Influences of Lateral Boundary Forcings on the 2020 Extreme Meiyu in the Yangtze‐Huaihe River Valley

Meiyu occurs in Yangtze‐Huaihe River valley (YHRV) every summer. Its intensity, distribution, and intraseasonal variation are greatly influenced by atmospheric forcings from different directions, such as the monsoon southwesterlies from the south, the western Pacific subtropical high (WPSH) in the east, the atmospheric longwave activities in the north, the southwest vortex from the west, and so on. In this study, to explore the contributions of the atmospheric forcings from different directions to 2020 extreme Meiyu, Regional Climate Model version 4.6 (RegCM4.6) is employed. A series of sensitivity experiments are conducted with realistic or climatological averaged lateral boundary conditions. The results show that the monsoon westerlies from the south transport moisture and heat to YHRV, converge with the cold air brought by the frequent atmospheric longwave activities in the north, and result in this extreme Meiyu. The frequent cold air from the north can lift warm air, provide unstable conditions, and make the distribution of precipitation similar to the teleconnection pattern in East Asia. The stable WPSH in the east anchors the 2020 Meiyu rainband to the YHTV for a long time. In addition, the contributions of the atmospheric forcings from different directions to evolution of 2020 Meiyu change with time. Before 10 June, the WPSH in the east mainly contributes to the Meiyu evolution. From 11 June to 3 July, the combined effects of atmospheric forcings from the south and north are dominant. From 3 July to 10 July, the cold air from the north plays a major role.

1961, causing economic losses of $11.75 billion and the deaths of 219 people (Wang et al., 2021).Many recent studies have emphasized the remote effect of the warm Indian Ocean or the North Atlantic Oscillation (NAO) on this extreme Meiyu.Ding et al. (2021) and Takaya et al. (2020) emphasized the role of Indian Ocean warming in enhancing precipitation by stretching the western Pacific subtropical high (WPSH) to the southwest, and it also favors exceptionally persistent and quasi-stationary Madden-Julian oscillation (MJO) activities, which is an important reason for this extreme rainfall (W.Zhang et al., 2021).In addition, B. Liu et al. (2020) considered that the Meiyu front regulated by the NAO was responsible for this unexpected extreme Meiyu event.However, the contributions of anomalous atmospheric signals in different directions during different periods have not been well analyzed.
The Meiyu front is a planetary-scale, quasi-stationary front formed by the interaction between warm and wet air from low latitudes and cold and dry air from middle and high latitudes (Ding, 2007;Ding et al., 2021;Ninomiya, 1984Ninomiya, , 2000)), which is influenced by factors from different directions (Ding et al., 2020;Y. Liu et al., 2019;Mao & Wu, 2006;Ninomiya & Shibagaki, 2007).A low-level monsoon southwesterly on the south side of the YHRV provides sufficient moisture for summer rainfall in East China (He et al., 2007; R. Zhang & Sumi, 2002).In addition, on the south side of the YHRV, quasi-biweekly oscillation (QBWO) and MJO are also closely related to Meiyu (J.Chen et al., 2015;Ding et al., 2020;Kemball-Cook & Wang, 2001;Song et al., 2016;L. Zhang et al., 2009;Zhu et al., 2022).The QBWO influences the onset, withdrawal, and active/break spells of Meiyu over the YHRV (Annamalai & Slingo, 2001;Seo et al., 2007), which shows the close connection between external anomalous atmospheric forcings and the internal dynamic processes of Meiyu.The northward leap of the upper-level westerly jet, which is located on the north side of the YHRV, is a precursor to the onset of Meiyu (C.Li et al., 2004;L. Li & Zhang, 2014), and the large-scale environmental forcings caused by the East Asian subtropical jet are important reasons for the occurrence and maintenance of the Meiyu rainband (Sampe & Xie, 2010).On the east side of the YHRV, the location and intensity of the WPSH is an important circulation background for Meiyu, and the locations of the rainbands depend largely on the position of the WPSH (Y.Liu et al., 2012;Lu et al., 2005;Sun et al., 2021;Yang et al., 2010).On the west side of the YHRV, Tibetan Plateau vortices and southwest vortices are important rainfall producers for Meiyu when they emigrate from where they originate (Huang et al., 2015;L. Li et al., 2020).Therefore, the factors affecting this extreme 2020 Meiyu are complex, and further research is needed to reveal the role of various factors from different directions in the YHRV on this record-breaking rainfall event.
The article is organized as follows.In Section 2, the model, data sets, and methods are presented.The distribution and evolution of rainfall due to different influencing factors are analyzed in Section 3, and the conclusions are presented in Section 4.

