Characteristics of Water Vapor Transport during the “7·20” Extraordinary Heavy Rain Process in Zhengzhou City Simulated by the HYSPLIT Model

Abstract

Water vapor transport is an important foundation and prerequisite for the occurrence of rainstorms. Consequently, the understanding of water vapor transport as well as the sources of water vapor during rainstorm processes should be considered as essential to study the formation mechanism of rainstorms. In this study, the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model is adopted for backward tracking of water vapor transport trajectories and sources during the “7·20” extraordinary heavy rain process in Zhengzhou City of China that occurred on 20 July 2021. On this basis, the trajectory clustering method is applied to quantitatively analyze the contributions of water vapor sources, aiming to provide a basis for exploring the maintenance mechanism of this extreme rainstorm event. The spatio-temporal characteristics of this rainstorm event show that there are 4 consecutive days with the precipitation reaching or exceeding the rainstorm level across the whole Zhengzhou City, with the daily rainfall amounts at eight national meteorological stations all breaking their respective historical extreme values. The regional-averaged rainfall amount in Zhengzhou City is 527.4 mm, while the maximum accumulated rainfall amount reaches 985.2 mm at Xinmi station and the maximum hourly rainfall amount at Zhengzhou national meteorological station reaches 201.9 mm h⁻¹. The water vapor sources for this rainfall process, ranked in descending order of contribution, are the Western Pacific, inland areas of Northwest China and South China, and South China Sea. The water vapor at lower levels is mainly transported from the Western Pacific and the South China Sea, while those from the inland areas of Northwest China and South China provide a supply of water vapor at upper levels to a certain extent. The water vapor at 950 hPa is mainly sourced from the Western Pacific and South China Sea, accounting for 56% and 44%, respectively. The water vapor at 850 hPa mainly derives from the Western Pacific and the inland areas of South China, contributing 58% and 34% of the total, respectively. The water vapor at 700 hPa mainly comes from the inland areas of Northwest China and South China Sea. Specifically, the water vapor from inland Northwest China contributes 44% of the total, acting as the primary source. The water vapor at 500 hPa is mainly transported from the inland areas of South China and Northwest China, with that from the inland South China (56%) being more prominent. The water vapor at all levels is mainly transported to the rainstorm region through the eastern and southern regions of China from the source areas. Additionally, there are some differences in the water vapor trajectories at a 6 h interval.

1. Introduction

Three necessary conditions are favorable for the formation of a rainstorm, i.e., abundant water vapor (water vapor condition), strong and persistent upward motion (dynamic condition), and unstable atmospheric stratification (thermal condition) [1]. The water vapor condition is particularly notable, which is an important foundation and prerequisite for the occurrence of a rainstorm [2,3,4]. However, the water vapor in a local area is not sufficient for the formation of rainstorms, thus requiring water vapor to be continuously transported from other regions and stably maintained. Understanding the transport and sources of water vapor during rainstorm processes is essential to study the formation mechanism of rainstorms [5,6], which is of great significance. During the study on rainstorm processes, analysis on the water vapor transport mechanism is involved in both the diagnostics on a synoptic scale and the statistics on a climate scale [1,7,8,9].

Previous studies on water vapor during rainstorm processes have mainly been conducted using diagnosis analysis from a climatic aspect and trajectory tracking using the Euler method [10,11]. In recent years, with the development of the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model, the research on the transport path, sink and source of water vapor during rainstorm processes using the HYSPLIT model (https://www.ready.noaa.gov/HYSPLIT.php) has gradually increased. For the study of water vapor sources, Ramos et al. [12] identified the main atmospheric rivers that resulted in landfall in western Europe with the Lagrangian analysis method. Using the Lagrangian method, Stohl et al. [13] and Brimelow et al. [14] diagnosed the water vapor sources of extreme precipitation processes in central Europe and the Mackenzie River basin, respectively. Wang et al. [2] quantitatively analyzed the water vapor transport paths and sources during a rainstorm process in the western Sichuan Basin. For the study of channels and characteristics of water vapor transport, Jiang et al. [15] and Yang et al. [16] analyzed the water vapor transport characteristics during Jianghuai Meiyu and Huaibei rainy seasons and the differences in between based on the HYSPLIT model. Through the Lagrangian method, Li et al. [17] investigated the water vapor transport characteristics during abnormal pre-summer rainy seasons in South China and proposed that the water vapor from the Western Pacific is currently always more than that in normal years during the frontal precipitating stage. Using the HYSPLIT model, Sun et al. [18] have found three main water vapor transport channels for rainstorms in Northeast China. These previous studies indicate that there are significant interannual variations in water vapor sources of rainstorms in a certain area. Moreover, the water vapor sources and transport channels vary obviously with geographical locations under different pressure fields.

