Synoptic-Scale Analysis on Development and Maintenance of the 19–21 July 2021 Extreme Heavy Rainfall in Henan, Central China

Heavy rainfall is one of the major weather disasters in China, which imposes great impacts on people’s life and property. In the context of global warming, frequent occurrences of extreme heavy rainstorms have evoked widespread concern on forecast and warning of heavy rainfall.

The rainbelt movement in China is largely controlled by the western Pacific subtropical high (WPSH). Following the north–south migration of the WPSH, the rainbelt also advances to the north or retreats to the south. From late July to early August when the WPSH jumps to the north, the rainbelt moves from the Yangtze–Huai River valley to North China, leading to heavy rainfall there (Tao, 1980). Heavy rainfalls in North China are featured with high intensity but short duration; they are usually highly localized with large interannual variability, and concentrated over a specific period; and they are often intimately related to orographic forcing. Ding et al. (1980) analyzed the synoptic weather systems associated with 33 heavy rainfall events that occurred during 1958–1976 in North China, and found that heavy rainfalls in North China mainly occur when a high-pressure system resides to the east while a low-pressure system to the west, or two high-pressure systems respectively occur to the east and west simultaneously. Low vortices, shear lines, and low-pressure troughs and cold fronts are the main weather systems that cause heavy rainfalls. When two or more weather systems interact or overlap, extreme precipitation may occur.

Historically, the extreme heavy rainfall in August 1963 (“63·8”) in North China (You, 1965), the extreme heavy rainfall in Beijing on 21 July 2012 (“7·21”) (Yu, 2012; Sun et al., 2013), and the torrential rain in North China on 20 July 2016 (“7·20”) (Lei et al., 2017) are all related to the development and movement of low vortices, shear lines, low-pressure troughs and cold fronts, and so on. In addition to the above-mentioned weather systems, the influence of typhoons on heavy rainfalls in North China is also important. While landfall typhoons can directly trigger heavy rainfalls (He et al., 2016, 2020; Yang and Duan, 2020), typhoons may also interact remotely with the westerly wind belt disturbances to generate relay of water vapor transport over long distances to North China leading to heavy rains there (Sun and Zhao, 2000; Hou et al., 2006).

Certain physical quantities such as frontogenesis function, moisture flux convergence, apparent heat source, apparent moisture sink, and others have been widely applied to diagnose the development and maintenance of extreme rainfall events. Petterssen (1936, 1956) investigated changes in front strength from the perspective of potential temperature gradient, and proposed that divergence, vorticity, and deformation all can contribute to frontogenesis. Further studies indicate that deformation and horizontal convergence terms both play a critical role in frontogenesis (Sun and Du, 1996; Wang and Wu, 2000) because they can lead to convergence of water vapor and energy, which subsequently affects the initiation and development of a rainstorm (Wu et al., 1995; Wu and Cai, 1997; Tao et al., 2008; Duan et al., 2018). Ninomiya (1984, 2000) calculated in detail the effect of each individual term of the frontogenesis function in the study of a Meiyu front and proposed that horizontal convergence term and deformation term are the main contributors to Meiyu front genesis. Han et al. (2005) found that the deformation term of the frontogenesis function plays an important role in extratropical transition of landfall typhoon. Compared with the divergence and tilting terms, the deformation term shows an earlier and more significant effect in landfall typhoon extratropical transition process. Du and Lan (2010) compared the structures of two Yunnan–Guizhou quasi-stationary fronts and found that the differences in the frontal structure are largely attributed to differences in the contributions from the horizontal convergence and deformation terms. The “7·21” extreme heavy rain in Beijing is a typical precipitation process induced by an upper-level trough accompanied with a cold front, during which strong frontogenesis occurred due to the local deformation effect at lower levels (Yu, 2012; Li et al., 2013; Sun et al., 2013).

Precipitation is closely related to vertical motion. Under the condition of sufficient water vapor supply, vertical motion actually determines the occurrence and total amount of precipitation. The moist air expands and cools during ascending. When it reaches the saturation level, condensation of water vapor leads to the release of latent heat, which heats up the air and results in stronger vertical movement. Apparently, there is a positive feedback between the latent heat release and the upward movement in the precipitation process. The apparent heat source (Q₁) and moisture sink (Q₂) are often applied to analysis of the heating characteristics during heavy rainstorms (Luo and Yanai, 1983; Yanai et al., 1992; Wang and Lu, 1994; Qin and Lu, 2013). Ding and Wang (1988) calculated Q₁ and Q₂ to analyze the latent heat release of Meiyu rainfall in the middle and lower reaches of the Yangtze River. Guo (2000) and Jiao et al. (2006) found that occurrences of heavy rainfall coincide with the evolution of Q₁ and Q₂, that is, heavy rainfall happens when large values of Q₁ and Q₂ appear, and Q₁ and Q₂ are dominated by vertical transport of heat and moisture. It is clear that quantitative analysis of dynamic and thermodynamic characteristics of the rainstorm system using multiple physical quantities helps to better understand and predict heavy rainfalls.

