in southwest China. During the third period, warm-wet showed a significant positive trend in large areas of China. Warm-dry showed a significant positive trend in the western China, especially in the southern Tibetan Plateau. Cold-wet showed a negative trend, especially in southwest China. Cold-dry showed a more general positive trend in most regions.
It should be noted that the spatial trend analysis of CEEs in the three periods ignores the change trend between the periods, that is, it basically does not include the sudden change in the transition stage between the two periods but only shows the trend within one period. For instance, in the second period , the trend of cold CEEs is not significant or even decrease. In fact, cold CEEs are more frequent in this period than those in the other two periods, however, cold CEEs do not always increase with time but always maintains a high frequency, so their change trends are not significant at the second period.
We use the probability density function (PDF) to represent the probability distribution and interdecadal variation of the frequency of CEEs (Figure 4). For the warm-wet (warm-dry) in the three periods, the frequency corresponding to the maximum probability of fitting has gradually increased from nearly 750 (1,000) in the first period 1901-1950 to approximately 4,500 (6,300) in the third period 1996-2019. The peak height of the fitted probability curve has gradually descended, and the width gradually expanded, suggesting that the range and variability of the frequency of warm-wet and warm-dry have increased. For cold-wet and cold-dry, the PDF of the first period has been somewhat similar to that of the third period, and the frequency corresponding to the maximum probability has been smaller, nearly half of that of the second period. Nevertheless, the fitting curve of cold-wet in the third period is wider, suggesting that the change range has been larger.

Correlation
The correlations between the temperature field and the precipitation field can be analyzed to explain some climatological characteristics of CEEs ( Figure 5). The frequency of CEEs can be partially explained by the correlation  Figure S1 in Supporting Information S1 suggests that the area with significant positive correlations (western China) is the area with frequent occurrences of warm-wet or cold-dry. The areas with negative correlations with frequent warm-dry or cold-wet, especially in summer, usually belong to summer monsoon regions, because anomaly of summer monsoon usually causes cold-wet or warm-dry weather. However, since strong winter monsoon usually causes cold-dry weather and eastern China belongs to winter monsoon region, there are relatively frequent cold-dry and warm-wet events and positive correlations in eastern China in winter.

Contribution of Different Factors to CEEs
Although the correlation and coupling between temperature and precipitation can partially explain the occurrence of CEEs (X. Yu & Zhai, 2020;Zhou & Liu, 2018), they do not separate the contribution of different scale factors to CEEs. In order to further study the causes affecting the changes of CEEs, we further isolated the contribution of different scale factors (long-term change and low frequency oscillations, variations in interannual variability, and changes in the temperature-precipitation coupling) to the changes of CEEs.
According to the method in Section 3.2, CEEs component excluding the effects of long-term change and low frequency oscillations (Case B) is shown in Figure S7 in Supporting Information S1. The CEEs component due to long-term change and low frequency oscillations (Case A-Case B) is calculated in Figure 6. Time series of cold CEEs component due to long-term change and low frequency oscillations decreased since the late 1970s, while the time series of warm CEEs component increased rapidly since the early 1970s, indicating that long-term change and low frequency oscillations since 1970s contributed negatively to cold CEEs, but was conducive to the increase of warm CEEs. In addition, warm-wet and warm-dry variation to follow global mean air temperature curve rather closely in terms of the interdecadal oscillation and overall trend. They all showed a weak increase trend before the 1940s, no significant trend changes from the 1940s-1970s, and showed a rapid increase trend after the 1980s.
Based on Case B, we calculate the CEEs frequencies assuming independency between temperature and precipitation extreme events (Equation 2) (Case C). Case C reflects contributions from changes in the variability of higher frequencies to CEEs (Figure 7). The results show that the component of warm-wet caused by the variability of  higher frequencies increased since 1970s, followed by cold-wet, while the components of warm-dry and cold-dry changed little. Figure 8 shows the CEEs component caused by the coupling between temperature and precipitation (Case B-Case C). The component represents the contribution of the coupling to the CEEs. Figure 8 shows that in history, the warm-dry and cold-wet caused by the coupling are generally positive values, while the warm-wet and cold-dry are generally negative values, which indicates that the relationship between temperature and precipitation in China is mainly negative coupling. It is worth noting that the amplitude of the four curves in Figure 8 increased since 1970s, indicating that changes in the coupling between temperature and precipitation have an increasing impact on CEEs.
In order to compare the contribution of different factors to the change of CEEs, we calculate the variance contribution rate of the different components to the CEEs in Table S2 in Supporting Information S2. The results show that the contribution of long-term change and low frequency oscillations to the CEEs is the largest, followed by the contribution from changes in the variability of higher frequencies, and the coupling between temperature and precipitation is the least contributing factor of the three factors to the CEEs.