On 20 July 2021, under the joint influence of weather systems including subtropical high, continental high and Typhoon In-Fa and Cempaka [19], rarely seen extraordinary heavy rain occurred in Zhengzhou City of China, causing heavy casualties and huge economic losses. Therefore, it is very necessary to understand the formation mechanism of the rainstorm process to explore where the huge amount of water vapor in the rainstorm process came from. Moreover, so far, there has been no detailed analysis of water vapor sources or paths in the study of the “7·20” rainstorm process in Zhengzhou, and no scholars have used the HYSPLIT model to quantitatively calculate the water vapor transport contribution of various sources during the rainstorm process. In this paper, the above two aspects are studied in detail.

Hence, this study adopts the HYSPLIT model to analyze the water vapor transport during the “7·20” extraordinary heavy rain process in Zhengzhou City. This model can effectively depict the three-dimensional motion trajectory of each fluid particle or air mass during the whole process through the backward trajectory, determine the water vapor source, and quantitatively calculate the contributions of each water vapor source and transport path during the rainstorm process [20,21]. Additionally, the combination with the 1° × 1° reanalysis data from the National Centers for Environmental Prediction (NCEP) for comprehensive diagnosis can provide relatively accurate and reliable diagnosis results [22,23]. The objective of this study is to deepen the understanding on water vapor transport mechanism during extraordinary heavy rain processes and improve the application level of the HYSPLIT model in diagnostic analysis for extreme weather processes, thereby providing theoretical and technical support for further clarifying the formation mechanism of rainstorms.

2. Materials and Methods

2.1. Study Area and Rainstorm Profile

The research area is Zhengzhou, Henan Province, China, which is located in the hinterland of China, near 112°–114° East longitude and 34° North latitude. The geographical location is shown in Figure 1. An unusually heavy rainstorm occurred in Zhengzhou around 20 July 2021. and this extraordinary heavy rain process was characterized by a long duration, a large accumulated rainfall amount and high short-term rainfall intensity. From 18 July to 22 July, the regional-averaged rainfall amount was 527.4 mm across the whole Zhengzhou City, with an accumulated maximum rainfall amount of 985.2 mm and a maximum hourly rainfall amount of 201.9 mm.

2.2. Data

The data used in this study consist of the following four types. The first type is the hourly precipitation data collected from 8 national meteorological stations, namely Gongyi (34.81° N, 112.98° E), Xingyang (34.79° N, 112.98° E), Dengfeng (34.49° N, 113.11° E), Zhengzhou urban district (34.71° N, 113.66° E), Songshan (34.49° N, 113.05° E), Xinmi (34.54° N, 113.34° E), Xinzheng (34.38° N, 113.72° E), and Zhongmu (34.71° N, 113.97° E) Station (Figure 1c). The second is water vapor flux data, which are derived from the fifth-generation atmospheric reanalysis dataset produced by the European Centre for Medium-Range Weather Forecasts (ERA5). Reanalysis data “Vertical integral of divergence of moisture flux” generated by the ERA5 were also used in this research, with the time resolution of 1 h, and “the daily accumulated vertical integrals of water vapor flux”. The third is the high-altitude meteorological data observed by the conventional meteorological sounding station. Moreover, the reanalysis data from the NCEP processed by the Global Data Assimilation System (GDAS) are also employed here (ftp://arlftp.arlhq.noaa.gov/pub/archives/gdas1, accessed on 20 October 2023). The GDAS interpolates data from various observing systems and instruments. Specifically, this study assimilates the GDAS reanalysis data and interpolates these assimilated data onto a map in a conformal projection. These data are wind direction, wind speed, air pressure, temperature and specific humidity, which have a spatial resolution of 1° × 1°, 17 vertical layers, and a temporal resolution of 6 h.