Model
The model used in this study is the Regional Climate Model version 4.6 (RegCM4.6), which was developed by the Abdus Salam International Center for Theoretical Physics (ICTP) (Giorgi et al., 2011), and can be downloaded at https://zenodo.org/record/5766326#.ZCw8cbxBwuU.RegCM4.6 has a hydrostatic dynamic core on an Arakawa B horizontal grid.It is widely used because of its various applications, including studies of modern-day climate simulation, land-atmosphere interactions, climate change projections, and further model developments (M.Chen et al., 2002;Gao et al., 2017;Giorgi et al., 1999).
The model domain has 100 grid points in the west-east direction and 100 grid points in the south-north direction, with the center located at 32°N, 110°E.The resolution of the model domain is 50 km, and the Lambert projection is adopted in this study.Figure 1 shows the scope and topography of the model domain.There are 23 vertical levels within the sigma coordinate.The buffer zone is 10 grid points for each lateral boundary.The model is initialized at 0000 UTC on 1 May 2020 and ends at 0000 UTC on 1 August 2020.The first month is used as the spin-up period and excluded from subsequent analysis.The time step is 120 s, and the model outputs are produced every 6 hr.
The physical schemes used in this study include the nonlocal planetary boundary layer scheme of Holtslag (Holtslag et al., 1990), the ocean flux scheme of Zeng (Zeng et al., 1998), the cumulus scheme of Emanuel (1991), and the land surface scheme of Community Land Model version 4.5 (CLM4.5)(Lawrence et al., 2011) with a time step of 600 s.The lateral boundary conditions are provided through the relaxation/diffusion technique, and details can be found in Giorgi et al. (1993) and Marbaix et al. (2003).

Data Sets
The initial and lateral boundary conditions used to drive RegCM4.6 are four times daily National Centers for Environmental Prediction and the National Center for Atmospheric Research (NCEP/NCAR) reanalysis data (Kalnay et al., 1996), with a resolution of 2.5° × 2.5° for the period from 1991 to 2020; Daily averaging is performed first before the reanalysis data are input into the model.The daily precipitation data are derived from the Climate Prediction Center Global Unified Gauge-Based Analysis (Xie et al., 2007), with a resolution of 0.5° × 0.5°.The precipitation data and the NCAR/NCEP reanalysis data are used as observations to evaluate the model's ability.To unify the spatial resolutions of all the data, the observational data and model outputs are interpolated into the grids with horizontal resolutions of 0.5° × 0.5°.

Methods
One control experiment and seven sensitivity experiments are designed (see Table 1 and Figure 2).The control experiment (Exp_Control) uses the realistic reanalysis data from 2020 as the lateral boundary conditions.Exp_30ave uses the 30-year climatological averaged data from 1991 to 2020, which are changing in time, as the lateral boundary conditions.The other sensitivity experiments replace the boundary condition of the Exp_30ave at the south boundary (Exp_South), the north boundary (Exp_North), the south and north boundaries (Exp_ South&North), the east boundary (Exp_East), the south and east boundaries (Exp_South&East), and the north and east boundaries (Exp_North&East) with 2020 realistic data, respectively, while the other boundaries are consistent with the Exp_30ave.The experimental configurations are shown in Table 1.

