Weather pattern conducive to the extreme summer heat in North China and driven by atmospheric teleconnections

Extreme summer heat can have severe socioeconomic impacts and has occurred frequently in North China in recent years, most notably in June–July 2023, when North China experienced the most widespread, persistent, and high-intensity extreme heat on record. Here, typical weather patterns covering North China and its surrounding areas were classified into seven types based on the Cost733class package, and the weather pattern type 4 (T4), characterized by the strengthened ridge and anticyclone anomaly in northeastern China, was found as the most favorable for the occurrence of extreme summer heat in North China (NCSH). Diagnostic and wave activity flux analyses indicate that the Eurasian teleconnection (EAT) pattern from the atmosphere and the Victoria mode (VM) from the ocean are the top two dominant climate drivers of the T4 weather pattern. The empirical models constructed based on the EAT and the VM can effectively simulate the number of days of the T4 weather pattern and the NCSH, respectively. Our results suggest that, with the help of the seasonal forecast from climate models, the EAT and the VM can be used to predict the number of days of the T4 weather pattern and the NCSH for the coming summer, enabling us to protect human health and reduce its socioeconomic impacts through proactive measures in advance.


Introduction
North China (34 • -44 • N, 110 • -120 • E) has a large population and is one of the major political, economic, and agricultural centers in China and even the world, where it is more likely to be severely affected by extreme summer heat due to its higher risk of heat exposure (Ding et al 2010, Zhang et al 2015, Wang et al 2018, Gao et al 2022, Wu et al 2022, Xie et al 2022).Since the onset of the rapid global warming in the 1980s, extreme summer heat events have become more frequent in many regions of the world, including North China, adversely influencing human health and socioeconomic development (Patz 2005, Lobell and Field 2007, Zhang et al 2013, 2023, King and Harrington 2018, Na et al 2019, Xie et al 2019, 2023c, Perkins-Kirkpatrick and Lewis 2020, Yu et al 2021, Wang et al 2022).In June-July 2023, North China, including Beijing, Tianjin, southern Hebei, and northern Henan, experienced the most widespread, persistent, and high-intensity extreme heat on record.In particular, the daily maximum temperature at the Beijing Observatory, a national basic weather station, exceeded 40 • C for three consecutive days from 22 to 24 June 2023, which is the longest continuous and highest intensity of the extreme heat process since the station was established in 1951.Hence, it is of great scientific significance and social value to study the extreme summer heat in North China (NCSH).
Previous studies suggested that various climate factors, such as the western Pacific subtropical high (WPSH; Wang et al 2016, Guan et al 2019, An and Zuo 2021, Li et al 2023), the continental high pressure over Eurasia (Gong et al 2004, You et al 2011, Deng et al 2018), the East Asian jet stream (Chen and Lu 2016, Wang et al 2016, Deng et al 2020), the El Niño-Southern Oscillation (Gao et al 2020), the North Pacific sea surface temperature (SST) variability (Qiu and Yan 2020, Xie et al 2023c), the North Atlantic SST variability (Ding et al 2019a, Xie et al 2023a), the Indian Ocean SST variability (Xie et al 2023c), and the circumglobal teleconnection (Kornhuber et al 2019, Lee et al 2023, Wang et al 2023), are correlated with the variability of the NCSH on interannual to decadal time scales.These studies have enhanced our understanding of the mechanism and external forcing of the NCSH variability.Nevertheless, these were mostly explorations of the NCSH from a long-term climatological perspective.Summer is not only the main occurrence period of the extreme heat in North China, but also the main occurrence period of precipitation, which means that the weather patterns for North China and its surrounding areas change particularly quickly (Xie et al 2023b).Thus, studying the NCSH solely from a long-term climate perspective may obscure some extreme information.
The Cost733class package from the European Union Cost733 project contains a variety of objective or subjective weather pattern classification algorithms that can be used to quickly and objectively classify a large number of weather patterns even on a day-by-day basis (Demuzere et al 2011).In recent years, the Cost733class package has been widely used in the analysis of air pollution, environment and other weather science scales.On the basis of the Cost733class package, Zhang et al (2012) studied the effects of circulation patterns on regional transport pathways and air quality over Beijing and its surroundings, Wang et al (2020) studied the effects of atmospheric circulations on the interannual variation in high levels of particulate matter (PM2.5)concentrations over the Beijing-Tianjin-Hebei region, and Li et al (2022) studied the winter particulate pollution severity in North China and found that it is driven by atmospheric teleconnections.Hence, the Cost733class package may be used for the analysis of the extreme summer heat in North China, where the summer weather patterns shift rapidly, avoiding the signal annihilation caused by analysis on the monthly or even 10 d time scales.
In this paper, the Cost733class package, which is widely used for weather analysis, is applied to the climatological analysis of the NCSH, and the typical weather patterns covering North China and its surrounding areas are classified.Then, the most favorable weather patterns for the NCSH are identified, and the dominant climate drivers are further analyzed.Finally, the empirical models of the number of days of the favorable weather pattern and the variability of the NCSH are constructed using the dominant climate drivers.This paper is organized as follows: section 2 describes the datasets and methods used in this research.Section 3 describes the results, including the weather pattern which is favorable for the formation of the NCSH in section 3.1, the dominant climate drivers of the weather pattern type 4 (T4) in section 3.2, and the prediction ability of the dominant climate drivers for the number of days of the T4 weather pattern and the NCSH in section 3.3.Finally, section 4 gives the conclusions and discussion.