2.3. Hybrid Single-Particle Lagrangian Integrated Trajectory Model

2.3.1. Trajectory Tracing

The HYSPLIT model is an air mass backward trajectory model based on the Lagrangian method developed by NOAA [24], which uses the GDAS data as input. HYSPLIT continues to be one of the most extensively used atmospheric transport and dispersion models in the atmospheric sciences community [25,26]. The principle of analyzing airflow trajectory by the HYSPLIT model involves assuming that the air parcel is drifting with the wind. The advection of an air parcel is computed from the average of the three-dimensional velocity vector for the initial position $P\left(t\right)$ and the first-guess position $P^{\prime }\left(t+∆t\right)$. The velocity vectors are interpolated in both space and time. The first-guess position and the final position at $t+∆t$ are as follows [27]:

$$P^{\prime }\left(t+∆t\right)=P\left(t\right)+V(P,t)∆t$$

(1)

$$P\left(t+∆t\right)=P\left(t\right)+0.5[V\left(P,t\right)+V(P^{\prime },t+∆t\left)\right]∆t$$

(2)

where $V$ is the velocity of the air parcel, and $∆t$ is the time step. The horizontal coordinates of the HYSPLIT model are consistent with those of the input data [28], and the vertical coordinate is interpolated into the terrain-following coordinate (Equation (3)).

$$\sigma ={\displaystyle \frac{(z_{top}-z_{mst})}{(z_{top}-z_{gt})}}$$

(3)

where $z_{top}$ is the model top, $z_{gt}$ is the elevation, and $z_{mst}$ is the lower boundary height of the model.

2.3.2. Clustering Analysis

According to the initial positions of air parcels, large amounts of trajectories can be obtained through the HYSPLIT model. As the exceptional number of trajectories is difficult to analyze intuitively, the clustering analysis is adopted [29]. The total spatial variance (TSV) is applied in trajectory clustering analysis. The trajectories that are near each other are merged and grouped into several clusters by analyzing the variations of the TSV of all clusters, where the final clustering result is achieved at the minimum TSV considering both the cluster number and the meanings of clustering. Specifically, assuming that there are $N$ trajectories, the cluster spatial variance of each cluster refers to the sum of the squared distances between the endpoints of the cluster’s component trajectory and the mean of the trajectories in this cluster. At the initial time, every trajectory is defined to be a cluster, and the spatial variance is set to 0. Then, any two clusters are randomly selected to be combined, and the pair of clusters combined are the ones with the lowest increase in the TSV. However, after a certain number of clustering iterations, the TSV again increases rapidly, indicating that the two clusters being combined are not very similar. Thus, this latter increase in the TSV suggests where to stop the clustering, and the mean trajectories of all clusters are finally obtained.

2.3.3. Schemes of Water Vapor Trajectory Tracking and Clustering Analysis

The water vapor trajectory tracking scheme is as follows. Firstly, the geographic locations of the 8 meteorological observation stations in Figure 1c with large rainfall amounts during this rainstorm process in Zhengzhou are selected as the initial positions for water vapor trajectory tracking. Next, considering that the atmospheric water vapor mostly concentrates in the middle and lower levels of troposphere [30,31], and the water vapor sources and channels vary with height during rainstorm processes, the isobaric levels 950 hPa, 850 hPa, 700 hPa and 500 hPa are selected as the initial heights for water vapor trajectory tracking. Finally, according to the temporal variation in rainfall intensity in Figure 2, the tracking period is set to 7 days. Starting from 0200 Beijing Time (BJT), 0800 BJT, 1400 BJT, and 2000 BJT from 19 to 21 July 2021, the backward trajectories of air parcels over a 7-day (168 h) period are simulated by the HYSPLIT model. The trajectory positions are output every 6 h to generate 384 trajectories in total (8 spatial initial points × 4 levels × 12 temporal initial points).