Overview of the 2020 Extreme Meiyu and Evaluation of RegCM4.6
In order to evaluate whether the RegCM4.6 model is able to capture the Meiyu pattern and large-scale circulation in 2020, the observation and the simulation of the control experiment are compared.From the observed precipitation distribution in June-July 2020 (Figure 3a), the intensity of the 2020 Meiyu is very strong, and the maximum value of rainfall center exceeds 20 mm per day.The main rainband is located along the lower and middle reaches of the Yangtze River near 30°N from west to east.The distribution of 30-year climatological averaged precipitation shows that the precipitation is mainly distributed in the southern and eastern China, with precipitation centers distributed in Guangdong Province, Guangxi Province of South China, and the YHRV, respectively, during June-July (Figure 3c).Comparing the observed precipitation in 2020 (Figure 3a) with the climatological averaged precipitation (Figure 3c), the precipitation in the lower reaches of the YHRV is much stronger than the climatological averaged precipitation, while the precipitation in Guangdong Province and Guangxi Province in 2020 is much weaker than the climatological averaged precipitation.The precipitation anomalies are mainly distributed along the YHRV, with the maximum value center in the lower reaches of YHRV (Figure 3d).Comparing the simulated precipitation of Exp_Control (Figure 3b) with the observations (Figure 3a), it can be found that the Exp_Control can reproduce the characteristics of the Meiyu rainfall distribution well.However, the simulated precipitation intensity is weaker than the observations to some extent, because the intensity of the 2020 extreme Meiyu is too strong, and the simulation of the 2020 extreme Meiyu is very challenging for regional climate models.
The extreme Meiyu in 2020 occurs under specific circulation conditions (Ding et al., 2021).Figure 4 shows the observed and Exp_Control-simulated large-scale atmospheric circulation at lower, middle, and higher levels of the atmosphere in 2020 and their anomalies.From the observed circulation at 850 hPa, the monsoon southwesterly during the summer of 2020 is weaker than the climatological averaged state, which only reaches the YHRV (Figures 4a-4c).The southwesterlies anomalies converge with the northeasterlies in the YHRV, which are favorable for the extreme Meiyu in 2020 (Figure 4c).The WPSH at 500 hPa in 2020 is much stronger and stretches much further westward than the climatological averaged condition with large positive anomalies in East China (Figures 4e-4g).At the same time, the South Asia high at 200 hPa in 2020 is also much stronger and stretches much further eastward than the climatological averaged state (Figures 4i-4k).In addition, the upper-level westerly jet at 200 hPa in the summer of 2020 is stronger than the climatological averaged state.The weak monsoon westerlies, strong WPSH, South Asia high and upper-level jet all favor the occurrence of the 2020 extreme Meiyu.Comparing the Exp_Control with the observations, RegCM4.6 can basically reproduce the large-scale atmospheric circulation, although the simulated WPSH and South Asia high are slightly weaker than the observations.Therefore, the evaluation of the simulation results of the Exp_Control indicates that the Exp_Control is able to basically capture anomalous patterns of precipitation and large-scale circulation in 2020.In the following text, the effects of different lateral boundary forcings on the extreme 2020 Meiyu in the YHRV are analyzed specifically through comparisons between observations with Exp_Control and different sensitivity experiments.