In July 2021 (“21·7”), a large-scale extreme heavy rainfall event occurred in Henan Province, central China. Torrential rain took place at multiple places of Henan, and concentrated precipitation occurred during 17–22 July 2021. Accumulated precipitation reached 1122.6 mm in Hebi Prefecture of northern Henan. The strongest precipitation occurred on 20 July at Zhengzhou, the capital city of Henan; specifically, Zhengzhou meteorological station witnessed the heaviest rainfall during 1500–1700 Beijing Time (BT), with an hourly rain rate of 201.9 mm in 1600–1700 BT, breaking the historical record of 1-h rainfall in the Chinese mainland and resulting in great casualties and property damages (Su et al., 2021; Bueh et al., 2022; Liang et al., 2022; Xie et al., 2022; Yin et al., 2022).

Chyi et al. (2022) investigated the mesoscale systems responsible for the initiation and development of this extreme rainfall, based on minute-interval surface observations, radar data, and high-resolution satellite data. At the synoptic scale, the “21·7” extreme rainfall event also presents some unique features that deserve in-depth analysis. For example, the heavy rainfall occurred under a weak saddle pattern in the pressure field, with two remote tropical cyclone in East and South China respectively, and a weak Huang–Huai cyclone lingering nearby. More questions remain as follows—is there any frontal activity involved in sustaining this event? Given the unprecedented rainfall amount, how is the role of diabatic heating? Where is the source for the excessive moisture supply to this extreme rainfall? As a follow-up study to Chyi et al. (2022), the current paper tries to reveal the development and maintenance mechanisms at the synoptic scale for the “21·7” heavy rainfall event in Henan Province, central China. The ECMWF Reanalysis version 5(ERA5) data are used to derive frontogenesis, atmospheric diabatic heating, and water vapor budget during the extremely heavy rainfall event. The results obtained are expected to help better understand and forecast future extreme rainfall events in this region.

The hourly ERA5 reanalysis data (Hersbach et al., 2020; https://cds.climate.copernicus.eu/) on a 0.25° × 0.25° horizontal resolution with 21 vertical levels from 1000 to 100 hPa for 19–22 July 2021 are used. Hourly geopotential height, temperature, zonal and meridional winds, vertical velocity, and specific humidity are analyzed. Comparison with sounding data suggests that the synoptic weather systems in ERA5 are consistent with those in observations (figure omitted), while the high resolution ERA5 data provide extra information for refined analysis.

Hourly surface station observations from 19 to 22 July 2021 and the sounding data from 20 to 21 July 2021 are provided by the National Meteorological Information Centre of the China Meteorological Administration (CMA) (https://data.cma.cn/data/cdcdetail/dataCode/A.0012.0001.html). Note that the daily average of meteorological variables refers to the 24-h average from 0800 to 0800 BT the next day. The average over the heavy rainfall event refers to the 72-h average from 0800 BT 19 to 0800 BT 22 July 2021.

Large-scale precipitation and disastrous weathers are often caused by fronts and associated horizontal wind convergence and uplift of air mass. The frontogenesis function is widely used in diagnosis of fronts and other discontinuous meteorological factors. To investigate the frontogenesis during the “21·7” extreme rainfall event in Henan, both empirical knowledge and principles of synoptic meteorology are employed to objectively define the front that brought the heavy precipitation. The frontogenesis function (F) is calculated by using the approach proposed by Zhu et al. (2000), in which the thermodynamic parameter is the pseudo-equivalent potential temperature (θ_se) and the influence of the diabatic heating term is ignored. The complete (without omission) formula is expressed as:

where F₁, F₂, F₃, and F₄ are diabatic heating term, horizontal divergence term, horizontal deformation term, and tilting term related to vertical motion, respectively. Here, $ {A}_{\rm f}=\dfrac{\partial u}{\partial x}-\dfrac{\partial v}{\partial y} $ is extension-induced deformation term, ${B}_{\rm f}= $$ \dfrac{\partial v}{\partial x}+\dfrac{\partial u}{\partial y} $ is shear-induced deformation term, and $ {D}_{\rm h}= $$\dfrac{\partial u}{\partial x}+\dfrac{\partial v}{\partial y} $ is divergence term. When F and its individual components are larger than zero, the horizontal gradient of $ \theta $_se increases, indicating frontogenesis; otherwise the horizontal gradient of $ \theta $_se decreases, indicating frontolysis.

Apparent heat source (Q₁) and moisture sink (Q₂) are calculated by using the methods proposed by Yanai et al. (1992):

where p is pressure, θ is potential temperature, V is horizontal wind, κ = R/c_p, and R and c_p are the gas constant and specific heat capacity at constant pressure, respectively. Note, p₀ = 1000 hPa, $ \omega $ is vertical velocity, L is latent heat coefficient, and q is specific humidity. The apparent heat source (Q₁) and moisture sink (Q₂) over the entire atmosphere can be obtained by vertical integration of Q₁ and Q₂ from the surface to 100 hPa. Q₁ includes radiative cooling, net water vapor condensation, and vertical transport by small-scale eddies from cumulus and turbulence. Q₂ includes net water vapor condensation and vertical transport by small-scale eddies from cumulus and turbulence. Q₁ (Q₂) can be obtained from local variation term of potential temperature, horizontal advection, and vertical advection.