After calculating the backward trajectory based on the reanalysis data, clustering analysis is carried out for the trajectories at 950 hPa, 850 hPa, 700 hPa and 500 hPa, respectively.

3. Results

3.1. Overview of Rainstorm and Weather Situation

3.1.1. Spatio-Temporal Characteristics of Rainfall Amount

This extraordinary heavy rain process is characterized by a long duration, large accumulated rainfall amount and high short-term rainfall intensity. From 0800 BJT on 18 July to 0800 BJT on 22 July, the regional-averaged rainfall amount is 527.4 mm across the whole Zhengzhou City, with the accumulated rainfall amounts exceeding 500 mm at 107 stations. Moreover, there are 53 stations with the accumulated rainfall amount exceeding the annual average rainfall amount of Zhengzhou (641 mm). The maximum is found to be 985.2 mm at Xinmi station, and the accumulated rainfall amount at Zhengzhou national meteorological station reaches 817.3 mm. There are 4 consecutive days with the precipitation reaching or exceeding 50 mm across the whole city. Moreover, the daily rainfall amounts at the 8 national meteorological stations have all broken their respective historical extreme values. The spatial distribution of observed daily rainfall amounts in Henan Province from 18 to 21 July are displayed in Figure 2a–d, showing that the rainfall intensity is stronger and the influencing range is larger on 19 and 20 July during this rainstorm process, which impacts all counties in Zhengzhou City.

Figure 3a,b display the variations in hourly rainfall intensity from 18 to 22 July. It is revealed that the maximum hourly rainfall intensities are 47.4 mm·h⁻¹, 45.8 mm·h⁻¹, 29.2 mm·h⁻¹, 201.9 mm·h⁻¹, 53.9 mm·h⁻¹, 64.0 mm·h⁻¹, 46.2 mm·h⁻¹ and 33.4 mm·h⁻¹ at Gongyi, Xingyang, Dengfeng, Zhengzhou urban district, Songshan, Xinmi, Xinzheng and Zhongmu Station, respectively, among which the maximum hourly rainfall intensity of 201.9 mm·h⁻¹ recorded at Zhengzhou urban district station has broken the historical record since its establishment in 1951.

According to a national standard in China, the Grade of Precipitation (GB/T 28592-2012), the precipitation amounts are classified into the following three grades: rainstorm (with 24 h accumulated precipitation between 50.0 mm and 99.9 mm), downpour (with 24 h accumulated precipitation between 100.0 mm and 249.9 mm), and extraordinary heavy rain (with 24 h accumulated precipitation above 250.0 mm). Based on the area-averaged hourly rainfall intensity over the whole Zhengzhou city from 0000 BJT on 18 July to 2300 BJT on 22 July 2021, a moving average technique is performed on 24 h accumulated rainfall amounts, as shown in Figure 4. It can be seen that the precipitation amount reaches the grade of rainstorm during the period from 1400 BJT on 19 July to 2000 BJT on 19 July and 2000 BJT on 21 July to 0200 BJT on 22 July (with a duration of 12 h), reaches the grade of downpour from 2100 BJT on 19 July to 1100 BJT on 20 July and 1200 BJT on 21 July to 1900 BJT on 21 July (21 h), and reaches the grade of extraordinary heavy rain from 1200 BJT on 20 July to 1100 BJT on 21 July (23 h).

