Possible Relationship Between Korean Heat Wave and Western North Pacic Tropical Cyclone Genesis Frequency

This study conducted a correlation analysis between tropical cyclone genesis frequency (TCGF) in the western North Pacic (WNP) and heat wave days (HWD) in Korea during July and August for 46 years (1973-2018) and we found showed a strong positive correlation between them. This implied that the higher the TCGF in the WNP during July and August, the higher the HWD in Korea becomes. To examine the cause of the statistically signicant positive correlation between the TCGF during July and August in the WNP and the HWD in Korea, 15 years with the highest frequency and the lowest frequency out of the 46 years in the TCGF time series were selected and dened as high TCGF years and low TCGF years, respectively. An analysis of the difference in 2m air temperature (Air2m) between the two groups showed that in the mid-latitude region of Asia, the Air2m was higher during the high TCGF years. Thus, it could be seen from this analysis that the increase of HWD during the high TCGF years is likely to occur in the entire mid-latitude region of East Asia as well as in Korea. According to the difference in atmospheric circulations between the two groups, in all layers of the troposphere, anomalous anticyclonic and cycloninc circulations were strengthened in the mid-latitude region of East Asia and in the WNP, respectively, which was similar to the Pacic-Japan (PJ) teleconnection pattern. Furthermore, the anomalous anticyclone strengthened in the mid-latitude region of East Asia was associated with the weakening of the East Asian summer monsoon, and the anomalous cyclone strengthened in the WNP was associated with the WNP summer monsoon. The difference in the vertical meridional circulation averaged over the longitude range where Korea is located showed that anomalous upward and downward ows were strengthened in the WNP and in the latitude where Korea is located, respectively. This implied that the local Hadley circulation was strengthened during the high TCGF years. An analysis of the difference in the mean sea surface temperature during July and August showed that the eastern Pacic (EP) La Niña was strengthened during the high TCGF years. To determine the cause of the formations of anomalous anticyclones in the mid-latitude region of East Asia and in the WNP during the high TCGF years, the 500 hPa wave activity ux was analyzed. The wave activity ux originated from the North Atlantic, passed through the Scandinavian Peninsula, the North coast of Russia, and East Siberia before reaching Korea and the WNP. This spatial distribution was similar to the Scandinavia teleconnection pattern. Therefore, we conclude that the anomalous anticyclone formed in the mid-latitude region of East Asia and the anomalous cyclone formed in the WNP during the high TCGF years are associated with the Scandinavia teleconnection pattern.


Introduction
Heat wave and tropical cyclone (TC) are representative meteorological disasters that cause serious damage to South Korea. According to the Annual Disaster Report for recent two years, the number of deaths due to heat wave was 48 in 2018 and 30 in 2019 (total deaths caused by meteorological disasters in each year were 53 and 48, respectively), recording the biggest casualty in South Korea. According to Kim et al. (2014), the average number of deaths by heat wave was 21 per year and the number of deaths by heat wave increased exponentially with the increase of the heat wave days during the last two decades. TC is also one of the second most damaging natural disasters after heat wave in South Korea.
Moreover, because TCs are accompanied by destructive wind and heavy rainfall, even one TC event can cause enormous socioeconomic damage. In fact, TCs have long been the most devastating natural phenomenon on South Korea (Ministry of the Interior and Safety (MOIS), 2017; 2018; 2019).
The rise of global temperature and sea surface temperature (SST) is increasing severe weather phenomena such as heat wave and TC, which are likely to become more frequent in the future (IPCC, 2019). Consequently, the Asian region including South Korea is actively conducting related research, and there is concern that severe weather phenomena will be strengthened due to climate change. However, existing studies mostly research these two phenomena independently (heat wave: Choi  There are almost no studies on the complex interaction between TC and heat wave in South Korea. However, the correlations of these two phenomena were estimated by previous studies. TC is a mesoscale weather phenomenon, but the development of a TC depends on the effect of the surrounding synoptic-scale environment after it occurs in the tropical western Paci c and moves north. Because the TC moves along the ridge of the WNPSH, therefore, the expansion of WNPSH is also an important factor in addition to the traditional factors affecting TCs in lower latitude (e.g., monsoon gyre/trough, high SST, Heat wave also has been proven to develop through interactions with synoptic-scale weather events in many studies ( (2020b) distinguished three typical weather types that cause heat wave. Heat wave mainly occurred when the edge of WNPSH was located in Korea, in particular. By contrast, heat waves caused by orographic effect were frequent and showed the lowest number of occurrence days in June (Lee et al. 2020b). Choi et al. (2020) and Lee et al. (2019) showed that teleconnection and global warming related to tropical convective activity contributed to heat waves in 2016.  showed that the deep convection phenomenon of South China Sea (SCS) provides a condition that causes heat wave through Rossby wave train, suggesting that tropical forcing is another major cause of heat wave.
In addition, we need to pay attention to the fact that both heat waves and TCs occur during the peak season of July and August. It is very unlikely to be actually affected by compound phenomena because South Korea is in uenced by only 1 or 2 TCs in this season (only 4 TCs occurred in 1979-2017 in the study of Mattews et al. (2019) as well). However, unlike other seasons, most TCs that occur in July and August move north to the mid-latitude and the Northeast Asia, and cause indirect in uence on the Korean Peninsula in various forms such as extreme wind, storm, and heavy rainfall even though they do not directly land in Korea (KMA 2011). Moreover, recently, the TC genesis/activity region has been found to gradually shift toward the north (Kossin et al. 2014;Moon et al. 2015;Choi et al., 2015). Thus, the possibility of TC directly or indirectly affecting the Korean Peninsula increases in a relatively short period of time after the TC genesis should also be considered.
This study analyzed the correlation between tropical cyclone genesis frequency (TCGF) in the WNP and heat wave days (HWD) in Korea during July and August and investigated the cause in large-scale environments and atmospheric circulations. We suggest that heat wave and TC are closely correlated because both are in uenced by summer synoptic pattern. Furthermore, since both are hazardous weather events that cause much damage, related research needs to be conducted to prepare for potential havoc in the future. Section 2 introduces the data and analysis method. Section 3 presents the correlation between heat wave in Korea and TC genesis in the WNP and a possible mechanism. Finally, Sect. 4 summarizes the results of this study.