A rainfall process includes large-scale (synoptic-scale) stable precipitation and mesoscale convective precipitation processes, and thus the latent heat release also includes both large-scale and mesoscale release of latent heat. To distinct the latent heat release caused by the two different processes, large-scale vertical motion and surface precipitation observations are utilized to estimate distributions of large-scale and mesoscale latent heat release. The latent heating rate (H_s) of condensation caused by large-scale vertical motion and the latent heating rate (H_c) caused by mesoscale convective precipitation can be written as (Lei et al., 2017):

where L is the latent heat coefficient, $ \omega $ is vertical velocity, q_s is saturation specific humidity, p is pressure, c_p is the specific heat capacity at constant pressure, R is precipitation, and p_B and p_T are pressure at the bottom and top of the cloud, respectively. The equilibrium level pressure and the uplift condensation level pressure in the sounding data are used to replace the top of cloud and bottom of cloud, respectively. Note that the calculation of large-scale latent heating needs to follow three conditions: (1) $ \omega $ < 0, (2) q/q_s > 0.8, and (3) $ -\dfrac{\partial \theta }{\partial p} $ > 0, while the calculation of the convective latent heating rate needs to follow $-\dfrac{\partial \theta }{\partial p} $ < 0. For the convenience of comparison, H_s and H_c in the saturated atmospheric layer are averaged over the depth of the cloud body; for the calculation of H_c, R is the average of the intensive ground observations of rainfall for the past 6 hours, and R, p_B, and p_T are interpolated onto a regular grid with intervals of 2.5° in both latitude and longitude.

To study the water vapor budget during the “21·7” extremely heavy rainfall in Henan, the Gaussian theorem is applied to transform the formula of precipitation rate:

where l represents the horizontal boundary of the area for water vapor budget calculation, s is the area of the critical region, and V_n is the normal wind component, which is orthogonal to the boundary. In Eq. (10), the average water vapor flux (Q) on each side of the boundary can be calculated by integrating the water vapor flux along the boundary. Then the difference between the water vapor flux in the east and west directions is A_East–West = Q_East − Q_West, and the difference in the north and south direction is A_{North–South} = Q_North – Q_South. The total water vapor budget is A_total = A_East–West + A_{North–South}.

From 17 to 22 July 2021, an extremely heavy rainfall process occurred in Henan. This process is characterized by extremely heavy precipitation, large amount of accumulative precipitation, extremely strong short-term precipitation, long duration, and severe impacts. During this process, record-breaking precipitation occurred at 1/6 of all the surface weather stations in Henan Province. Daily precipitation at 20 national meteorological stations including Zhengzhou, Xinxiang, Jiaozuo, Anyang, Hebi, etc. broke respective historical records since the stations were built (Fig. 1). Maximum daily precipitation exceeded 500 mm at 32 weather stations and exceeded 600 mm at 5 weather stations. The maximum daily precipitation reached 777.5 mm at Hegang and 624.1 mm at Zhengzhou meteorological station, while the latter one was 3.39 times of the historical maximum daily precipitation (Chyi et al., 2022). The heaviest precipitation occurred on 20 July, centered around Zhengzhou. The maximum hourly precipitation occurred at Zhengzhou meteorological station from 1600 to 1700 BT 20 July (Fig. 1b). Following the eastward moving of the synoptic systems on 21 July, the area of heavy precipitation shifted to northern Henan and southern Hebei provinces (Fig. 1c).

Figure 1.  Distributions of (a) the surface weather stations with record-breaking daily accumulative rainfall (black dots), superimposed with terrain height of the area (shadings; m); (b) daily accumulative rainfall (mm) from 0800 BT 20 to 0800 BT 21 July 2021; and (c) daily accumulative rainfall (mm) from 0800 BT 21 to 0800 BT 22 July 2021. In (b) and (c), the black dot denotes Zhengzhou meteorological station and the black box indicates the area of heavy rainfall (33.5°–36°N, 112.5°–115°E).

The July 2021 extreme heavy rainfall in Henan Province was associated with an abnormal distribution of multiscale weather systems in the middle and low latitudes. The continuous and stable transport of water vapor under this circulation condition superimposed by the orographic effect eventually led to the extreme rainfall. The upper-level circulation pattern was featured with a ridge to the east of Henan Province at 200 hPa (Fig. 2a). Zhengzhou was located in front (east) of the trough and thus dry and cold air from the high latitudes was continuously transported to the study region, providing a favorable condition for upper-level divergence and upward pumping of the low-level airmass. Convection developed continuously under such a situation. In the middle troposphere (500 hPa), the WPSH abnormally shifted to the north (Fig. 2b) and extended westward during its intensification. Henan Province was located in the low-pressure zone between the WPSH and the continental high, and a stationary low vortex maintained over the northwest of Henan.

Figure 2.  Synoptic weather maps at 0800 BT 20 July 2021 of (a) geopotential height (black contours; dagpm) and wind (barbs) at 200 hPa, (b) geopotential height (black contours; dagpm), temperature (red solid lines; °C), and wind (barbs) at 500 hPa, (c) specific humidity (green solid lines; g kg⁻¹) and wind (barbs) at 850 hPa. The black dot denotes Zhengzhou meteorological station where maximum hourly rainfall was observed.