3.1.2. Weather Situation and Water Vapor Condition

This precipitation process mainly occurs during the eastward extension of South Asia high (SAH). At 0800 BJT on 20 July, as the 500 hPa subtropical high is located more northward than in normal years, the Huanghuai region lies in the low-pressure area between the SAH and subtropical high (Figure 5a). In this situation, ZhengZhou city is affected by the east–southeast airflows east of the low vortexes and shear lines at 850 hPa, with the southeasterly wind of 14 m·s⁻¹ at 0800 BJT (Figure 5b). The warm and humid cyclonic airflows from north of Typhoon In-Fa (centered at 131° E, 34° N) and east of Typhoon Cempaka (centered at 112° E, 21° N) form a strong easterly water vapor transport belt, which brings abundant water vapor to the northwest. The water vapor converges in front of the Taihang Mountain and the Funiu Mountain and remains there for a long time, forming an intense convergence area. In addition, the blocking effect of the subtropical high in the east facilitates the supply of adequate water vapor for this heavy rainfall in Zhengzhou City [32].

Under the weather situation mentioned above, the transport amount of water vapor during this extraordinary heavy rain amount is large. The water vapor flux and its divergence can quantitatively describe the direction and volume of water vapor transport and where the water vapor converges, which help us to further understand the water vapor condition that favors rainstorm. Here, the daily accumulated vertical integrals of water vapor flux divergence from 18 to 21 July are shown in Figure 6. It is indicated that a strong convergence of water vapor flux was observed in the north-central Henan Province on 19 July, with the strongest convergence ranging from −30 g·m⁻²·s⁻¹ to −40 g·m⁻²·s⁻¹. On 20 July, the convergence of water vapor flux prevailed over the whole Henan Province. Moreover, due to the gradual westward extension and enhancement of the subtropical high and Typhoon In-Fa, as well as the topographic uplifting effect, abundant water vapor converged in Zhengzhou City and in areas north of the Yellow River, with the strongest convergence ranging from −70 to −80 g·m⁻²·s⁻¹. On 21 July, the water vapor flux convergence was evidently weaker.

3.2. Transport Trajectories and Sources of Water Vapor

In order to explore the water vapor sources of this rainstorm process, the HYSPLIT model is applied in the backward tracking of water vapor trajectories during the “7·20” extraordinary heavy rain process in Zhengzhou City. The precipitation concentration period from 19 to 21 July is selected as the study period for water vapor trajectory tracking, and the backward tracking period is set to 7 days. Considering the temporal resolution of GDAS data adopted by the HYSPLIT model, 0000 UTC, 0600 UTC, 1200 UTC and 1800 UTC are selected as the initial time of the simulation.

Figure 7 and

Table 1. Transport directions and sources of water vapor in the target region at 0200 BJT, 0800 BJT, 1400 BJT and 2000 BJT from 19 to 21 July 2021.

Time	950 hPa		850 hPa		700 hPa		500 hPa
	Arrival Direction	Source	Arrival Direction	Source	Arrival Direction	Source	Arrival Direction	Source
0200 BJT on 19 July	East	Western Pacific	East	Western Pacific and East China Sea	Southeast	Philippine Sea and Western Pacific	South	South China Sea and Philippine Sea
0800 BJT on 19 July	East	Western Pacific	East	Western Pacific	South	South China Sea and Philippine Sea	South, northwest and north	South China Sea, Northwest China and North China
1400 BJT on 19 July	East	Western Pacific	East	Western Pacific	South	South China Sea	South, northwest and southwest	Beibu Gulf, Northwest China and Southwest China
2000 BJT on 19 July	East	Western Pacific	East	Western Pacific	East	Western Pacific	South, southwest and northwest	South China Sea, Southwest China and Northwest China
0200 BJT on 20 July	East	Western Pacific	East	Western Pacific	Southeast and south	South China Sea, Philippine Sea and Western Pacific	Southeast and south	Western Pacific and South China Sea
0800 BJT on 20 July	East	Western Pacific	East	Western Pacific	East and south	Western Pacific, East China Sea and South China Sea	East and south	Western Pacific and South China Sea
1400 BJT on 20 July	East	Western Pacific	East and southeast	Philippine Sea and Western Pacific	East and south	Western Pacific, South China and Southeast Asia	East, south, west and north	Western Pacific, South China, Central Asia and Mongolia
2000 BJT on 20 July	East	Western Pacific	East	Western Pacific	East and south	Western Pacific and South China Sea	East, north and southeast	North China, Mongolia and South China Sea
0200 BJT on 21 July	East	Western Pacific	East	Western Pacific	South	South China Sea	South and north	South China Sea and North China
0800 BJT on 21 July	East	Western Pacific	East	Western Pacific	South	South China Sea and South China	East and south	Western Pacific, South China Sea and Central China
1400 BJT on 21 July	East	Western Pacific	East	Western Pacific	Southeast and south	Western Pacific, the South China Sea and South China	South	South China Sea, Southeast Asia and South China
2000 BJT on 21 July	East	Western Pacific	East	Western Pacific	Southeast and south	Western Pacific and South China	South	South China Sea, Southeast Asia, South China and Central China