Interdecadal Change of Atmospheric Circulation
There have been some previous studies show that stationary anticyclones (e.g., blocking, associated with enhanced adiabatic heating through increased subsidence), land-atmosphere feedbacks (associated with soil moisture impacts on surface temperature), and large-scale modes of variability (e.g., El Niño-Southern Oscillation, or ENSO for short) can induce the negative dependence between dry and hot extremes, resulting in the concurrence of the two extremes of different time scales (Alizadeh et al., 2020;Hao et al., 2018a;Zscheischler, Van Den Hurk, et al., 2020).
To physically analyze the interdecadal changes of CEEs, characteristics of the atmospheric circulation corresponding to the three periods are presented in Figure S8 in Supporting Information S1. It is suggested that the situation of mean circulation in 500 hPa in three periods is relatively consistent but tends to rise northward ( Figure S8a in Supporting Information S1). One ridge and one trough are in the middle latitudes of Eurasian continent; the low-pressure system in India and the subtropical high in the subtropical western Pacific; a low-pressure system characterized by polar vortex is near the polar region. However, the anomaly fields of the three periods reveal that the abnormal characteristics of the respective circulation systems are not consistent with each other (Figures S8b-S8d in Supporting Information S1). East Asian trough deepened in the south of 50°N between 1901 and 1950, which was conducive to cold air entering into northeast China and the cold CEEs frequently in this region. The anomalous characteristics of the second and third periods are quite different, the height field in 500 hPa has basically showed a negative anomaly when cold CEEs occurred frequently during 1951-1995, while the height field changed to generally significant positive anomaly when warm CEEs dominated during 1996-2015. Figure S8 in Supporting Information S1 reflects that global warming is the main factor for the transition from cold CEEs in the second period to warm CEEs in the third period, but covers the changes of the circulation factors themselves. In order to focus on the changes of the circulation factors in the three periods, it is necessary to subtract the zonal average geopotential height from the original geopotential height. Figure 9 calculates the zonal mean deviation of the geopotential height at 500 hPa during the full period   (Figure 9a) and the anomalies of the zonal mean deviation of each period compared with the full period (Figures 9b-9d).   1901 and 1950, (c) between 1951 and 1995, (d) between 1996 and 2015. sity changes of related circulation factors may not be conducive to the increase of warm CEEs during this period. However, the previous analysis ( Figure 1) has shown that warm CEEs in China during this period increased rapidly and occupied a dominant position. This indicates that global warming caused by climate change is the main factor causing the increase of warm CEEs in China after 1996.

Conclusion
In this study, centurial changes of temperature-precipitation CEEs in China are analyzed. The reasons for the interdecadal change of CEEs are explained in accordance with the correlation between temperature and precipitation, the changes of CEEs components caused by different scale factors and the changes of circulation factors.
In the past 119 years, the sum of four types of temperature-precipitation CEEs in China has been increasing, its linear trend is 0.31 months/decade, and since 2000 it has been nearly three times more than that in the 1900s. For four different CEEs, warm-wet and warm-dry show a significant positive trend, of which the largest is the trend of warm-dry, reaching 0.16 months/decade, while the trend of cold-dry is not significant. Warm-wet and warm-dry have increased rapidly in the past three decades, making the greatest contribution to the growth of CEEs in the past three decades, while cold-wet and cold-dry have made the most significant contribution during the period of 1950s-1980s.
For spatial distribution, the frequency of warm-wet and cold-dry is similar, mainly concentrated in northwest China. The frequency of warm-dry and cold-wet have similar distribution patterns and primarily concentrated in the north, southwest and southeast regions of China.
For interdecadal changes, the four types of CEEs in the past 119 years can basically fall into three period exhibiting different spatio-temporal characteristics, including 1901-1950, 1951-1995 and 1996-2019. The period between 1901 and 1950 is the period of relatively less CEEs, especially warm-wet. Between 1951 and 1995, cold CEEs occurred frequently, cold-wet were the most widely distributed and frequent, followed by cold-dry. The period 1996-2019 represents frequent warm CEEs. In general, for more than 100 years, both frequency and area, warm-wet and warm-dry have been increasing, while cold-wet and cold-dry have been frequent at the second period and then decreased significantly. Consistent with the conclusion of the changing characteristics of CEEs since 1960s in existing researches (X. Wu et al., 2021;X. Wu, Hao, Hao, Li, & Zhang, 2019;Xiao et al., 2020).
In recent 119 years, the CEEs in China were affected by long-term change and low frequency oscillations, the variability of higher frequencies, and temperature-precipitation coupling changes, among which long-term change and low frequency oscillations are the dominant factors of CEEs relative to other scale factors. Our study shows that global warming caused by climate change is the main reason for the rapid increase and dominance of warm CEEs since the 1990s. The East Asian trough, monsoon and subtropical high are vital circulation factors in the interdecadal change of CEEs in China.