Data and methods
The NCEP-DOE reanalysis 2 (NCEP2) dataset for atmospheric data including geopotential height and winds on a daily basis for the period 1979-2021 with a horizontal resolution of 2.5 • × 2.5 • grid were used (Kanamitsu et al 2002).The European Centre's Medium-Range Weather Forecasts Reanalysis 5 dataset (ERA5; Hersbach et al 2020) on a horizontal resolution of 0.25 • × 0.25 • grid has also been used here to re-examine the results and gives similar results, which are included in the supplementary information.Daily maximum temperatures were derived from the daily meteorological dataset of basic meteorological elements of the China National Surface Weather Station version 3, which contains measured data from 2474 meteorological stations in China, covering the period 1951-2021.The SST data were obtained from the Hadley Centre Sea Ice and SST (HadISST) dataset covering the period 1870-2021 with a horizontal resolution of 1 • × 1 • grid (Rayner et al 2003).In addition, SST data from the NOAA Extended Reconstructed SST (ERSST) version 5 (Huang et al 2017) were also used here to reexamine the results.
In this research, an extreme heat day is defined as the daily maximum temperature equal to or greater than 32.3 • C, which corresponds to the 90th percentile of the summer (1 June-31 August) daily maximum temperature in North China.In this calculation, the daily maximum temperature is first interpolated onto a 0.5 • × 0.5 • grid before the area-weighted average is calculated.The NCSH is calculated based on whether the area-weighted average daily maximum temperature in North China is equal to or greater than the threshold of 32.3 • C during summer.
Here, the typical summertime weather patterns for 1979-2021 were identified by using obliquely rotated principal component analysis in T mode (T-PCA), which is commonly used for circulation pattern classification (Zhang et al 2012, Wang et al 2020, Li et al 2022).In this research, the T-PCA method in the Cost733class software package from European Union Cost733 project was applied to identify the typical summer circulation patterns in T Xie et al North China during the period 1979-2021.The classification was based on the climatic factors of the 500 hPa geopotential height anomalies, the 200 hPa latitudinal wind anomalies, and the 850 hPa meridional wind anomalies.We have tried to classify the summertime weather patterns for 1979-2021 into 6, 7 and 8 categories and selected the results classified into 7 categories (types 1-7) for further analysis by evaluating the explained variation, the pseudo F-values and the identified conducive weather pattern types.The pseudo F-values can be calculated as follows: where SS A is the sum of squared dissimilarities between groups (or categories), SS W is the sum of squared dissimilarities within groups, k is the number of groups, and the N is the total number of objects.
The larger pseudo F-value represents a relatively larger difference between the categories.The pseudo Fvalue is 9.88 when the weather patterns are classified into 6 categories, while it increases to 22.05 for 7 categories and 24.13 for 8 categories.The classification in the study also examines the differences in the daily maximum temperatures and the number of extreme heat days corresponding to the various weather patterns when classified into different numbers of categories, and it was found that the classification into 7 categories was the most effective in classifying weather patterns that favored the occurrence of high temperatures.Considering the pseudo F-value and relationships between the weather patterns and maximum temperature, 7 categories are selected for further analysis.
The Victoria mode (VM) index is constructed as follows: where SSTa(x, y, cor(SSTa, T4)) is the spatial pattern on the number of days of the type 4 (T4) weather pattern correlated with the summer SST anomalies over the North Pacific (0 for the period 1979-2021, as shown in figure 5(a).SSTa(x, y, t) is the SST anomalies over the North Pacific for the period 1979-2021.The result multiplied by −1 is because the above correlation pattern SSTa(x, y, cor(SSTa, T4)) is usually considered to be the negative VM pattern.Following Di Lorenzo and Mantua (2016), this procedure generates the historical time series for the strength of the spatial pattern above.