shows the backward trajectories of water vapor at different levels at 0200 BJT, 0800 BJT, 1400 BJT and 2000 BJT from 19 to 21 July 2021. It can be seen that the water vapor at 950 hPa and 850 hPa are basically transported from the east and southeast, while those at 700 hPa and 500 hPa are mainly from the south. The water vapor at most levels is transported from the source regions to the rainstorm region through eastern and southern China.

3.3. Clustering Analysis on Water Vapor Trajectories

Figure 8 demonstrates the TSV variations during the clustering process of water vapor trajectories at different levels. The TSV at 950 hPa increases sharply after the number of clusters drops to below 4. Therefore, the optimal clustering number at 950 hPa is 4. That is, there are four main water vapor transport channels at 950 hPa. Similarly, the optimal values of clustering numbers are 3, 4 and 2 for water vapor trajectories at 850 hPa, 700 hPa and 500 hPa, respectively. Hence, there are three, four and two main water vapor transport channels at the above three levels, respectively.

Figure 9 displays the spatial distributions of the main water vapor transport channels at different levels. At 950 hPa, the trajectories belonging to Channel 1, where the water vapor is mainly sourced from the Western Pacific, account for 36% of the total trajectories, which is the most among all channels at this level. The water vapor at Channel 2 originates from the northern Philippines, extends westwards to the Hainan Island and then is transported northward to Zhengzhou City through Guangxi and Hunan Province. The trajectories of Channel 2 account for 20% of the total. The water vapor at Channel 3 derives from the southern Taiwan Island, and then extends westwards to Guangdong Province and continuously northwards to Zhengzhou City. The trajectories belonging to Channel 3 account for 24% of the total. For Channel 4, the water vapor mainly comes from the east of the Strait of Japan to Zhengzhou City through the Yellow Sea, where 20% of the total trajectories are included. For the three main water vapor channels at 850 hPa, the water vapor at Channel 1 originates from the Western Pacific, which is transported westward to Shanghai and further northwards to Zhengzhou. The trajectories of Channel 1 account for 58% of the total trajectories at 850 hPa. For Channel 2 at 850 hPa, the water vapor is mainly sourced from Europe, which is then transported southeastward to Zhengzhou. The trajectories in Channel 2 only account for 8%, being the least among the three channels at 850 hPa. The water vapor of Channel 3 is from south of the Yangtze River, which is transported northward to Zhengzhou. The trajectories belonging to Channel 3 amount to 34% of the total. For the four water vapor channels at 700 hPa, the water vapor at Channel 1 comes from the South China Sea and extends northward to Zhengzhou through Guangdong Province, containing 36% of the total trajectories. The water vapor at both Channel 2 and Channel 3 originates from the inland areas of Northwest China and is then transported eastward to Zhengzhou. Specifically, the water vapor at Channel 2 is sourced from Tibet, consisting 20% of the total trajectories, and that at Channel 3 originates from Xinjiang with the trajectories accounting 24% of the total. The water vapor at Channel 4 comes from the Western Pacific, which is then transported northwestward to Fujian Province and then northward to Zhengzhou. The trajectories in Channel 4 account for 20% of the total at 700 hPa. For the two water vapor channels at 500 hPa, the trajectories from the inland areas of South China and Northwest China account for 56% and 44% of the total, respectively.