Meanwhile, in the low-latitude Pacific Ocean to the east of the Taiwan Island, Typhoon In-Fa (2106) gradually intensified while moving towards the northwest. Stable easterly wind shears maintained at 850 hPa (Fig. 2c). Following the gradual northwestward movement of Typhoon In-Fa and westward extension of the WPSH, the pressure gradient between the typhoon and the WPSH increased. The southeasterly and the easterly wind shear both intensified at 850 hPa and a large amount of water vapor was continuously transported to Henan Province. The easterlies were forced to rise due to the blocking and orographic forcing of the Song Mountain, the Funiu Mountain, and the Taihang Mountain (Fig. 1a), and the low-level easterly wind shear further strengthened.

Such a synoptic-scale circulation pattern was favorable for water vapor transport to local convective systems and updrafts, which were then converted to precipitation. According to the 500-hPa temperature field (Fig. 2b) and 850-hPa specific humidity field (Fig. 2c), it is clear that a moisture front was significant near Henan Province, although the temperature front was weak. An obvious frontogenesis was observed during the heavy rainfall event, and the front activity was mainly generated by the intersection of the continental warm, dry airmass and the warm, moist airmass from north of Typhoon In-Fa and south of the WPSH.

Apparently, Henan Province was located in the front of the trough at 200 hPa and the weak low pressure zone at 500 hPa between the WPSH and the continental high. Easterly wind shear appeared at 850 hPa. The configuration of low-level convergence and high-level divergence was favorable for the development and maintenance of the synoptic weather systems. The westward extension and intensification of the WPSH and the westward movement of Typhoon In-Fa increased the pressure gradient between them, leading to stronger easterlies and continuous water vapor transport to Henan Province. Under the blocking and uplifting effects of the Song Mountain, the Funiu Mountain, and the Taihang Mountain, the gradually intensified southeasterlies that carried a large amount of water vapor eventually resulted in the extreme heavy rainstorms in Henan Province.

Figure 3 shows distributions of frontogenesis function at 925 hPa and pseudo-equivalent potential temperature during 20–21 July 2021. A positive value of frontogenesis function indicates frontogenesis. The distribution of the frontogenesis function was consistent with the area of dense θ_se isolines, and the moisture gradient in the front area was strong at this time. At 0800 BT 20 July, the θ_se contours were relatively sparse, with the value of 348 K near Zhengzhou. There was a weak frontogenesis zone to the east of Zhengzhou that was oriented along the southwest–northeast direction (Fig. 3a), and the value of frontogenesis function was about 6–8 × 10⁻⁹ K m⁻¹ s⁻¹. In northern Henan, in the vertical direction, there were two large-value centers of frontogenesis indicated by dense θ_se contours at 850 and 600 hPa, respectively (Fig. 4a). Note that during the entire rainfall process, the front zone showed a vertical distribution with an obvious neutral atmospheric stratification (Fig. 4). With the 352- and 356-K contours gradually moving to the north near Zhengzhou, warm dry airmass from the inland area gradually moved northward, converging with the warm humid airmass from Typhoon In-Fa and WPSH. The $ {\theta }_{\rm se} $ contours became denser at Zhengzhou, and the system was intensified. The frontogenesis also strengthened and was mainly concentrated at lower levels (Figs. 3b, 4b). A large-value center of frontogenesis formed in the warm moist area in southern Zhengzhou, and the intensity reached 20 × 10⁻⁹ K m⁻¹ s⁻¹ at the center.

Figure 3.  Distributions of frontogenesis function (shadings; 10⁻⁹ K m⁻¹ s⁻¹) and ${\theta }_{\rm se}$ (contours; K) at 925 hPa at (a) 0800 BT 20 July, (b) 2000 BT 20 July, (c) 0800 BT 21 July, and (d) 2000 BT 21 July. The black dot denotes Zhengzhou meteorological station.

Figure 4.  Latitude–height sections of frontogenesis function (shadings; 10⁻⁹ K m⁻¹ s⁻¹) and θ_se (contours; K) along the longitude of Zhengzhou meteorological station at (a) 0800 BT 20 July, (b) 2000 BT 20 July, (c) 0800 BT 21 July, and (d) 2000 BT 21 July.

Meanwhile, the secondary circulation gradually enhanced (Fig. 4b). The ascending branch of the secondary circulation further strengthened, leading to more intense precipitation there. On 21 July, following the northeastward movement of the synoptic system, the front zone also moved northeastward. Due to the orographic influence (Figs. 4c, d), the front zone tilted to the north with increasing height, and the area of large value was located in eastern Taihang Mountain, where the maximum value exceeded 50 × 10⁻⁹ K m⁻¹ s⁻¹. Large-scale precipitation occurred over northeastern Henan Province and southwestern Hebei Province. Later, the $ {\theta }_{\rm se} $ contours became less dense, the weather system weakened, and the rainfall ended (Figs. 3c, 3d, 4c, 4d).