Weather pattern favorable for formation of the NCSH
Figures 1(a)-(g) shows the composite maps of the geopotential height anomalies at 500 hPa and the horizontal wind anomalies at 850 hPa over North China and its surrounding areas corresponding to types 1-7 (here defined as T1-T7, respectively) of the summertime typical weather patterns during the period 1979-2021.It can be seen that the weather pattern of the T4 in figure 1(d) is the most favorable for the formation of the extreme summer heat in North China.The detailed analyses are as follows: the occurrence of the extreme summer heat in North China is usually accompanied by the positive geopotential height anomalies and south wind anomalies over North China, corresponding to the T4 weather pattern (figure 1(d)).This is also consistent with previous study showing that the extreme summer heat in North China is closely linked to the northwestern extension of the WPSH (Luo and Lau 2018).Figure 1(h) shows the percentage of total days for T1-T7 during the summer of 1979-2021.The number of days of T1-T7 weather patterns accounted for 14.9%, 19.0%, 16.8%, 14.1%, 14.7%, 10.0%, and 10.4%, respectively.Meanwhile, the weather patterns derived from the ERA5 data are also given in figure S1 (supplementary information), which are in good agreement with those obtained from the NCEP2 data in terms of spatial distribution.It can also be seen that the variability of the ERA5 data is only about half that of the NCEP2 data, which has been noted in previous studies (Huang et al 2021, Kim and Lee 2022) and may be related to differences between the datasets.
The mean values of the daily maximum temperature anomalies in North China corresponding to each weather pattern were shown in figure 2(a).It can be seen that the mean value of daily maximum temperature anomalies relative to the 1979-2021 summer mean climatology associated with the T4 weather pattern reached 1.044 • C, and at this point, the values associated with the other weather patterns are negative.The values calculated based on the day-by-day maximum temperature climatology of summer 1979-2021 are similar, and the mean value of daily maximum temperature anomalies in North China associated with the T4 weather pattern reached 1.046 • C, much higher than the values for the other weather patterns.These are also consistent with the analysis in figure 1, suggesting that the weather pattern T4 is most favorable to the occurrence of the extreme summer heat in North China.And a similar conclusion can be drawn from figure 2(b), the number of the extreme summer heat days in North China associated with the T4 weather pattern and its ratio relative to the number of days of the corresponding weather pattern are also significantly higher than those associated with the other weather patterns.
Next, the interannual series of days associated to the T4 weather pattern and the NCSH are shown in figure 3(a).It can be seen that there is some consistency between the two series, and the correlation coefficient exceeds 0.43, which passes the 99% significance level test.Meanwhile, the time series of the NCSH and the frequency of the T4 weather pattern both show an increasing trend, which may be related to the long-term climate warming.The interdecadal variabilities of the above two series obtained based on the 13 year low-pass Gaussian filtering are shown in figure 3(b), and it can be seen that they have a consistent interdecadal variability with a correlation coefficient of 0.92 (significant above the 99% confidence level based on the effective degrees of freedom).The series for the remaining part after subtracting the corresponding interdecadal variabilities are shown in figure 3(c), and the correlation coefficient between them exceeds 0.40, which also passes the 99% significance level test.These suggest that the relationship between the NCSH and the number of days of the T4 weather pattern exists on both interdecadal and interannual scales, and the occurrence of the T4 weather pattern is favorable to extreme heat in North China.