The contributions of water vapor from main channels at each level during the rainstorm process are further calculated, as shown in

Table 2. Contributions of water vapor sources at different levels in Zhengzhou City on 20 July 2021.

Level	Western Pacific	South China Sea	Inland Areas of Northwest China	Inland Areas of South China
950 hPa	56%	44%	0	0
850 hPa	58%	0	8%	34%
700 hPa	20%	36%	44%	0
500 hPa	0	0	44%	56%
Total contribution	134%	80%	96%	90%

. The results show that at 950 hPa, the water vapor is mainly from the Western Pacific and South China Sea, contributing 56% and 44% to the total, respectively. At 850 hPa, the water vapor is mainly derived from the Western Pacific and the inland areas of South China, with the contribution ratios being 58% and 34%, respectively. At 700 hPa, the water vapor at 700 hPa mainly comes from the inland areas of Northwest China and South China Sea. Specially, the water vapor from inland Northwest China plays a dominant role, with the contribution rate reaching 44%. At 500 hPa, the water vapor is mainly transported from the inland areas of South China and Northwest China, where the inland area of southern China contributes more with the ratio being 56%.

For the regional reversing of water vapor source contribution in 500 hPa (Western Pacific, South China Sea) and 950 hPa (inland areas of Northwest China; inland areas of South China), the mechanism and reasons are relatively complex. Here, the conclusion of research from Tang et al. [33] with a sensitivity test on water deviation distribution in this rainstorm process is introduced. The possible mechanism of water vapor transport in the lower layer is considered to be that the abnormal vortex belt of water vapor formed by the zonal distribution of low-latitude multiple vortices resulted in transport eastward from the Indian Ocean. The mechanism of water vapor transport in the upper level is considered as twin typhoons and subtropical high from the Pacific Ocean.

In summary, the clustering results reveal that the water vapor sources are ranked in descending order of their contributions as follows: Western Pacific, inland areas of Northwest China and South China, and the South China Sea. The water vapor in the lower levels mainly comes from the western Pacific and South China Sea, while the inland areas of Northwest China and South China play a supply role for water vapor in the upper levels to a certain extent.

4. Discussion

The results of this study are basically consistent with previous studies on the “7·20” extraordinary rainstorm process in Zhengzhou City. Tang et al. [33] considered that the Western Pacific typhoon not only directly dominated the moisture transport in the Northwest Pacific but also transported water vapor to the South China Sea, which determined the strength of typhoons and moisture distribution in this region. Sun et al. [34] obtained similar results using the optical flow method, and the results showed that the southwest–northeast-oriented transport belt of water vapor and cloud water extending from South China to North China through Henan Province provides favorable conditions of water vapor and cloud water for the Zhengzhou area. Chyi et al. [35,36] proposed that the net water vapor during this rainstorm process is mainly influenced by the net inflow of water vapor at both east and west boundaries, and the extremely strong water vapor transport in the boundary layer by easterly trajectories plays a key role in the maintenance and enhancement of this rainstorm. Bueh et al. [37] pointed out that on July 20, the strong northward water vapor flux zone formed on the southern side of Henan Province converges with the water vapor flux zone facilitated by typhoons and the Western Pacific subtropical high, providing abundant water vapor for the rainstorm.

However, the research perspectives and results of this study are different from those of other studies. The above research only confirmed that the water vapor source of the rainstorm process was Typhoon Chapakah, and gave the anomaly of the distribution of water vapor flux in the single layer of 925 hPa or the whole layer from the ground to the upper layer, but did not give the specific transport path in each direction from the initial position of typhoon to the rainfall site, nor did it give the characteristics of water vapor transport in different vertical pressure layers. This paper presents a detailed supplementary study on these two aspects.