Data
This study used the data of the Regional Specialized Meteorological Center (RSMC)-Tokyo Typhoon Center to analyze TCs. In addition, to analyze large-scale environments and atmospheric circulation, the National Centers for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) Reanalysis dataset (Kalnay et al., 1996) was used. As for monthly global SST data, the National Oceanic and Atmospheric Administration (NOAA) Extended Reconstructed SST analysis (version 3) was used (Smith et al. 2008). The NOAA interpolated outgoing longwave radiation (OLR) was used to examine the state of convective activity (Liebmann B and Smith CA 1996). The precipitation data utilized is Global Precipitation Climatology Project (GPCP) version 2.3 (Adler et al. 2003).
Furthermore, this study used the surface air temperature (SAT), precipitation, and Palmer Drought Severity Index (PDSI) data observed at 58 weather observation stations in Korea. This data can be obtained from the website of Korea Meteorological Administration (KMA) (https://www.kma.go.kr), and the spatial distributions of the 58 weather observation stations are shown in Fig. 1a. Since the Ulleungdo and Jeju regions show unique weather characteristics of islands, the observation data in these two regions were excluded from this study. Since the number of weather observation stations in Korea has rapidly increased since 1973, this study used data since 1973. The data of HWD and tropical night in Korea are o cially provided on the KMA website (https://data.kma.go.kr/climate/). KMA de nes HWD as the number of days with daily maximum temperatures of 33 o C or higher, and tropical night days as the number of days with nighttime minimum temperatures of 25 o C or higher. Since the HWD mainly appears high in July and August as shown in Fig. 1b, this study used July-August mean data.
This study also used the East Asian summer monsoon (EASM) index (EASMI) of Li and Zeng (2002, 2005, and the western North Paci c summer monsoon (WNPSM) index (WNPSMI) provided on the website of the Asia-Paci c Data Research Center (APDRC) of the University of Hawaii (http://apdrc.soest.hawaii.edu/projects/monsoon/seasonal-monidx.html).

Methodology
The two-tailed Student's t test was used to determine the signi cance of the results of this study (Wilks 1995). The horizontal components of Stationary Rossby wave propagation were calculated using the formula of Takaya and Nakamura (2001). The Paci c-Japan index is de ned as a difference between two grid points (40°N, 150°E and 25°, 125°E) for 850 hPa geopotential height anomalies (Wakabayashi and Kawamura 2004;Kawamura and Ogasawara 2006).
In this study, TC was de ned as TCs that occurred in the WNP and developed more than tropical depression (TD). Furthermore, extratropical cyclone (EC) was also included in this study because it causes signi cant human and property damage in the mid-latitude region of East Asia after a TC is changed to an EC. trend of the TCGF is statistically insigni cant because it shows little change, but the HWD in Korea shows an increasing linear trend due to the effect of global warming. This increasing linear trend is signi cant at the 95% con dence level. A close observation of these two time-series reveals an in-phase trend between the two variables. Thus, a correlation analysis between the two variables shows a strong positive correlation of 0.57. This correlation is signi cant at the 99% con dence level. This means that the higher the TCGF in the WNP during July and August, the higher the HWD in Korea. However, this correlation may change if the linear trend is removed from the two variables. Therefore, the correlation was analyzed again after removing the linear trend from the two variables. However, it did not show a signi cant difference from the original correlation (Corr = 0.55, which is signi cant at the 99% con dence level). Furthermore, the correlation between tropical night in Korea and the TCGF in the WNP during July and August also shows a high correlation of 0.60 (not shown). This correlation is also signi cant at the 99% con dence level.
Meanwhile, to examine in more detail on the correlation between HWD in Korea and TCGF in the WNP during July and August, the WNP was divided into the following four areas (Fig. 3a): Northwest (NW) area, Southwest (SW) area, Northeast (NE) area, Southeast (SE) area. This distinction was based on 16°N and 141°E, according to as the average TC genesis location for 46 years. Among these four areas, the area that showed the highest correlation between TCGF and HWD in Korea during July and August is SE area with a correlation of 0.63. This correlation is signi cant at the 99% con dence level. The next is NW area in which the TCGF during July and August in this area and the HWD in Korea have a correlation of 0.52. This correlation is also signi cant at the 99% con dence level. The correlation between the two variables in the other two regions is low at 0.25 or lower. This correlation in these two regions is signi cant at 90% con dence level. Consequently, the two variables tend to have a high correlation only in the NW and SE areas in the WNP. This appears to be because a monsoon trough develops from the northwest to southeast direction in the tropical and subtropical WNP in general, and TCs tend to occur along this monsoon trough (Matsuura et al. 2003). Therefore, the time series of TCGF during July and August in the NW and SE areas were analyzed (Figs. 3b and 3c). The time series of TCGF in the NW area shows the lowest frequency in the late 2000s and tend to increase rapidly since 2010 (Fig. 3b). However, the linear trend changed very little during the total analysis period of 46 years. However, a correlation analysis was conducted after removing the linear trend from the two variables because the linear trend of the HWD tended to increase in Korea, but it was not much different from the original correlation (Corr = 0.51, which is signi cant at the 99% con dence level). The TCGF time series in the SE area has a strong interannual variation in general, but a weak interdecadal variation can be seen as well. They also show a rapidly increasing trend since the early 2010s (Fig. 3c). This time series also shows very little change in linear trend, and the correlation did not change even when the linear trend was removed from the two time series (Corr = 0.62, which is signi cant at the 99% con dence level).