Dominant climate drivers of the NCSH
Here, the dominant climate drivers including atmospheric and oceanic are analyzed, and it is found that an atmospheric teleconnection here defined as the Eurasian teleconnection (EAT) pattern located at mid-to-high latitudes in Eurasia and an SST pattern Figure 4(a) shows the correlation map of the interannual series of days associated to the T4 weather pattern with the 500 hPa geopotential height anomalies over East Atlantic to East Asia in summer for the period 1979-2021.It can be seen that there are four significant centers of positive and negative correlation in Eurasia, located in Central-Western Europe, the Eastern European Plain on the western side of the Ural Mountains, Central Siberia, and North China, respectively, showing a 'negative-positive-negativepositive' latitudinal wave train structure here defined as EAT pattern.The positive geopotential height anomalies over North China generally correspond to subsidence warming, which is favorable to the occurrence of extreme summer heat over North China.Further investigation revealed that the EAT pattern corresponds to the first Empirical orthogonal function (EOF1) mode of the detrended geopotential height anomalies at 500 hPa over the North Atlantic to East Asia in summer (figure 4(b)).These suggest that the EAT may be an important climatic driver of the T4 weather pattern.Ding et al (2019a) found that a similar teleconnection could influence the extreme summer heat in North China.And the horizontal stationary wave activity flux associated with the T4 weather pattern (figure 4(c)) and the above EOF1 mode (figure 4(d)) further confirm that the EAT pattern is an important driver of the T4 weather pattern.These results can also be obtained based on the  EAT can influence the two western polaritons of the EAWPQ pattern, but the two eastern polaritons are unexplained, implying that there may be signals, such as on the Pacific Ocean, that can drive the T4 weather pattern in addition to the EAT. Figure 5(a) shows the correlation map of the interannual series of days associated to the T4 weather pattern with the SST anomalies over North Pacific in summer for the period 1979-2021.VM mode is characterized by a band of positive SST anomalies extending from off California across the Pacific to the western Bering Sea and a band of negative SST anomalies extending from the central North Pacific to the coast of Asia, proposed by Ding et al (2015).From figure 5(a), it can be seen that there is an SST pattern similar to the negative VM, with a potential possible link to the T4 weather pattern.The constructed VM indices are shown in figure 5(b), and it can be seen that the constructed VM indices (after multiply by −1) have a stronger consistency with the series of days associated to the T4 weather pattern, and the correlation coefficient between the two reaches −0.56 (significant above the 99% confidence level).This suggests that the VM from the ocean may be another dominant climate driver of the T4 weather pattern in addition to the EAT from the atmosphere.
As shown in figure 5(c), a negative VM index corresponds to positive geopotential height anomalies at 500 hPa for almost all of the North Pacific north of 30 • N, suggesting that the appearance of the negative VM favors the occurrence of anomalous high pressures in the above region, especially in the Sea of Japan to the eastern part off Japan, which would block the propagation of the EAT downstream, causing the atmosphere to accumulate and form anomalous positive geopotential height anomalies in the north to northeast China region, thus favoring the occurrence of the T4 weather pattern.From another perspective, these are also consistent with the previous study that the northwestern extension of the WPSH favors the extreme summer heat in North China (Luo and Lau 2018).The wave activity flux and stream function (figure 5(d)) associated with the negative VM can show a similar finding that the atmospheric circulation from north to northeast China is also influenced by the VM, as the wave train propagating near the Kuroshio area to the Japan Sea region can block the propagation of the

Prediction ability of the dominant climate drivers
It has been found that the EAT from the atmosphere and the VM from the ocean are the dominant climate drivers of the T4 weather pattern.Further analysis shows that the correlation coefficient between the VM and EAT indices is 0.03 for the period 1979-2021, indicating that they are independent of each other.
Here an empirical model for the number of days of the T4 weather pattern will be constructed based on the EAT and the VM as follow: where T4 denotes the modeled number of days for the T4 weather pattern, t is the time in years, the EAT index is represented by the corresponding principal component series (PC1) of the EOF1 mode of the geopotential height anomalies in figure 4(b), the coefficients a, b, and c are 1.84, −3.53, and 0.0065, respectively, and are obtained empirically by multiple linear regression.Figure 6(a) shows the observed and modeled number of days for the T4 weather pattern, and it can be seen that the above two series have similar variation characteristics, with a correlation coefficient of 0.63 (significant above the 99% confidence level), indicating that the number of days of the T4 weather pattern can be effectively simulated based on the EAT and the VM.Since the T4 weather pattern favors the occurrence of extreme summer heat in North China, similarly, an empirical model of the NCSH was also constructed based on the EAT and the VM, and the corresponding scatterplot is shown in figure 6(b).It can be seen that the empirical model of the NCSH constructed based on the EAT and the VM can also reproduce the NCSH to some extent, and the correlation coefficient between the modeled NCSH and the observed NCSH exceeds 0.48, which passes the 99% significance level test.In addition, the average value of the NCSH is 6.7 d per year for the period 1979-2021, with 16 years of the NCSH values above the average (6.7 d), of which this empirical model is able to predict 12 years with 75% accuracy.Therefore, with the help of seasonal forecasts from climate models, the EAT and the VM can be used to predict the future number of days of the T4 weather pattern and the NCSH.