The HYSPLIT model has some shortcomings in studying the water vapor transport process. The spatiotemporal resolution of GDAS reanalysis data adopted by the model is not high enough, and there may be some deviation between the simulated trajectory and the real situation, which may affect the final clustering results. In future work, we will take the use and processing of data as one of the most important tasks.

Although the Lagrange method used in this paper can accurately provide a detailed water vapor transport path and source information and quantitatively analyze the source and transport process of water vapor by tracking specific gas blocks or tracks, the Euler method is needed to obtain the transport flux of water vapor at fixed locations and qualitatively study the characteristics of water vapor transport. It should be noted that the Euler method is suitable for large-scale meteorological analysis; for example, we need to assess the water vapor transport profile of a season or a year, and the Euler method may be more efficient. In some cases, the two methods can be combined to obtain a more comprehensive analysis of water vapor transport. The research object of this study is the characteristics of water vapor transport during specific weather events, so the Lagrange method is considered here.

In this study, the characteristics of water vapor transport trajectories and sources during the “7·20” extraordinary heavy rain process in Zhengzhou City are preliminarily explored based on the HYSPLIT model. However, the specific mechanism of moisture transport caused by the interaction between tropical cyclones and mid-latitude systems remains unclear [33], and the mechanisms involved in this rainstorm remain to be solved. For example, what is the specific role of the topographic factors in organizing and maintaining this extraordinary heavy rain process in the west of Zhengzhou City? What is the mechanism for water vapor supply by typhoons Cempaka and In-Fa during the rainstorm process? We will continue to conduct research on these issues.

5. Conclusions

In this study, the precipitation features during the “7·20” extraordinary heavy rain process in Zhengzhou City were analyzed, and the HYSPLIT model was employed to investigate the characteristics of water vapor transport during this process based on the GDAS data generated by NCEP. The objective is to provide an essential and reliable research basis for exploring the water vapor sources of this process. The main conclusions are as follows.

During this extraordinary heavy rain process, there are four consecutive days with daily precipitation reaching or exceeding the rainstorm level across the whole Zhengzhou City. The daily rainfall amounts at the eight national meteorological stations have all exceeded their historical extreme values since their establishment. The regional-averaged rainfall amount over Zhengzhou City is 527.4 mm, with the maximum accumulated rainfall amount reaching 985.2 mm at Xinmi Station. For Zhengzhou national meteorological station, the accumulated rainfall amount is 817.3 mm, with the maximum hourly rainfall intensity reaching 201.9 mm h⁻¹. This extraordinarily heavy rain lasts for 23 h.

The clustering of water vapor transport trajectories reveals that the water vapor sources in descending order of their contributions are as follows: Western Pacific, inland areas of Northwest China and South China, and the South China Sea. The water vapor in the lower levels mainly derives from the Western Pacific and the South China Sea, while the inland areas of Northwest China and South China play a supply role for water vapor in the upper levels to a certain extent. The water vapor is transported from source regions to the rainstorm region through the eastern and southern regions of China. The water vapor trajectories at a 6 h interval show some differences. At 950 hPa, the water vapor is mainly sourced from the western Pacific and South China Sea, accounting for 56% and 44%, respectively, and the arrival directions of water vapor are essentially east and southeast. At 850 hPa, the water vapor mainly derives from the Western Pacific and the inland areas of South China, with the contribution ratios being 58% and 34%, respectively, and the arrival directions are also essentially east and southeast. At 700 hPa, the water vapor mainly comes from the inland areas of Northwest China and South China Sea. Specifically, the water vapor from inland Northwest China plays a dominant role, with the contribution rate being 44%, and the arrival direction is mainly south. At 500 hPa, the water vapor is mainly transported from inland areas of South China and Northwest China, where the contribution from the inland areas of South China (56%) is more prominent, and the arrival direction is mainly south.