The Effect of Atmospheric Forcings on Precipitation
Driven by the 30-year climatological averaged reanalysis data, the summer rainfall distribution simulated by Exp_30ave (Figure 5a) is similar to the observed 30-year climatological averaged condition to some extent  (Figure 3c).The areas with much precipitation are mainly located in Guangdong Province and Guangxi Province in South China.
To avoid the impacts of the systematic weak bias simulated by RegCM4.6, the simulated rainfall of Exp_Control is assumed to be accurate.The differences between Exp_Control and other sensitivity experiments can be obtained by subtracting the Exp_Control results from the sensitivity experiment results to quantify the contributions of the lateral boundary forcings in different directions.From the precipitation bias between Exp_30ave and Exp_Control (Figure 6a), it can be found that the precipitation bias shows a "+ -+" pattern, with a strong negative Summer rainfall in China is closely related to atmospheric signals over the ocean south of China, such as the summer monsoon, convection over the tropical ocean, and boreal summer intraseasonal oscillation transmitted from south to north (J.Chen et al., 2015;Ding et al., 2020;He et al., 2007;Kemball-Cook & Wang, 2001;Song et al., 2016;Xu et al., 2022;R. Zhang & Sumi, 2002).After adding the 2020 realistic boundary forcings to the south boundary, no precipitation occurs over most parts of East China (Figure 5b).Large negative deviations between Exp_South and Exp_Control can be seen in East China, especially in the YHRV (Figure 6b).The southwesterlies from the ocean can transport moisture and heat to the YHRV; however, without cold air from the north, water vapor cannot be lifted to produce precipitation.
Frequent atmospheric longwave activities in upper-level westerlies at mid-latitudes, such as blocking highs, troughs, and ridges, can transport cold air to the YHRV, lift moist air, and result in thermal instability for the occurrence of precipitation, which is important for the occurrence and maintenance of Meiyu in the YHRV (C.Li  , 2004;Sampe & Xie, 2010).The results of Exp_North show that after adding the realistic atmospheric forcing to the north boundary, the precipitation increases significantly compared to both Exp_South and Exp_30ave, with a large precipitation center located in South China, the lower reach of the Yangtze River, and parts of North China (Figure 5c).From the precipitation bias between Exp_North and Exp_Control (Figure 6c), the distribution of precipitation bias shows a "+ -+ -+" pattern from south to north in the meridional direction, indicating that when the cold air from the north is active, it also makes the summer precipitation distribution in East China similar to a teleconnection pattern.
If the realistic south and north anomalous atmospheric signals in 2020 are added to the lateral boundary condition at the same time (Exp_South&North), a large amount of precipitation occurs in North China, and some precipitation appears in the Yangtze River basin (Figure 5d).However, the precipitation bias shows that the intensity in Exp_South&North is still weak in the YHRV, and the main rainband is located in North China, not in the YHRV (Figure 6d).It can also be noted that both the Exp_South and Exp_South&North have the 2020 realistic south boundary forcing, which is characterized by weaker southwesterlies than the climatological averaged state (Figure 4).However, the realistic southwesterlies in 2020 alone cannot result in extreme Meiyu in the YHRV but rather result in less precipitation in both the YHRV and South China.Although the interaction of frequent cold air from the north and the moist warm monsoon southwesterlies from the south can produce precipitation, the location of the rain belt is to the north of the YHRV.
The WPSH is located in eastern China and is one of the most important systems influencing Meiyu rainfall (Y.Liu et al., 2012;Lu et al., 2005;Sun et al., 2021;Yang et al., 2010).After adding the 2020 realistic east boundary forcing, the Exp_East simulates the location of the rainband well.The northeast-southwest rainband is mainly located in South China and the middle and lower reaches of the Yangtze River (Figure 5e), indicating that the interaction of the south and north atmospheric forcing is responsible for the occurrence of the Meiyu rainfall, while the location of the Meiyu rain belt is controlled by the WPSH in 2020.The intensity that was stronger than the 30-year averaged state and the stable southern location of the WPSH in 2020 resulted in the extreme Meiyu in the YHRV (Figure 4).From the distribution of the bias between Exp_East and Exp_Control (Figure 6e), a positive bias can be found in the middle reaches of the Yangtze River, South China, and North China.
Exp_South&East uses both the realistic south boundary forcing and east boundary forcing, Figure 5f shows that the precipitation in South China is significantly reduced, and the northeast-southwest rainband appears in the middle and lower reaches of the Yangtze River.The distribution of precipitation bias between Exp_South&East and Exp_Control (Figure 6f) indicates that the south boundary forcing can decrease the precipitation in South China and produce a northeast-southwest rainband rather than an east-west rainband as observed and in Exp_ Control.In addition, the location of the main rainfall belt is slightly southward of the observation.
After adding both the 2020 realistic north and east boundary forcing (Exp_North&East), the simulated precipitation in South China is further reduced, and the location of the Meiyu rainband shifts northward, which is close to the observation.In addition, the simulated rainband shows the same east-west zonal distribution characteristics as the observations and Exp_Control (Figures 5g and 6g).These results indicate that the combination of the active cold air from the north and strong WPSH can generate an east-west zonal rainband in the Yangtze River basin.Therefore, the interaction of the cold air from the north and the WPSH is very important for the location and direction of the Meiyu rainbands.
By comparing the precipitation bias between sensitivity experiments and Exp_Control, for the extreme Meiyu in summer 2020, the interaction of lateral boundary forcings from the south, north and east sides are the key factors for the formation of 2020 Meiyu.The warm moist southwesterlies can reduce precipitation in South China and provide moisture and heat for heavy Meiyu rainfall.However, the south boundary forcing alone cannot generate much rainfall without the lift and thermal instability conditions provided by the cold air from the north.The WPSH in the east can decide and anchor the position of the Meiyu rainband in the YHRV. Figure 7 shows the time-latitude profiles of the zonal-averaged precipitation at 107°-122°E from June to July.
From the observed evolution of precipitation (Figure 7a), the Meiyu precipitation evolution exhibits an obvious intraseasonal oscillation, and the rainband position oscillates north and south at approximately 30°N.By comparing Exp_Control (Figure 7b) and the observations, it can be found that the Exp_Control can basically simulate the evolution of rainband location and intensity under the 2020 realistic atmospheric forcings, but the simulated precipitation by Exp_Control is much weaker than the observation after 15 July, indicating that the model cannot represent the physical processes causing strong rainfall after 15 July.
By comparing the Exp_30ave with Exp_Control, there is no significant oscillation in the rainfall in the Yangtze River basin until 10 July (Figure 7c).Between 13 June and 5 July, precipitation is concentrated in South China and the South China Sea.The simulated rainband evolution of Exp_30ave after 10 July is very similar to the 2020 condition with two northward movements and one southward movement, indicating that the atmospheric conditions after 10 July in 2020 are similar to the climatological averaged state.However, the two northward shifts simulated by Exp_30ave reach further north than the observations, indicating that the climatological averaged monsoon southwesterlies from the south are stronger than those in 2020, and the climatological averaged WPSH can jump to a more northerly position than that in 2020 (Figure 4).
The results of Exp_South (Figure 7d) show that if only the realistic south boundary forcing is added, there is nearly no precipitation occurring in the Yangtze River basin, and the precipitation does not have characteristics of intraseasonal oscillation before 15 July.The rainband moves southward during 15-20 July, then moves northward after 20 July, which is consistent with the Exp_30ave (Figure 7c), further demonstrating that the precipitation evolution after 15 July can be generated even without atmospheric anomalies in 2020.
After adding the realistic north boundary forcing, the precipitation simulated by Exp_North increases significantly and shows an obvious intraseasonal oscillation similar to the observations.However, the location of the rainbands reaches more northward than the observation before 20 June.The southward shift of the rainband simulated by Exp_North from 3 July to 10 July is nearly the same as that simulated by Exp_Control, indicating that the anomalous atmospheric forcing in the north plays an important role in the precipitation evolution during this period.
If both the realistic north and south boundary forcings are added at the same time, the simulated precipitation evolution by Exp_South&North becomes closer to the observations than that of Exp_North (Figure 7f).The pattern correlation coefficient between the simulated precipitation evolution of Exp_South&North and Exp_ Control is the highest among all the sensitivity experiments (Figure 8), so it can be inferred that the intraseasonal oscillation of precipitation is mainly caused by the interplay of warm moist air from the south and cold air from the north.Many studies have demonstrated that the combined effects of warm moist air from the south and cold air from the north are important for the intraseasonal oscillation of the summer precipitation (J.Chen et al., 2015;Ding et al., 2020).However, due to the lack of realistic WPSH forcing in 2020, the rainband simulated by Exp_ South&North is located further northward than the observation, with precipitation mainly concentrated north of 30°N (Figure 7f).
By comparing the Exp_East (Figure 7g) with Exp_Control (Figure 7b) and the observations (Figure 7a), the southward movement of the rainband is consistent with the observation before 10 June, indicating that the north-south swing of the WPSH plays a major role in the precipitation evolution before 10 June.However, with only the east boundary forcing, Exp_East simulates no precipitation oscillations in YHRV from 11 June to 3 July.The north-south oscillation of the rainband after 10 July simulated by Exp_East is similar to Exp_30ave, further indicating that the atmospheric conditions during this period are close to the 30-year climatological averaged state and that no atmospheric forcing in any one direction plays a dominant role.
After adding both the south and east boundary forcings, the precipitation oscillations simulated by Exp_ South&East (Figure 7h) are very close to those simulated by Exp_Control (Figure 7b), and the pattern correlation coefficient with the Exp_Control is the second highest (Figure 8).It can also be found that the pattern correlation coefficients of Exp_South&East and Exp_South&North with the Exp_Control are high because both experiments have a realistic south boundary forcing, indicating that the south atmospheric forcing is crucial for the intraseasonal oscillation of the Meiyu rainband.
Figure 7i shows that the rainband simulated by Exp_North&East experiences a southward process and a northward process before 17 June, and the intensity of precipitation is strong in the YHRV, suggesting that both the cold air from the north and the warm moist southeasterlies from the south near the WPSH are mainly responsible for the occurrence of the strong Meiyu during this period.However, Exp_North&East fails to simulate the oscillation of the Meiyu rainband from 18 June to 3 July, and the simulated rainband is further northward, indicating that the south atmospheric forcing is critical to the oscillation of Meiyu rainfall in this period.
To objectively evaluate the model skills of different sensitivity experiments in simulating precipitation evolutions, Figure 8 shows pattern correlation coefficients between different sensitivity experiments and Exp_Control of time-latitude profiles of zonal-averaged precipitation, indicating that the contributions of the anomalous atmospheric forcings in different directions change with time.Among them, the east atmospheric forcing contributes to the precipitation evolution before 10 June.The precipitation oscillation process from 11 June to 3 July is mainly caused by the combined effect of south and north atmospheric forcings.From 3 July to 10 July, the north atmospheric forcings are important in the southward movement of the rainband.The atmospheric conditions in 2020 are similar to the climatological averaged state after 10 July, so the rainband evolution in 2020 is also similar to the climatological averaged state, and no atmospheric forcing in any direction plays a dominant role during this period.