Our calculation results indicate that compared with the diabatic heating term and tilting term related to vertical motion, the horizontal divergence term and deformation term contributed more to the frontogenesis at lower levels. Therefore, the horizontal divergence term and deformation term were mainly discussed. Figure 5 shows that distribution of the horizontal divergence term agreed well with the front zone, indicating that it made positive contribution to the frontogenesis. The horizontal divergence term is the product of gradient and divergence, which reflects the intensity of the front and the divergence/convergence condition. The denser the ${\theta }_{{\rm{se}}}$ contours, the stronger the confrontation between dry and moist airmasses and the more significant the frontogenesis. At 0800 BT 20 July, obvious easterly wind shears prevailed at 925 hPa over Henan Province. Zhengzhou City and the region to its south were located in the convergence zone between southerlies and northerlies, where obvious frontogenesis was seen at the lower levels (Fig. 5a). Accompanying the westward extension and enhancement of the WPSH and approaching of Typhoon In-Fa, the pressure gradient between the WPSH and the typhoon gradually increased. Southeasterlies enhanced accordingly and the frontogenesis also increased due to the horizontal divergence term. As a result, precipitation became stronger (Fig. 5b). At 0800 BT 21 July, the weather systems gradually moved to the north. Due to the influence of topography, convergence significantly enhanced, while the frontogenesis caused by divergence gradually moved northward along the topography. Since the frontogenesis remained strong, heavy precipitation developed on the windward slope of the Taihang Mountain (Fig. 5c). Since then, the convergence, the rainfall system, and the frontogenesis all weakened significantly, and the rainstorm process came to a stop (Fig. 5d).

Figure 5.  Distributions of the horizontal convergence term of frontogenesis function (shadings; 10⁻⁹ K m⁻¹ s⁻¹), wind (barbs), and divergence (contours; 10⁻⁵ s⁻¹) at 925 hPa at (a) 0800 BT 20 July, (b) 2000 BT 20 July, (c) 0800 BT 21 July, and (d) 2000 BT 21 July. The black dot denotes Zhengzhou meteorological station.

Deformation field could lead to water vapor convergence, increasing both the temperature gradient and moisture gradient. A long and narrow belt of strong temperature and humidity contrast formed, which played a key role in frontogenesis. Distributions of horizontal deformation term and shear deformation term were displayed in Fig. 6, which shows that the horizontal deformation had great contribution to frontogenesis. Note that deformation frontogenesis corresponded well with shear deformation, and their changes were consistent. Shear deformation played a major role in frontogenesis, triggering deformation frontogenesis. Elongation deformation had relatively little effect on frontogenesis. At 0800 BT 20 July, weak frontogenesis occurred in the northern part of Henan Province and horizontal deformation term made a major contribution to the frontogenesis there. The frontogenesis was attributed to shear deformation caused by topographic influence (Fig. 6a). Following gradual intensification of the low-level southeasterlies, the effect of deformation frontogenesis also enhanced. Frontogenesis occurred in the eastern flank of the Taihang Mountain as well as in southwestern Zhengzhou City. The frontogenesis in the east of the Taihang Mountain was mainly attributed to the effect of shear deformation (Fig. 5b), while the one in southwestern Zhengzhou City was largely influenced by elongation (figure omitted). On 21 July, as the system moved to the northeast, the effect of shear deformation became more distinct, triggering strong frontogenesis in the east of the Taihang Mountain and heavy rainfall (Fig. 6c). Later, the deformation shear weakened, the effect of horizontal deformation term and frontogenesis decreased, and frontolysis occurred (Fig. 6d).

Figure 6.  Distributions of horizontal deformation term (shadings; 10⁻⁹ K m⁻¹ s⁻¹) and shear deformation term (contours; 10⁻⁹ K m⁻¹ s⁻¹) at 925 hPa at (a) 0800 BT 20 July, (b) 2000 BT 20 July, (c) 0800 BT 21 July, and (d) 2000 BT 21 July. The black dot denotes Zhengzhou meteorological station.

For the frontogenesis during the extreme rainfall in Henan Province, horizontal divergence term and deformation term played a key role and their contributions were equivalent. On 20 July, horizontal divergence term made major contribution to frontogenesis in Zhengzhou and the region to its southwest, while the horizontal deformation term led to frontogenesis in the east of the Taihang Mountain. On 21 July, as the rainfall was more enhanced and more concentrated, moving to the northeast, the horizontal divergence term and deformation term were superimposed with each other, both contributing to enhanced frontogenesis.

The atmospheric heat sources have very important forcing effects on the development and maintenance of various weather systems. Water vapor not only affects the atmospheric moisture content, but also absorbs or releases a large amount of latent heat during its phase change, which imposes heating or cooling effects on the atmosphere. Using apparent heat source (Q₁) and moisture sink (Q₂), we can quantitatively analyze horizontal and vertical structure of the atmospheric heat sources, and explore the dynamic and thermodynamic processes for the initiation and maintenance of rainfall.