Spatiotemporal distribution of air temperature and rainfall
To examine the reason for the high positive correlation between the TCGF during July and August in the WNP and the high HWD in Korea from the above analysis results, 15 years with the highest frequency in the TCGF time series in the WNP during July and August out of the 46 years and 15 years with the lowest frequency out of the 46 years were selected and de ned as high TCGF years and low TCGF years, respectively ( Table 1). The selected 30 years have a number of samples large enough to occupy about 70% of the total analysis period. Then the mean difference between high TCGF years and low TCGF years was analyzed. The mean HWD in Korea for 46 years is 10.9 days, and there are only four years among the high TCGF years that did not exceed this mean HWD (1981,1989,1992,2002) and only four years among the low TCGF years that did not exceed this mean HWD (1983,1995,2008,2010). Hence, the mean HWD of high TCGF years is 15.4 days, but the mean HWD of low TCGF years is 8.0 days, and the difference in the mean HWD between the two groups is 7.4 days. This mean difference is signi cant at the 95% con dence level.  Meteorologically, this region is known to have relatively high temperatures during summer in the country (Ko et al. 2006). The daily time series of the difference between the two groups for SAT indicate that the high TCGF years have higher values from January to August whereas the low TCGF years have higher values from September to November (left panel of Fig. 4d). The largest variations in the SAT during the year appear in these two periods.
The reason that the SAT is high (low) may be because there are few (many) clouds in the atmosphere and there is a low (high) possibility of precipitation. Hence, the spatial distribution of total rainfalls in July and August for the two groups were analyzed (right panel of Figs. 4a and 4b). As expected, the spatial distributions of the two groups appear similar overall. The rainfall is relatively large in the northern region of Korea, whereas the rainfall is low in the eastern region and the west coast region. During the high TCGF years, the rainfall in the northern region of Korea is only 600mm (right panel of Fig. 4a), but during the low TCGF years, it is higher than 700 mm in the region (right panel of Fig. 4b).
In particular, the rainfall in the eastern region of Korea is lower during the high TCGF years. The difference between the two groups shows negative anomaly in most areas excluding the northeast coastal region and part of the southeast coast region (right panel of Fig. 4c). This means that there were large rainfalls during the low TCGF years than during the high TCGF years. The largest difference appears in the northeast and southeast regions of Korea. The daily time series of the difference in rainfall between the two groups showed strong negative anomaly during July and August, indicating that the HWD could increase in Korea during the high TCGF years (right panel of Fig. 4d).