The Effect of Atmospheric Forcings on the Wind at 850 hPa
The summer rainfall in East China is closely related to the EASM, and the evolution and strength of the EASM can lead to precipitation anomalies (Ding et al., 2021;He et al., 2007).Figure 9 shows the time-latitude profiles of the 850 hPa meridional wind, horizontal wind, and divergence of the vertically integrated water vapor transported from the surface to 300 hPa averaged over 107°-122°E during June-July.From the observations (Figure 9a), the large value center of the southerlies is stable south of 30°N.The intensity and evolution show an obvious characteristic of intraseasonal oscillation during June and July in 2020.The convergence of water vapor, which is located north of the large value area of the southerly wind, provides favorable dynamic lifting conditions for heavy rainfall.The monsoon southwesterlies from the south and the cold air from the north continue to converge near 30°N, leading to the super Meiyu in 2020.By comparing Exp_Control (Figure 9b) with the observations, the Exp_Control can basically reproduce the evolution of the southwesterlies, but the large value region of the southerlies extends more northward than the observations after 15 July, resulting in a more northward rainband and less Meiyu rainfall during this period.The monsoon southwesterlies simulated by Exp_30ave (Figure 9c) are stronger and extend further northward than those simulated by Exp_Control, resulting in the convergent zone of water vapor and the rainband moving northward of 30°N by 21 June.From 21 June to 5 July, the intensity of the southerlies simulated by Exp_30ave is weak, making the rainband move southward.After 10 July, the wind evolution simulated by Exp_30ave is similar to Exp_Control, so the simulated rainband evolutions are also consistent with the Exp_Control, further indicating that the atmospheric circulation evolution after 10 July in 2020 is similar to the climatological averaged state.
Adding the realistic boundary forcings in the south or north direction alone cannot simulate the evolution of the monsoon southwesterlies well.For example, the southerlies simulated by Exp_South are stronger than the observations, which extend more northward, resulting in a more north location of the Meiyu rainband than the observations.After 15 July, the monsoon southwesterlies evolution of Exp_South is close to Exp_Control, indicating that the atmospheric anomalies are also close to the climatological averaged state during this period (Figure 9d).The southwesterlies simulated by Exp_North extend further northward and have weaker intensity than Exp_Control in June (Figure 9e), resulting in a north rainband in June (Figure 7e).The evolution of monsoon southwesterlies in July simulated by Exp_North is similar to that simulated by the Exp_Control; therefore, the rainband evolution is also reproduced well by Exp_North.The realistic south and north boundary forcings together make the Exp_South&North reproduce a reasonable evolution of the southwesterlies, resulting in the simulated evolution of the rainband by Exp_South&North being the best in all the sensitivity experiments.Thus, the interaction between cold air from the north and warm moist air from the south is very important for the maintenance and oscillation of the rainband.
Exp_South&East (Figure 9h), Exp_North&East (Figure 9i), and Exp_East (Figure 9g) all have realistic east boundary forcings.From the results of the three experiments, the east boundary forcing can control the strong southerlies that remain south of 30°N, and the convergence zone of the northerlies and southerlies is maintained near 30°N.However, without the north boundary forcing and the south boundary forcing, the simulated evolution of the southwesterlies by Exp_East deviates from that of Exp_Control in June, further indicating that both the north boundary forcing, and south boundary forcing are very important to the intraseasonal oscillation of monsoon southwesterlies.
By analyzing the 850 hPa atmospheric circulation deviations between sensitivity experiments and the Exp_ Control (Figure 10), the circulation deviation has a south-to-north distribution of "cyclone," "anticyclone," and "cyclone," thus causing the precipitation deviation to show a "+ -+" pattern from south to north (Figure 6).