Figure 7 shows vertical profiles of Q₁, Q₂, and vertical velocity averaged over the heavy rainfall area (33.5°–36.5°N, 112.5°–115°E). It is seen that the atmospheric heating in the vertical direction mainly occurred over the layer of 700–400 hPa and the heating center of Q₁ was located at 500 hPa, where the heating rate was close to 2.5 K h⁻¹. The vertical distribution of Q₂ was relatively homogeneous over 850–400 hPa, and the heating center was slightly lower than that of Q₁, with the heating rate of 0.7 K h⁻¹, which was also smaller than that of Q₁. Meanwhile, strong ascending motion could be found in the rainstorm area with the center located at 600–500 hPa and the intensity of 0.9 Pa s⁻¹. The vertical structure of Q₁ shows a distinct unimodal pattern, while that of Q₂ is uniformly distributed over 850–400 hPa. This is because the warm moist airmass at lower levels was uplifted to the middle troposphere by strong ascending motion, leading to temperature increase over the rainstorm area (Fig. 7a). On 21 July, the synoptic systems gradually moved northeastward and weakened. Correspondingly, Q₁, Q₂, and vertical velocity all significantly decreased (Fig. 7b).

Figure 7.  Vertical profiles of Q₁ (K h⁻¹), Q₂ (K h⁻¹), and vertical velocity (Pa s⁻¹) averaged over the heavy rainfall region (33.5°–36.5°N, 112.5°–115°E) on (a) 20 July and (b) 21 July 2021.

Figure 8 indicates that the distributions of <Q₁> and <Q₂> basically were consistent during the rainfall period, and <Q₁> was larger than <Q₂>, which was consistent with the distribution of heavy rainfall (Figs. 1b, c). In the middle and upper troposphere, Q₁ was larger than Q₂, indicating that the release of regional-scale strong condensational latent heat had a positive feedback on precipitation. In the heat source area, the heat flows diverged in upper levels and converged in lower levels, while the opposite was true in the heat sink area (mainly located in the WPSH area). This positive thermodynamic circulation was important for the east–west transport of angular momentum, heat, and water vapor. The effective convergence of water vapor and the uplifting and mixing of water vapor caused by ascending motions greatly raised the condensation height of water vapor, which intensified the humidity contrast with the surrounding atmosphere, and the release of huge amounts of latent heat could further promote strong development of convective system.

Figure 8.  Distributions of (a, b) <Q₁> (W m⁻²) and (c, d) <Q₂> (W m⁻²) on (a, c) 20 July and (b, d) 21 July 2021.

Figure 9 shows vertical profiles of individual components of area-averaged Q₁ (Q₂), including vertical advection term, horizontal advection term, and local variation term in the heavy rainfall region (33.5°–36.5°N, 112.5°–115°E) during 20–21 July 2021. The dominant factors that led to changes in Q₁ and Q₂ were analyzed. For Q₁, the horizontal advection term of potential temperature made a negative contribution at lower levels, and the contribution was very small. Local variation term in potential temperature remained near zero. The vertical advection term of potential temperature was always positive and its value of 2.5 K h⁻¹ was close to Q₁. Therefore, the vertical advection of potential temperature played a decisive role in Q₁ (Fig. 9a). As the vertical motion weakened, Q₁ also gradually decreased (Fig. 9b).

Figure 9.  Vertical profiles of area-averaged Q₁ (K h⁻¹) in the heavy rainfall region (33.5°–36.5°N, 112.5°–115°E) on (a) 20 July and (b) 21 July 2021. (c, d) As in (a, b), but for Q₂ (K h⁻¹).

For Q₂, horizontal advection of specific humidity played a major role on 20 July in the rainfall area, indicating that there existed horizontal advection of water vapor in a deep layer. Meanwhile, local variation and vertical advection of specific humidity remained near zero (Fig. 9b). On 21 July, as the weather system gradually moved to the north and weakened, horizontal advection and Q₂ also decreased (Figs. 9c, d), and the rainfall weakened too. The above analysis indicates that for the changes of Q₁ and Q₂, Q₁ was greatly affected by ascending motion and the vertical advection of potential temperature played a key role. Q₂ was affected by horizontal advection of water vapor in a deep layer, and the horizontal advection of specific humidity played a critical role.

It should be noted that the calculation of apparent heat source (Q₁) and moisture sink (Q₂) is different between theory and actuality. Theoretically, the water vapor condensation term includes latent heating of condensation due to both large-scale and convective precipitation. However, in actual calculations, averaged upward movement of the rainstorm convective system rather than the upward movement of individual convective cells is used. This is because it is hard to obtain the ascending motion scenario of a single convective cell. Therefore, the results of the calculation are more representative of large-scale latent heat release. For this reason, Q₁ and Q₂ cannot fully and accurately reflect the latent heat release due to precipitation since the extreme rainfall process in the present study included both stable latent heat release from large-scale precipitation and the latent heat release from meso- and micro-scale convective precipitation. Following the approach of Lei et al. (2017), we further calculated mean large-scale condensational latent heating rate (H_s) and mesoscale convective latent heating rate (H_c) over the periods 0800–2000 BT 20 and 0800–2000 BT 21 July.