Discussion
Most of the existing research focus on the change of CEEs after 1950s, especially warm-dry events (e.g., Feng et al., 2020;Haqiqi et al., 2021;Kong et al., 2020). We used centennial scale data to study CEEs to more clearly analyze the impact of climate change. Compared with previous studies, this study can more completely evaluate variation of CEEs and separate and compare the contributions of different scale factors to CEEs, and further analyze the dynamic factors that affect CEEs. In general, CEEs in China have been increasing over the past one hundred years, and there are two increasing periods: one is the period before 1950s and the other is the last three decades. In the periods of 1900s-1920s and 1950s-1980s, cold CEEs contributed significantly to the total numbers of CEEs. The contribution of warm CEEs is significantly greater than that of cold CEEs in the last three decades, which has also appeared in 1930s-1940s.
Warm-dry has oscillated and risen over the past 120 years, especially the last three decades, which is basically consistent with the previous studies (e.g., X. Wu et al., 2021;X. Wu, Hao, Hao, Li, & Zhang, 2019;Xiao et al., 2020), whereas increasing trend before 1950s is found, especially between 1920s and 1950s. The inter-decadal oscillation and variations are basically consistent with trend of global change. Notable is, warm-wet has maintained a rapid growth rate over the past 20 years and its frequency has actually exceeded the frequency of warm-dry over the last few years, especially in summer. The new findings will be valuable for evaluating and dealing with climate change. The emissions scenario predicts an exponential increase in temperature over the twenty-first century. Therefore, the linear trend analysis in Section 3.3 May not fully explain the trend of CEEs, and we hope to propose more accurate trend analysis methods in this regard in the future.
Moreover, the definition based on the monthly data and the 25th percentile threshold is somewhat flawed and subjective. So far there are many definitions of CEEs. Different thresholds (e.g., Kong et al., 2020;Mukherjee et al., 2020) can influence the evaluating results. There are many other types of CEEs and some new types of CEEs worthy of attention (Zscheischler, Martius, et al., 2020). On the other hand, previous studies on the driving factors of warm-dry have suggested that warm-dry are influenced by human activities (Vogel et al., 2019), urbanization (Yu & Zhai, 2020), atmospheric circulation anomalies (L. Ren et al., 2020) and Sea surface temperature anomalies (Lyon & Barnston, 2004;Mukherjee et al., 2020). But there is less research on the drivers of the other three events. Although previous studies have shown that Atlantic Multidecadal Oscillation as a mode of climate variability is significantly related to the spatial extent of CEEs in China (X. , the specific influencing mechanism is still unclear. It is very necessary to further investigate the specific mechanisms affecting a certain type of CEEs in different regions in depth (X. Wu, Hao, Hao, Li, & Zhang, 2019), especially in summer. Because the process by which summer temperature differences lead to changes in rainfall patterns or rainfall distribution influences overall summer temperature is a complex issue (Seneviratne et al., 2002;Wasco & Nathan, 2019). In addition, capturing early signals and even evaluate corresponding health risks through the mechanism research of CEEs is also an urgent need for improvement in our future work.