Large-scale environments and atmospheric circulations
This study rst analyzed the difference in 2m air temperature (Air2m) between the two groups in Asia (Fig. 5). In the mid-latitude region of East Asia (30°-40°N), the Air2m is higher during the high TCGF years, whereas the spatial distribution is higher during the low TCGF years in other regions. Thus, it can be seen from this analysis that the increase of HWD during the high TCGF years is likely to occur in the entire midlatitude region of East Asia as well as in Korea. Then the East Asia was divided into Northeast (NE) Asia area (30°-40°N, 115°-140°E) and South China (SC) area (20°-30°N, 110°-120°E). The time series of Air2m area-averaged in each region and the TCGF time series during July and August in the WNP are shown in Figs. 2b and 2c. First, the Air2m time series area-averaged in the NE Asia area shows strong interannual variations and weak interdecadal variations (Fig. 2b). The linear trend of this time series shows an increasing linear trend, which is signi cant at the 95% con dence level. This increasing trend may be due to the effect of global warming. Furthermore, the two time series shows an in-phase trend, and an analysis of the correlation between these two variables showed a positive correlation of 0.43. This correlation is signi cant at the 99% con dence level. This implies that if the TCGF in the WNP increases during July and August, there is a possibility that the HWD in the NE Asia area can increase as well. Since the two time series show a distinctly increasing linear trend, the correlation may be changed if the linear trend is removed from the two time series. Therefore, the correlation was reanalyzed after removing the trend from the two time series. As a result, the correlation became higher than the original correlation (Corr = 0.46, which is signi cant at the 95% con dence level). The Air2m time series area-averaged in the SC area shows interannual and interdecadal variations (Fig. 2c). This time series also shows an increasing linear trend due to the effect of global warming, and this increasing linear trend is signi cant at the 95% con dence level. Furthermore, since these two time series show a trend of out-of-phase, the correlation between the two variables was analyzed. The result showed a negative correlation of -0.47 and this correlation is signi cant at the 99% con dence level. This indicates the possibility of decreasing HWD in the SC area where the TCGF increases during July and August in the WNP. When the correlation was reanalyzed after removing the linear trend from the two time series, the negative correlation was strengthened to some extent (Corr = -0.49, which is signi cant at the 99% con dence level).
Meanwhile, the differences in air temperature between the two groups were analyzed at lower-level (850 hPa), middle-level (500 hPa), and upper-level (300 hPa) (left panel of Fig. 6). At 850 hPa, a warm anomaly is located in the mid-latitude region of East Asia, at the center of which Korea exists (left panel of Fig. 6a). The warm anomaly in Korea is signi cant at the 95% con dence level. In addition, a weak warm anomaly appears in the WNP in which TCs occur. Although weak, this warm anomaly must have in uenced the increase of TCGF. At 500 hPa, a warm anomaly located in the mid-latitude region of East Asia was shifted further north (left panel of Fig. 6b). Thus, the southern region of Korea is included in the cold anomaly, but it is not statistically signi cant. In the WNP, the area of warm anomaly was also shifted to the north. At 300 hPa, the warm anomaly in the mid-latitude region of East Asia moved much further to the north and the entire Korea shows a cold anomaly (left panel of Fig. 6c). The warm anomaly in the WNP is a little strengthened than at 850 and 500 hPa.
If the relative humidity is also high when the air temperature is high, people tend to feel more uncomfortable. Thus, the difference in relative humidity between the two groups was analyzed (middle panel of Fig. 6). At all levels, a negative anomaly is located in the Korean Peninsula. This is believed to be due to the evaporation of water vapor as the subsidence was strengthened by the formation of anomalous anticyclone in Korea. Meanwhile, in the WNP where TCs were generated, the positive anomaly appears at every level. This is an important factor in increasing the TCGF. Gray (1975) analyzed relative humidity at lower-level and middle-level is one of the important factors among the six physical parameters that in uence TC genesis.
To examine the variations of the spatial distributions of air temperature and relative humidity according to the level analyzed above, the differences in atmospheric circulations at the three levels were analyzed (right panel of Fig. 6). In all the layers of the troposphere, anomalous anticyclonic circulations are strengthened in the mid-latitude region of East Asia and anomalous cyclonic circulations are strengthened in the WNP. This is similar to the Paci c-Japan (PJ) teleconnection pattern (Nitta 1986(Nitta , 1987(Nitta , 1989. The PJ pattern is an important pattern that in uences the East Asian climate in summer (e.g., rainfall and air temperature) and TC genesis ). Therefore, the correlations of the PJ index with the HWD in Korea and the TCGF in the WNP during July and August were analyzed (not shown). The results showed positive correlations of 0.43 and 0.41, respectively, and these correlations were signi cant at the 99% con dence level. This suggests that in Korea, subsidence developed and water vapors evaporated with the strengthening of the anomalous anticyclone, resulting in negative relative humidity at every level. Moreover, as it moves from the lower level to the upper level, the anomalous anticyclone in the mid-latitude of East Asia is shifted to the north. Consequently, as analyzed above, the warm anomaly in the mid-latitude region of East Asia was shifted to the north toward the upper level. Meanwhile, the anomalous cyclone strengthened in the WNP at the three levels became an important background in the increase of TCGF.
The increase or decrease of HWD in Korea is associated with the locations of western North Paci c subtropical high (WNPSH) and Tibetan high (TH) (Wu et al. 2012). Hence, the degree of development of WNPSH and TH was analyzed for the two groups (left panel of Fig. 7). Here, WNPSH (red line) and TH (blue line) are de ned as areas larger than 5,870 gpm and 12,480 gpm. The WNPSH is strengthened northwest to the western sea of Korea during the high TCGF years (left panel of Fig. 6a), whereas during the low TCGF years, it is expanded southwest to the southeastern region of China (left panel of Fig. 6b). TH is located a little north in the high TCGF years than in the low TCGF years, and is further expanded to the east during the low TCGF years. Hence, two high pressure systems are overlapped in Korea in high TCGF years. Thus, it can be seen that the HWD increased as the solar radiation increased due to the strengthening of subsidence as the high pressure systems were located in the upper and lower layers of the troposphere in Korea. Furthermore, during the high TCGF years, the two high pressure systems shifted to the north and low pressure systems were strengthened in the WNP, thus increasing the TCGF.
To examine this, the 850 hPa geopotential height mean eld was analyzed for the two groups (right panel of Fig. 7). Here, the dashed and solid lines indicate the monsoon trough and the ridge of WNPSH. The ridge of WNPSH is expanded northwest to the northeastern region of China in high TCGF years (right panel of Fig. 7a), but in low TCGF years, it is expanded west to the eastern coast of central China (right panel of Fig. 7b). As a result, in the high TCGF years when the ridge of WNPSH is shifted more to the north, the monsoon trough is strengthened east to 145°E (right panel of Fig. 7a). By contrast, in the low TCGF years, the monsoon trough is weakened only to 130°E. This suggests that the strengthening of the ridge of WNPSH and monsoon trough in high TCGF years is associated with the increase of HWD in Korea and the increase of TCGF in the WNP during July and August.