Figure 10 displays estimates of mean heating rates during the two periods. It is indicated that H_c is much stronger than H_s, and their heating centers are located in different places (Fig. 10). On 20 July, H_s was distributed from central Henan to southern Shanxi provinces, with its maximum value centers of 4–5 K h⁻¹ over northwestern Henan and southern Shanxi provinces (Fig. 10a). The distribution of H_c estimated from intensive ground rainfall observations was more concentrated. The heating center was located near Zhengzhou, where the maximum value was 15 K h⁻¹, which was more than three times that of H_s (Fig. 10c). This indicates that there existed strong convective latent heat release during the record-breaking rainstorm in Zhengzhou and its surrounding areas on 20 July. On 21 July, following the northward movement of the rainfall, areas of latent heat release also shifted to the north (Figs. 10b, d). Large centers of H_s could be found in northern Henan, eastern Shanxi, and southwestern Hebei, with values of around 5–6 K h⁻¹ (Fig. 10b). The distribution of convective latent heating rate (H_c) had a better correspondence with the heavy rainfall area, mainly located in the areas from northern Henan to southwestern Hebei, and the maximum value of H_c in northwestern Henan was 8–10 K h⁻¹ (Fig. 10d).

Figure 10.  Distributions of (a) H_s (K h⁻¹) corresponding to averaged rainfall from 0800 to 2000 BT 20 July and (b) H_s corresponding to averaged rainfall from 0800 to 2000 BT 21 July 2021. (c, d) As in (a, b), but for H_c (K h⁻¹). The black circle denotes Zhengzhou meteorological station.

The above analysis reveals that strong latent heat release occurred during this rainstorm process. Convective latent heat release had a significant heating effect on the atmosphere, and the positive feedback of latent heat release on the heavy rain was favorable for the maintenance of rainfall. In addition to the stable latent heat release from large-scale precipitation, convective heating effect played a more important role over the entire rainfall area. On 20 July, the convective latent heating rate was over three times the large-scale latent heating rate, confirming that this rainstorm fell in the category of severe convection. Compared with other rainfall events, the value of H_s for the current “21·7” rainfall was close to that of some similar previous cases (Lei et al., 2017), yet extreme value of H_c was found for this case. For example, H_c for this rainstorm was twice that for the “7·20” rainstorm in Beijing. In addition, it is noted that the horizontal distribution of Q₁ was similar to that of H_s in Figs. 10a, b. It could be qualitatively deduced that Q₁ could better reflect the effect of large-scale latent heat release H_s, but it could not sufficiently represent the effect of convective latent heat release H_c.

Previous studies have shown that there existed strong boundary layer water vapor transport during the July 2021 extreme rainfall in Henan Province. The normalized anomaly of water vapor flux at 925 hPa has exceeded 6 standard deviations, which is highly extreme (Chyi et al., 2022). However, despite the large amount of southeasterly water vapor transport into the rainstorm area, there were also huge amounts of water vapor outflow leaving this area. In this section, we analyze the “21·7” rainfall in Henan Province from the perspective of water vapor flux and water vapor budget.

Distributions of water vapor flux vertically integrated over 1000–700 hPa and water vapor flux divergence during 20–21 July 2021 are displayed in Fig. 11. It is shown that with the gradual westward extension of the WPSH and intensification of Typhoon In-Fa moving northwestward, strong water vapor convergence occurred over Zhengzhou City and southeastern Henan Province during 0800–2000 BT 20 July. The area of water vapor convergence obviously increased, leading to the formation of easterly wind convergence and wind shear near Zhengzhou. The strong dynamic convergence and the orographic lifting effects in northwestern Henan Province were favorable for the development of abnormally large water vapor convergence (Figs. 11a, b). On 21 July, as the easterly wind and inverted trough moved to the north, the water vapor convergence region gradually shifted to northeastern Henan Province and southwestern Hebei Province, leading to heavy rains in these areas (Figs. 11c, d).

Figure 11.  Distributions of vertically integrated (1000–700 hPa) water vapor flux (arrows; kg m⁻¹ s⁻¹) and the corresponding flux divergence (shadings; 10⁻⁵ kg m⁻² s⁻¹) at (a) 0800 BT 20 July, (b) 2000 BT 20 July, (c) 0800 BT 21 July, and (d) 2000 BT 21 July. The black dot denotes Zhengzhou meteorological station and the black box indicates the region of water vapor flux convergence (33°–38°N, 112°–116°E).

Figure 12 shows vertical profiles of the water vapor budgets along the zonal (west–east; WE) and meridional (north–south; SN) directions, as well as the whole area average, over the water vapor flux convergence region (32°–38°N, 111°–116°E) on 20 and 21 July 2021. At the initial stage of the rainfall, changes in the water vapor budget mainly occurred below 700 hPa. The maximum total water vapor influx of close to 0.9 × 10⁷ kg s⁻¹ occurred at 925 hPa. The strongest net water vapor inflow along both zonal and meridional directions was located at 925 hPa, but the net water vapor inflow on the zonal direction was much stronger and occurred at lower levels. Above 850 hPa, the total water vapor outflow prevailed, which was attributed to the net outflow along the zonal direction. The net water vapor budget along the meridional direction remained inflow, although the value was small. The net water vapor inflow along the zonal direction accounted for more than 70% of the total water vapor budget (Fig. 12a). As Typhoon In-Fa was moving northwestward, the easterly jet in the boundary layer gradually intensified, and the moisture transport into Henan Province increased correspondingly. Specifically, the net water vapor inflow along the east–west direction increased and accounted for more than 90% of the total net water vapor inflow, which was up to 1.0 × 10⁷ kg s⁻¹. The level of maximum water vapor inflow along the north–south direction was raised to 850 hPa, yet the intensity did not change a lot (Fig. 12b).

