Quantitative Analysis of Water Vapor Transport during Mei-Yu Front Rainstorm Period over the Tibetan Plateau and Yangtze-Huai River Basin
There are continuous precipitation systems moving eastward from the Tibetan Plateau to the middle and lower reaches of the Yangtze-Huai River during the Mei-yu period. We selected 20 typical Mei-yu front precipitation cases from 2010 to 2015 based on observational and reanalysis data and studied the characteristics of their environmental fields. We quantitatively analyzed the transport and sources of water vapor in the rainstorms using the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT_4.9) model. All 20 Mei-yu front precipitation cases occurred in a wide region from the Tibetan Plateau to the Yangtze-Huai River. The South Asian high and upper level jet stream both had strong intensities during the Mei-yu front rainstorm periods. Heavy rainfall mainly occurred in the divergence zone to the right of the high-level jet and in the convergence zone of the low-level jet, where strong vertical upward flows provided the dynamic conditions required for heavy rainfall. The water vapor mainly originated from the Indian Ocean, Bay of Bengal, and South China Sea. 52% of the air masses over the western Tibetan Plateau originated from Central Asia, which were rich in water vapor. The water vapor contribution at the initial position was only 41.5% due to the dry, cold air mass over Eurasia, but increased to 47.6% at the final position. Over the eastern Tibetan Plateau to the Sichuan Basin region, 40% of the air parcels came from the Indian Ocean, which was the main channel for water vapor transport. For the middle and lower reaches of the Yangtze River, 37% of the air parcels originated from the warm and humid Indian Ocean. The water vapor contribution at the initial position was 38.6%, but increased to 40.2% after long-distance transportation.
Mei-yu front rainstorms are major meteorological disasters in the Yangtze River drainage basin in summer and are characterized by their rapid onset and diversity. The structures and properties of Mei-yu fronts vary among events and regions, or even between different episodes in one precipitation event . Although there are still some difficulties in determining the timing, location, and quantitative forecasting of rainstorm cases , their water vapor transport characteristics in terms of the climatic region and the timing of rainstorms are worth exploring .
There have been many studies of precipitation in Mei-yu fronts. Great progress has been made in many aspects, including the classification of Mei-yu fronts [4, 5], cloud structure [6–8], terrain effects [9, 10] and mesoscale systems [11–13]. In addition, abundant water vapor is necessary to form the heavy rainfall in Mei-yu fronts . Therefore, an analysis of the sources and transport of water vapor is important in studies of the mechanisms of Mei-yu front precipitation . Qin et al. analyzed rainstorms in Mei-yu fronts in the middle and lower reaches of the Yangtze River from June 21 to 24, 2002, using water vapor imagery. It was found that tropical water vapor in the upper troposphere extends from the eastern Tibetan Plateau to the Yangtze River area and interacts with water vapor in the mid-latitudes; rain clouds form in this tropical water vapor [16, 17]. Most of the water vapor in Mei-yu front rainstorms is concentrated below 800 hPa, with the low-level jet being the largest provider of water vapor, which means that heavy rainfall occurs in the water vapor transport convergence zone at middle and low levels .
A previous study by Sun and Zhang  pointed that when short-wavelength troughs move eastward from the Tibetan Plateau, they can interact with the western section of Mei-yu fronts and favor the formation of new cloud clusters. The Mei-yu front is seen as a band-like cloud area on satellite cloud images, with a horizontal extent of up to thousands of kilometers [20–22]. Mei-yu front precipitation usually shows a low brightness temperature at the cloud top and a heavy and uneven distribution of rainfall [23, 24]. The Tibetan Plateau Atmospheric Scientific Experiment has shown that most of the continuous heavy rain events in the Yangtze River Basin are associated with a continuous series of convective clouds spreading eastward from the Tibetan Plateau [25, 26]. Other studies have shown that these convective clouds produce a southwesterly vortex, triggering convective activity in the Mei-yu front around the middle and lower reaches of the Yangtze River [27–29]. However, previous studies selected different samples and used different data and methods, so their conclusions vary [6, 24, 30]. It is, therefore, necessarythat the characteristics of large-scale circulation in both the Tibetan Plateau and the Yangtze-Huai River Basin should be investigated from a climate mean period and a region with a large number of cases.
Most early research on the source and transportation of water vapor in Mei-yu fronts was based on single cases and focused on the qualitative features of water vapor transport using the Euler method [14, 16]. Studies of the quantitative characteristics of water vapor transport are rare [15, 18]. We selected examples of rainstorms on Mei-yu fronts from 2010 to 2015 and used the Lagrangian trajectory model HYSPLIT_4.9  to analyze the sources and transportation of water vapor from the Tibetan Plateau to the middle and lower reaches of the Yangtze River during the precipitation process of Mei-yu fronts. The goal of this study was to reveal the quantitative moisture transport characteristics of Mei-yu front rainstorms from the Tibetan Plateau to the Yangtze-Huai River Basin, so as to improve the monitoring skill of these destructive rainstorms.