Local Hadley circulation
Meanwhile, the anomalous anticyclone strengthened in the mid-latitude region of East Asia and the anomalous cyclone strengthened in the WNP can be associated with local Hadley circulation. To examine this, the difference in the vertical meridional circulation averaged over the longitude range of 120°-130°E where Korea is located was analyzed (Fig. 8a). Anomalous upward ows are strengthened at 10°-30°N where the WNP is located and anomalous downward ows are strengthened at 30°-40°N where Korea is located. In particular, the center of anomalous upward motion in the WNP is located at 10°-20°N, and this is signi cant at the 95% con dence level. Furthermore, the center of anomalous downward motion at 30°-40°N is distributed at 30°-35°N where South Korea is located,which is placed in a region of the 95% con dence level. This anomalous vertical structure between the WNP and Korea means that the air rising in the WNP subsides in Korea. Therefore, the TCGF can increase in the WNP and can be an important background in which the HWD can increase in Korea. As a result of this, in the vertical structure of air temperature, a warm anomaly appears in the lower level of 30°-40°E where Korea is located, and in the higher level, a cold anomaly appears (Fig. 8b). This result is consistent with the result in Fig. 6, which showed that the warm anomaly in the mid-latitude region of East Asia is shifted to the north toward the upper level. In the vertical structure of relative humidity, a negative anomaly appears in every level of the latitude range of Korea, and this characteristic is also consistent with the result in Fig. 6 (Fig. 8c).
This local Hadley circulation that developed in high TCGF years can be also seen in the analysis of the differences in the horizontal divergence at lower and upper levels between the two groups ( Fig. 9). At 850 hPa, a negative anomaly appears in the WNP and a positive anomaly appears in the longitude range of 20°-30°N (Fig. 9a). This means that anomalous convergence is strengthened in the WNP and anomalous divergence is strengthened at 20°-30°N. However, Korea shows a weak negative anomaly. At 200 hPa, a positive anomaly is strengthened in the WNP and a negative anomaly is located in the longitude range of 30°-35°N where South Korea is located (Fig. 9b). This implies that anomalous divergence is strengthened in the WNP and anomalous convergence is strengthened in the mid-latitude region of East Asia. Thus, such a structure of horizontal divergence formed in the lower and upper levels implies that the air rising in the WNP falls in the mid-latitude region of East Asia, and this indicates that local Hadley circulation is strengthened in high TCGF years.
In summer, the higher the HWD, the lower the precipitation becomes, which can cause a drought. Thus, the mean PDSI in Korea during July and August and the time series of HWD in Korea were analyzed (Fig. 10a). A smaller PDSI implies a worse drought. The time series of the PDSI has strong interannual variations and the drought has worsened until recently due to the effect of climate change. This linear trend is signi cant at the 95% con dence level. Meanwhile, there is a distinct in-phase trend between the two variables. An analysis of the correlation between the two variables shows a high negative correlation of -0.60. This high negative correlation is signi cant at the 99% con dence level. This means that in Korea, drought worsens as the HWD increases. The two time series show a distinct change in the linear trend, but the correlation did not show a signi cant difference from the original correlation even when the linear trend was removed from the two time series (Corr = -0.58, which is signi cant at the 99% con dence level). In addition, the mean PDSI during July and August in Korea and the time series of TCGF in the WNP during July and August were analyzed (Fig. 10b). There is a distinct out-of-phase trend between the two variables. Therefore, the correlation between the two variables was analyzed and it showed a negative correlation of -0.46. This correlation is signi cant at the 99% con dence level. This indicates that an increase of TCGF in the WNP strengthens drought in Korea, and this result is associated with the local Hadley circulation as analyzed above.