The Effect of Atmospheric Forcings on the WPSH
Figure 11 shows the time-latitude profiles of the zonal-averaged ridgeline of the WPSH and vorticity at 500 hPa in June-July, where the ridgeline of the WPSH represents the position of the WPSH.The negative vorticity region in the south represents the region controlled by the WPSH.As shown in Figure 11a, the north-south movement of the WPSH is basically consistent with the north-south movement of the Meiyu rainband (Figure 7a) and has obvious characteristics of intraseasonal oscillation.By comparing Exp_Control with the observations, the Exp_Control can basically reproduce the northward and southward movements of the WPSH (Figure 11b).The evolution of the positive vorticity is also consistent with the northward and southward movements of the rainband in Exp_Control (Figure 7b).After 15 July, the simulated WPSH and negative vorticity region by Exp_Control are further northward than the observations, resulting in a weak and northerly Meiyu rainband during this period.
The WPSH simulated by Exp_30ave is further northward than that in 2020 (Figure 11c), which is consistent with the climatological averaged state (Figure 4).However, the WPSH simulated by Exp_30ave is closer to Exp_Control and the observations after 10 July, further indicating that the atmospheric state is close to the climatological averaged state after 10 July.Without the east realistic boundary forcing, the Exp_South (Figure 11d), Exp_North (Figure 11e), and Exp_South&North (Figure 11f) cannot simulate the position of the WPSH well before 10 July.They produce a more northerly WPSH, resulting in a northerly rainband.In contrast, in the control of the realistic east boundary forcing, the Exp_East (Figure 11g), Exp_South&East (Figure 11h), and Exp_North&East (Figure 11i) can greatly improve the simulation of the WPSH.Among them, the simulated evolution of the WPSH in Exp_North&East is closest to Exp_Control and the observations, with the ridgeline becoming stabilized near 25°N, suggesting that the interplay of north and east atmospheric signals plays a key role in the evolution of the WPSH and Meiyu rainband.