Conclusion and discussion
In this paper, firstly, the weather patterns in North China and its surrounding areas were classified into seven types based on the T-PCA method in the Cost733class software package, and it is found that the T4 weather pattern is the most favorable weather pattern for the occurrence of extreme summer heat in North China.Then, the dominant climate drivers of the T4 weather pattern were explored, and the EAT from the atmosphere and the VM from the ocean are the dominant climate drivers of the T4 weather pattern.Finally, two empirical models were constructed based on the EAT and the VM for the number of days of the T4 weather pattern and the NCSH, respectively.And the above empirical models can effectively simulate the number of days of the T4 weather pattern and the NCSH, respectively.
Here, the weather pattern that favors the occurrence of extreme summer heat in North China was analyzed and two dominant climate drivers in summer were identified for this pattern.Some previous studies have also suggested that the Pacific Decadal Oscillation (PDO), Atlantic Multidecadal Oscillation (AMO), IPOD, and other climate factors may be related to the occurrence of extreme summer heat in the Chinese region.For example, Zhang et al (2020) found the PDO and AMO can influence the interdecadal variability of the extreme high temperatures in North China.Interdecadal variability in the NCSH and the frequency of the T4 pattern was also found here, but instead of exploring this specifically by performing a scale separation, the dominant climate drivers of the T4 weather pattern was investigated by putting the interdecadal and interannual variability together.In the future, further study of the relevant antecedent climate factors by performing a scale separation or evaluation and revision of climate model prediction products for the EAT and the VM will be able to effectively improve the prediction capability of the NCSH.

Figure 1 .
Figure 1.(a)-(g) Composite maps of the geopotential height anomalies (shaded; units: gpm) at 500 hPa and the horizontal wind anomalies (vectors; units: m s −1 ) at 850 hPa corresponding to the types 1-7 of the weather patterns during summer 1979-2021.The areas in the blue box denote North China (34 • -44 • N, 110 • -120 • E).(h) Percentage of total days for each weather pattern during summer 1979-2021.

Figure 2 .
Figure 2. (a) Mean anomalies of the daily maximum temperatures relative to the mean of 1979-2021 summer (red bar; units: • C) and the mean of 1979-2021 summer day-by-day (orange bar; units: • C) for each weather pattern.(b) As in (a), but for the cumulative number of heat days (red bar; units: days) and the ratio (orange bar) of the number of heat days relative to the number of days corresponding to each weather pattern.

Figure 3 .
Figure 3. (a) Time series of the NCSH (red line; units: days) and the number for T4 days (black line; units: days) for the period 1979-2021.(b), (c) As in (a), but for the corresponding decadal variability and the remainder, respectively.

Figure 4 .
Figure 4. (a) Correlation map of the time series of T4 with the 500 hPa geopotential height anomalies over East Atlantic to East Asia in summer for the period 1979-2021.The dotted area indicates the correlation coefficient is significant at the 95% confidence level.(b) Spatial pattern of the first EOF mode for the 500 hPa geopotential height anomalies in summer for the period 1979-2021, which is displayed as the normalized 500 hPa geopotential height anomalies in summer regressed on the normalized PC.Horizontal stationary wave activity flux at 500 hPa associated with (c) the T4 and (d) the PC1, based on T-N wave activity flux.Contours and vectors indicate the velocity potential and divergent wind anomalies, respectively.

T
Figure 5. (a) Correlation map of the time series of T4 with the SST anomalies over North Pacific in summer for the period 1979-2021.The dotted area indicates the correlation coefficient is significant at the 95% confidence level.(b) Time series of the T4 (black line) and the constructed VM index (red line).(c) As in (a), but for the correlation of the 500 hPa geopotential height anomalies and the VM index, and multiplied by −1.(d) Horizontal stationary wave activity flux at 500 hPa associated with the negative VM, based on T-N wave activity flux.Contours and vectors indicate the velocity potential and divergent wind anomalies, respectively.

Figure 6 .
Figure 6.(a) Time series of the observed (black line) and modeled (red line) T4 for the period 1979-2021.(b) Scatterplot of the modeled NCSH versus the observed NCSH for the period 1979-2021.Each black dot represents a year, and black line is a linear fit.