EASM and WNPSM
The anomalous anticyclone strengthened in the mid-latitude region of East Asia and the anomalous cyclone strengthened in the WNP can be associated with EASM and WNPSM, respectively. Hence, the time series of the HWD in Korea and the mean EASMI and WNPSMI during July and August were analyzed (Figs. 11a and 11b). The HWD in Korea and the EASMI showed an out-of-phase trend. When the correlation between the two variables was analyzed, the result showed a negative correlation of -0.57 (Fig. 11a). This correlation is signi cant at the 99% con dence level. This implies that as the EASM is strengthened (weakened), the HWD in Korea becomes lower (higher). The linear trend of the EASMI shows little change. Consequently, the correlation did not show a signi cant difference from the original correlation even when the linear trend was removed from the two time series (Corr = -0.56, which is signi cant at the 99 con dence level). The time series of the HWD in Korea and the mean WNPSMI during July and August were also analyzed (Fig. 11b). The two variables showed a distinct in-phase trend. When the correlation was analyzed, it showed a positive correlation of 0.56. This correlation is signi cant at the 99% con dence level. The WNPSMI has shown an increasing linear trend until recently, and this increasing linear trend is signi cant at the 90% con dence level. Therefore, the correlation was reanalyzed after removing the linear trend from the two time series, and it did not show a signi cant difference from the original correlation (Corr = 0.58, which is signi cant at the 99% con dence level). The above result suggests that the anomalous anticyclone strengthened in the mid-latitude region of East Asia during high TCGF years implies weakening of the EASM, and the anomalous cyclone strengthened in the WNP implies the strengthening of the WNPSM. Regarding this, Choi et al. (2016) have found that when the WNPSM is strengthened, the TCGF increases, but the EASM tends to be weakened.
To examine whether the EASM was weakened in the mid-latitude region of East Asia during high TCGF years and whether the WNPSM was strengthened in the WNP, the differences in the mean OLR, total cloud cover, and rainfall during July and August were analyzed (Figs. 11c, 11d, and 11e). In the analysis result of OLR, a negative anomaly appears in the WNP, whereas a positive anomaly appears in the mid-latitude region of East Asia (Fig. 11a). In particular, the center of positive anomaly in the mid-latitude region of East Asia is located in the zonal direction from the central eastern China to Korea and Japan. The positive anomaly in this region is signi cant at the 95% con dence level. From the above result, it can be seen that convective activity is strengthened in the WNP whereas the convective activity is weakened in the mid-latitude region of East Asia. As a result, a large amount of clouds is observed in the WNP regarding the total cloud cover and a negative anomaly appears in the mid-latitude region of East Asia (Fig. 11d). The negative anomaly in Korea is signi cant at the 95% con dence level. The large amount of clouds formed by the strengthening of convective activity in the WNP leads to a large amount of rainfalls, and the opposite characteristic appears in the mid-latitude region of East Asia (Fig. 11e).

ENSO
Meanwhile, to examine whether the El Niño-Southern Oscillation (ENSO) in uences the HWD in Korea, the difference in the mean SST during July and August between the two groups was analyzed (Fig. 12a). In general, a strong cold anomaly appeared in the equatorial eastern Paci c, and this implies that the eastern Paci c (EP) La Niña strengthened during the high TCGF years. When the analysis result of the difference in velocity potential at lower and upper levels is examined, at the lower level, the center of anomalous convergence is located in the east-west direction from the eastern sea of Australia to the offshore of Peru. At the upper level, the center of anomalous divergence is located in subtropical and tropical central paci c (Figs. 12b and 12c). Therefore, the correlation between the Niño-3 index and the HWD in Korea was analyzed (not shown). These two variables showed a negative correlation of -0.25 at the 90% con dence level. However, the correlation analysis between the Niño-3 index and the TCGF in the WNP during July and August showed a negative correlation of -0.17, which is statistically insigni cant.
Thus, a partial correlation analysis was conducted to examine whether the ENSO in uenced the high positive correlation between HWD in Korea and TCGF in the WNP during July and August (Table 2). When the Niño-3 index was set as a control variable, the HWD in Korea and the TCGF in the WNP during July and August showed a positive correlation of 0.50. This correlation was signi cant at the 99% con dence level. This result suggests that the effect of ENSO on the high positive correlation between the two variables is small.

Scandinavia teleconnection pattern
To determine the cause of the formation of anomalous anticyclone in the mid-latitude region of East Asia and the formation of anomalous cyclone in the WNP during the high TCGF years, the 500 hPa wave activity ux was analyzed (Fig. 13a). The wave activity ux originates from the North Atlantic passes through the Scandinavian Peninsula, the North coast of Russia and East Siberia before reaching Korea and the WNP. The analysis of the difference in the 500 hPa geopotential height between the two groups shows a positive anomaly in the western sea of the UK, the Scandinavian Peninsula, northwestern coast of Russia, and the regions from Central Asia to the Bering Sea (shading in Fig. 13a). This spatial distribution is similar to the Scandinavia teleconnection pattern. Therefore, it can be seen that the anomalous anticyclone formed in the mid-latitude region of East Asia and the anomalous cyclone formed in the WNP during the high TCGF years are associated with the Scandinavia teleconnection pattern. To examine whether the HWD in Korea and the TCGF during July and August are associated with the Scandinavia teleconnection pattern, the Scandinavia index and the time series between the two variables were analyzed (Figs. 13b and 13c). Both variables showed an in-phase trend with the Scandinavia index.
The Scandinavia index showed a signi cant positive correlation of 0.45 with the HWD in Korea at the 99% con dence level (Fig. 13b). The correlation analysis with the TCGF in the WNP during July and August showed a positive correlation of 0.49 at the 99% con dence level (Fig. 13c). This shows that when the Scandinavia teleconnection pattern is strengthened, the HWD in Korea and the TCGF in the WNP during July and August are increased.