The Effect of Atmospheric Forcings on the Upper-Level Westerly Jet
Figure 12 shows the time-latitude profiles of the zonal-averaged 200 hPa zonal winds, geopotential height and divergence in June-July.As shown in the observations (Figure 12a), the subtropical westerly jet is maintained between 30°N and 40°N from the beginning of June to 20 July, with a constant strong intensity.The strong subtropical westerly jet is related to the continuous convergence of cold air from the north and warm air from the south, which intensifies the south-north temperature gradient.The south side of the jet is the positive divergent zone, which can provide good lifting conditions for precipitation, and the evolutions of both the positive divergence and the rainband (Figure 7a) are basically consistent.Moreover, the position of the South Asia High is relatively stable and maintained south of 30°N.The weak jet simulated by Exp_Control also makes the total precipitation intensity simulated by Exp_Control weak (Figure 12b).
If only the realistic south boundary forcing is used, the westerly jet simulated by Exp_South is weak before 20 June, so no precipitation occurs in the YHRV during this period.Between 20 June and 10 July, the westerly jet simulated by Exp_South is stronger but located further northward than the observations, resulting in no precipitation in YHRV.According to the previous analysis, the anomalous atmospheric conditions in 2020 are similar to the climatological averaged state after 10 July, so the westerly jet simulated by Exp_South is as weak as that simulated by Exp_Control and the observations (Figure 12d).
If only the realistic north boundary forcing is included (Exp_North), although the subtropical jet is close to the north boundary and the realistic atmospheric forcing can diffuse from the north boundary through the buffer zone to the interior of the region, Exp_North cannot reproduce the intensity and evolution of the subtropical jet, and the simulated subtropical jet is much weaker than that in Exp_Control (Figure 12e).These results suggest that there is a systematic interaction between the north cold air, south warm air, and Meiyu rainfall.The upper-level jet is in the subtropical frontal region with a large north-south temperature gradient.The south boundary forcing in Exp_North is the climatological averaged state, leading to strong monsoon southwesterlies and weak Meiyu.
The weak precipitation releases less latent heat of condensation, causing a weak temperature gradient and weak westerly jet in Exp_North (Figure 12e).
Without the realistic north boundary forcing, the Exp_East (Figure 12g) and the Exp_South&East (Figure 12h) cannot simulate the westerly jet well.Among all sensitivity experiments, only Exp_South&North (Figure 12f) and Exp_North&East (Figure 12i) can reproduce the intensity and evolution of the upper-level jet well, suggesting that the north boundary forcing combined with south or east boundary forcing can reproduce the continuous convergence of the north cold air and south warm air, intensifying the temperature gradient and favoring the strong westerly jet and Meiyu.Without the north atmospheric forcing, the westerly jet is weak.