Figure 12.  Vertical profiles of the water vapor flux budgets (10⁷ kg s⁻¹) along the east–west direction (solid line), along the north–south direction (dashed line), and averaged over the entire domain of the water vapor flux convergence region (32°–38°N, 111°–116°E) (dotted line) at (a) 0800 BT 20 July, (b) 2000 BT 20 July, (c) 0800 BT 21 July, and (d) 2000 BT 21 July.

On 21 July, as the weather system gradually moved to the northeast and weakened, the total water vapor budget was still dominated by the net inflow along the east–west direction. Below 850 hPa, water vapor inflow prevailed along north–south direction, but the inflow was much smaller than that along the east–west direction. Above 850 hPa, net outflow of water vapor prevailed (Figs. 12c, d). These results indicate that under the joint influences of Typhoon In-Fa and the WPSH, abundant water vapor was transported to Henan Province by strong northeasterlies. Significant dynamic convergence and orographic lifting of air flow led to abnormally strong boundary layer moisture convergence. The total water vapor budget was dominated by the net inflow along the east–west direction. The robust boundary layer moisture transport by easterlies played a critical role in the maintenance and intensification of the rainstorm in Henan Province. Compared with the total net water vapor inflow of 0.6 × 10⁷ kg s⁻¹ (He et al., 2009) during the extreme heavy rainfall in June 2005 (“05·6”) in South China, the net inflow at 2000 BT 20 July during the “21·7” extreme rainfall in Henan Province was up to 1.0 × 10⁷ kg s⁻¹, which was 40% more than that during the June 2005 rainstorm in South China. Apparently, the water vapor transport during the rainstorm process in Henan Province was indeed extreme.

Based on the ERA5 reanalysis data, frontogenesis, diabatic heating, and water vapor budget during the “21·7” heavy rainfall in Henan Province, central China are analyzed in the present study. The development and maintenance mechanisms for the synoptic-scale systems sustaining this process are revealed. Major conclusions are summarized below.

(1) During this extremely heavy rainfall, Henan Province was located in the front of a 200-hPa trough and controlled by a 500-hPa low pressure system saddled between the WPSH and the continental high. Easterly winds prevailed at 850 hPa. The convergence at low levels, coupled with the divergence at upper levels, favored the development and maintenance of the rainstorm, leading to extremely heavy rainfall in Henan Province.

(2) Under the influence of the 500-hPa low pressure system, frontogenesis played a dominant role in development of the extreme heavy rainfall. Frontogenesis largely occurred in the lower troposphere and well corresponded to the area of dense $ \theta $_se contours. Compared with other frontogenesis terms, horizontal divergence and deformation terms made much more contributions to the frontogenesis (although these two terms played almost equal parts). On 20 July, frontogenesis areas were scattered similar to the rainfall distribution; horizontal divergence term contributed the most to frontogenesis in Zhengzhou and the area to its northwest, while horizontal deformation term was responsible for the frontogenesis adjacent to the east side of the Taihang Mountain. On 21 July, as the precipitation system and frontogenesis areas unified and moved northeastward, large values of the two terms were overlapped in northern Henan, and they worked together leading to enhanced frontogenesis and extremely heavy rainfall there.

(3) Horizontal distributions of apparent heat source <Q₁> and moisture sink <Q₂> agreed well with the area of heavy rainfall. Vertical transport of potential temperature dominated Q₁ while horizontal advection of specific humidity dominated Q₂. Vertically, a maximum center of Q₁ occurred in the middle and upper troposphere whereas large values of Q₂ appeared evenly within 850–400 hPa; Q₁ was larger than Q₂ in the middle and upper troposphere. These results indicate that the synoptic-scale latent heat release had a positive feedback on precipitation during this extreme rainfall event.

(4) Analysis of water vapor budget indicates that under the influence of the westward extension of the WPSH and the westward movement of Typhoon In-Fa, easterly winds converged and formed shearlines in Henan Province. The strong dynamic convergence and topographic uplift of airmasses triggered abnormally strong water vapor convergence in the boundary layer. The total water vapor budget was dominated by net inflow along the east–west direction in the study region. The extremely strong moisture transport in the boundary layer steered by the easterly winds played a key role in maintaining and intensifying the extreme rainfall process.

It should be noted that analyses of the latent heat release related to large-scale stable precipitation (H_s) and the convective heating (H_c) due to meso- to micro-scale convective precipitation show clearly that the latter (H_c) also played a significant role in the development and maintenance of the “21·7” Henan extreme rainfall. In what way did the meso- and micro-scale convective heating contribute to the extreme rainfall and how it interacted with large- and synoptic-scale weather systems to enhance and sustain this extreme rainfall event still require further investigations.

The authors are thankful to Professor Lifu He of National Meteorological Centre, China Meteorological Administration for his great guidance on this study. His help in various stages of this work including ideas, methods, result analysis, and paper writing are greatly appreciated.