TC activity
TCs can also in uence the HWD in Korea in summer. Therefore, the differences in the TCGF and TC passage frequency (TCPF) during July and August between the two groups were analyzed (Figs. 14a and   14b). The spatial distribution of the TCGF shows a higher genesis frequency during high TCGF years in general (Fig. 14a). In particular, signi cantly large differences at the 95% con dence level appear in the South China Sea (SCS) and the northeastern sea of the WNP. The SCS and the northeast seas of WNP show signi cantly large differences at the 95% con dence level. Meanwhile, the spatial distribution of the TCPF shows the spatial distribution of a dipole pattern. In other words, a negative anomaly appears in the Indochina Peninsula and SCS and the southeast region of China, and a positive anomaly in the remaining areas. The cause of such a difference in spatial distribution of the TCGF difference between the two groups can be seen from the analysis result of the difference in the 500 hPa stream ow between the two groups (right panel of Fig. 6b). In the Indochina Peninsula, SCS and the southeastern region of China, anomalous northerlies are strengthened by anomalous cyclonic circulation formed in the WNP. These anomalous northerlies can play the role of steering ows that block TCs that move toward this area. By contrast, the mid-latitude region of East Asia including Korea is in uenced by anomalous southesaterlies due to anomalous anticyclonic circulation, and these anomalous southesaterlies, which play the role of steering ows that move TCs to this region. This trend is associated with the location of WNPSH for the two groups (left panel of Fig. 7). TCs generally tend to move along the western edge of WNPSH. Since the WNPSH is extended in the northwest to the western sea of Korea. Thus, TCs can easily move toward Korea and the mid-latitude region of East Asia. By contrast, during the low TCGF years, the WNPSH is extended southwest toward the southern eastern China, and TCs move to the Indochina Peninsula, SCS and the southeast region of China. Therefore, the time series of the HWD in Korea and the TCPF in the SC area were analyzed (Figs. 14c and 14d). The TCPF in Korea shows an in-phase trend with the HWD in Korea (Fig. 14c), whereas it shows an out-of-phase trend with the TCPF in the SC area (Fig. 14d). In particular, the TCPF in Korea shows a decreasing linear trend, which is signi cant at the 90% con dence level. According to the correlation analysis, the HWD in Korea and the TCPF in Korea shows a signi cant positive correlation of 0.42 at the 99% con dence level, whereas the HWD in Korea and the TCPF in the SC area shows a signi cant negative correlation of -0.51 at the 99% con dence level. Therefore, it can be seen that the TCPF in Korea increases, but it decreases in the SC area during the high TCGF years.
Consequently, even if the TCPF in Korea increased during the high TCGF years, it could not have contributed to mitigating the heat wave.