Conclusions
In this study, we focus on the extreme Meiyu process in the YHRV from June to July 2020.The NCEP/NCAR reanalysis data were used as the large-scale boundary forcings to drive RegCM4.6.The contributions of lateral boundary forcings in different directions were explored through a series of sensitivity experiments with the realistic or climatological lateral boundary conditions in the south, north, and east directions, respectively.The main conclusions are given as follows.
Compared with the 30-year climatological averaged state, the overall atmospheric circulation in the summer of 2020 shows the following characteristics.First, the monsoon southwesterlies are weak.Second, cold air from the north is active.Third, the WPSH and South Asia high are strong.Driven by the 2020 realistic reanalysis data, Exp_Control can well reproduce the distribution of Meiyu rainfall during June-July, but the intensity of the simulated precipitation is slightly weaker than that of the observations.The simulated precipitation driven by 30-year averaged reanalysis data is very similar to the observed 30-year climatological averaged precipitation.The extreme Meiyu in 2020 exhibits obvious characteristics of intraseasonal oscillation in the YHRV, and the south atmospheric forcing is crucial for the intraseasonal oscillation of Meiyu.In addition, the contributions of anomalous atmospheric forcings in different directions change with time.Before 10 June, the east atmospheric forcing plays a major role in the precipitation evolution.From 11 June to 3 July, the precipitation oscillation process is mainly caused by the combined effect of south and north atmospheric forcings.From 3 July to 10 July, the north atmospheric forcing is important for the southward precipitation process.After 10 July, the atmospheric circulation conditions in 2020 are closer to the climatological averaged state, making its rainband evolution similar to the climatological averaged state.
The results also reveal that the upper-level divergence provided by the upper-level westerly jet is not the decisive condition for the strong Meiyu in 2020.There is an interaction between the north cold air, south warm air, Meiyu, and upper-level subtropical jet.Only the north atmospheric forcing cannot reproduce the strong westerly jet and Meiyu.The north atmospheric forcing must be combined with the south or east atmospheric forcing so that it can reproduce the continuous convergence of the north cold air and south warm air, which intensifies the temperature gradient and favors the strong westerly jet and Meiyu.

Figure 1 .
Figure 1.Model domain and distribution of topographic height (shading, m).
conditions for the north and east boundaries based on Exp_30ave are used

Figure 3 .
Figure 3. Distributions of the daily averaged precipitation from June to July (shading, mm day −1 ) in the observations in 2020 (a), Exp_Control (b), the 30-year climatological averaged data from 1991 to 2020 (c), the anomalies of 2020 from climatology (d), and the anomalies of Exp_Control from climatology (e).

Figure 4 .
Figure 4.The daily averaged wind vector (vector, m s −1 ) and geopotential height (shading, gpm) at 850, 500, and 200 hPa for the observations, Exp_Control and their anomalies, during June-July 2020 compared with climatological averaged state from 1991 to 2020.The solid green line denotes the geopotential height contours in 2020, and the dashed green line denotes the climatological averaged geopotential height contour, which can represent the shape and scope of the WPSH and South Asia high (850 hPa: 1,500 gpm; 500 hPa: 5,880 gpm; 200 hPa: 12,520 gpm).

Figure 5 .
Figure 5. Distributions of the daily averaged precipitation from June to July (shading, mm day −1 ) in different sensitivity experiments (Exp_30ave: (a); Exp_South: (b); Exp_North: (c); Exp_South&North: (d); Exp_East: (e); Exp_South&East: (f); Exp_North&East: (g)).The boundaries of each figure are marked with thick or normal lines to show the configuration of different sensitivity experiments.The thick solid line for the boundary represents the realistic boundary forcing in 2020, while the normal line represents the 30-year averaged boundary forcing.

Figure 6 .
Figure 6.Same as Figure 5, but the distributions of the precipitation deviations of each sensitivity experiment from the control experiment (shading, mm day −1 ).

Figure 8 .
Figure 8. Pattern correlation coefficients between different sensitivity experiments and Exp_Control of time-latitude profiles of zonal-averaged (107°-122°E) precipitation, where the data above the solid blue line are those passing the 95% significance test.

Table 1
Model Configurations for the Seven Experiments