Summary And Conclusions
This study analyzed the correlations between the TCGF in the WNP and HWD in Korea during July and August for 46 years . The correlation analysis between these two variables showed a strong positive correlation. This implied that the higher the TCGF in the WNP during July and August, the higher the HWD in Korea became. Furthermore, the tropical night in Korea and the TCGF in the WNP during July and August also showed a high correlation.
Meanwhile, to examine in more detail the correlation between HWD in Korea and TCGF in the WNP during July and August, the WNP was divided into the following four areas: Northwest (NW) area, Southwest (SW) area, Northeast (NE) area, Southeast (SE) area. Among these four areas, the area that showed the highest correlation between TCGF and HWD in Korea during July and August is SE area, followed by the NW area. However, the two variables showed a low correlation in the other two regions.
To examine the cause of the high positive correlation between the TCGF in the WNP during July and August and the high HWD in Korea from the above analysis results, 15 years with the highest frequency in the TCGF time series in the WNP during July and August out of the 46 years and 15 years with the lowest frequency out of the 46 years were selected and de ned as high TCGF years and low TCGF years, respectively. An analysis of the difference in Air2m between the two groups in Asia showed that in the mid-latitude region, the Air2m was higher during the high TCGF years, whereas the spatial distribution was higher during the low TCGF years in other regions. Thus, it could be seen from this analysis that the increase of HWD during the high TCGF years is likely to occur in the entire mid-latitude region of East Asia as well as in South Korea.
The differences in the air temperature at the lower, middle, and upper levels of the troposphere between the two groups were analyzed. At the lower level, the warm anomaly was located in the mid-latitude region of East Asia at the center of which Korea located. Furthermore, the WNP, which was the area of TC genesis showed a weak warm anomaly. At the middle level, the warm anomaly in the mid-latitude region of East Asia was shifted more to the north, and the area of the warm anomaly in the WNP was also shifted to the north. At the upper level, the warm anomaly in the mid-latitude region of East Asia moved much further to the north and the entire Korea shows a cold anomaly. In the WNP, the warm anomaly is more strengthened at the lower and middle levels. In addition, the differences in relative humidity between the two groups were analyzed. At all levels, Korea showed a negative anomaly. The reason of this was that subsidence was strengthened by the formation of anomalous anticyclone in Korea and evaporation had increased. By contrast, the WNP, which is a TC genesis area, showed a positive anomaly at every level.
The differences in atmospheric circulations between the two groups were analyzed. In all levels of the troposphere, anomalous anticyclonic and cyclonic circulations were strengthened in the mid-latitude region of East Asia and in the WNP, respectively. This was similar to the Paci c-Japan (PJ) teleconnection pattern (Fig. 15a). Furthermore, the anomalous anticyclone and cyclone strengthened in the mid-latitude region of East Asia and in the WNP was associated with the weakening of the EASM and the strengthening of the WNPSM, respectively.
The degrees of WNPSH and TH develoment were analyzed for the two groups. The WNPSH was strengthened northwest to the western sea of Korea during high TCGF years, whereas it was extended southwest to the southeastern region of China during the low TCGF years. TH moved further north during the high TCGF years than during the low TCGF years. Instead, it was further extended east during the low TCGF years. Therefore, during the high TCGF years, two high pressure systems were overlapped in Korea.
Thus, high pressure systems were located at the lower and upper levels in Korea, and the strengthening of subsidence increased solar radiation, resulting in a higher HWD. Furthermore, during the high TCGF years, two high pressure systems were shifted north, which strengthened the low pressure system in the WNP, thus increasing the TCGF.
An analysis of the difference in the vertical meridional circulation averaged over the longitude range over Korea showed that anomalous upward and downward ows were strengthened in 10 o -30 o N where the WNP is located and in 30 o -40 o N where Korea is located, repectively. This anomalous vertical structure implied that the air rising in the WNP subsided in Korea and the local Hadley circulation was strengthened during the high TCGF years.
The differences in the mean SST during July and August between the two groups were analyzed. In general, a cold anomaly appeared in the equatorial eastern Paci c, which implied the strengthening of the eastern Paci c (EP) La Niña during the high TCGF years. Consequently, a partial correlation analysis was conducted to examine whether the ENSO in uenced the high positive correlation between the HWD in Korea and the TCGF in the WNP during July and August. When the Niño-3 index was set as a control variable, the HWD in Korea and the TCGF in the WNP during July and August showed a high positive correlation. This result suggested that the in uence of ENSO on the high positive correlation between the two variables is small.
To determine the cause of the formation of anomalous anticyclone in the mid-latitude region of East Asia and the formation of anomalous cyclone in the WNP during the high TCGF years, the 500 hPa wave activity ux was analyzed (Fig. 15b). The wave activity ux originated from the North Atlantic, passed through the Scandinavian Peninsula, the North coast of Russia, and East Siberia before reaching Korea and the WNP. The differences in the 500 hPa geopotential height between the two groups showed a positive anomaly in the western sea of the UK, the Scandinavian Peninsula, the northwestern sea of Russia, and the areas from Central Asia to the Bering Sea. This spatial distribution was similar to the Scandinavia teleconnection pattern. Therefore, it could be seen that the anomalous anticyclone formed in the mid-latitude region of East Asia and the anomalous cyclone formed in the WNP during the high TCGF years are associated with the Scandinavia teleconnection pattern.
The differences in the TCPF during July and August between the two groups were analyzed. The Indochina Peninsula, SCS, and the southeastern region of China showed a negative anomaly, whereas the other regions showed a positive anomaly. This result suggests that during the high TCGF years, the TCPF it could not contribute to mitigating the heat wave even if the TCPF increased in Korea.     Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors. on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.

Figure 8
Composite differences of latitude-pressure cross-section of (a) vertical velocity (contours) and meridional circulations (vectors), (b) air temperature, and (c) relative humidity averaged along 120°-130°E between high and low TCGF years in JA. The values of vertical velocity are multiplied by −100. Dashed areas are signi cant at the 95% con dence level and shaded areas denote negative values. Contour intervals are 2−2hPa s−1, 0.2°C, and 1% for vertical velocity, air temperature, and relative humidity, respectively.

Figure 9
Composite difference in horizontal divergence (unit: s-1*107) between high and low TCGF years at (a) 850 hPa and (b) 300 hPa in JA. Hatched areas are signi cant at the 95% con dence level. Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.  the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors. Figure 12 (a) composite difference in SST between high and low TCGF years. In (a), hatched areas are signi cant at the 95% con dence level. (b) composite differences in (b) 850 hPa velocity potential and divergent wind and (c) 200 hPa velocity potential and divergent wind between high and low TCGF years. In (b) and (c), shaded areas denote negative anomalies and contour interval is 2 m2s-110-6. Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors. Figure 13 please see the manuscript le for the full caption Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.

Figure 15
Schematic illustration of (a) 850 hPa anomalous atmospheric circulations and (b) 500 hPa wave activity ux occurring during the high TCGF years. Note: The designations employed and the presentation of the