2. Selection of Mei-Yu Front Rainstorms
Luo et al.  analyzed the typical structure of Mei-yu fronts and found that the frontal surface is formed by a strong potential pseudo-equivalent temperature gradient. The south side of the front is warm and humid, whereas the north side is part of a different air mass. The south side forms part of the southwest monsoon system, whereas the north side has an easterly flow. Upwelling and heavy precipitation mainly occur at the frontal surface. A high-level subtropical front is seen in the upper troposphere above the Mei-yu front, which matches the subtropical upper level jet. The tropical easterly wind in the upper troposphere and the subtropical upper level westerly jet form the upper level divergent flow field during the progress of the Mei-yu front rainstorm.
In order to select examples of Mei-yu front precipitation along the Yangtze River Basin from 2010 to 2015, we analyzed the shape and development of the Mei-yu front using FY2E_TBB satellite data and daily precipitation data at 753 observation stations (0800–0800 h Beijing time) from the National Meteorological Information Center of China Meteorological Administration, routine meteorological sounding data, and ERA-Interim reanalysis data . The definitions and standards of heavy rain in this article are obtained from the “Yearbook of Heavy Rain (2008)” from China Meteorological Administration: (1) 24 h total precipitation more than 50 mm; (2) one heavy rainfall process should have at least 15 stations exceeding 50 mm within 24 h, and more than 5 stations should have recorded heavy rain on the start and end dates; and (3) the precipitation area is the Yangtze-Huai River Basin (27°N–36°N, 97°E–120°E). Then, we chose Mei-yu front precipitation cases from all the heavy rain processes by using the TBB data, routine meteorological sounding data, and ERA-Interim reanalysis data.
Figure 1 shows the time-varying meridional profiles of the near-surface specific humidity and zonal wind velocity component in June–July 2010. The zonal wind velocity component changed significantly during the four precipitation periods of June 17–20, July 3–7, July 9–14, and July 16–20, 2010, together with the specific humidity, which has a good relationship with the beginning and end of rainfall processes. Figure 1 shows that the zero line of the zonal wind is the boundary between the easterly and westerly winds, which can be used to distinguish the horizontal shear in the low-altitude wind field of the Mei-yu front. At a certain latitude, when the wind direction changes from the northerly winds to southerly winds, the specific humidity also displays abrupt increases. These two variables are important indicators of the occurrence of Mei-yu front precipitation. In addition to the specific humidity and zonal wind, changes in other circulation elements are also taken into account in determining the rainstorm period of the Mei-yu front (figures omitted): (1) The front in the middle and lower troposphere is formed by a strong horizontal gradient of θse. (2) In the south of the Mei-yu front, there is a warm and humid air mass with high θse, while in the north there is low temperature and dry air mass. (3) The wind field of the Mei-yu front performs southwest monsoon and easterly wind in the south and north sides, respectively. Based on these criteria, we selected 20 Mei-yu front precipitation events along the Yangtze-Huai River Basin (27°N–36°N, 97°E–120°E) during the time period 2010–2015 (Table 1).
The total precipitation (amount of 76 days for all cases) in the Yangtze River Basin gradually increased from the Tibetan Plateau toward the east. Most of the Tibetan Plateau (region I, Figure 2) recorded ≤500 mm of precipitation, whereas the recorded precipitation in Shannan and western Nyingchi was 500–1000 mm. The amount of precipitation in the transitional region from the eastern Tibetan Plateau to the western Sichuan Basin (region II) was 500–2000 mm, reaching 1000–2000 mm in some areas of the western Sichuan Basin. The amount of precipitation in the middle and lower reaches of the Yangtze River (region III) was 1000–2000 mm and exceeded 2000 mm in southern Anhui and southern Jiangsu.
Zhang et al.  classified Mei-yu front rainstorms in China into three types: (1) β-mesoscale convective rainstorms, (2) torrential rain caused by the initial cyclone on the east side of the Mei-yu front (east of 115°E), and (3) persistent rainstorms in front of the deep high-level trough at the western end of the Mei-yu front. The 20 Mei-yu front precipitation events studied here fell into these three types (Table 1)—for example, the rainfall that occurred on June 17–20, 2010, belongs to the third type of rainstorm. Persistent heavy rainfall occurs to the south of the Yangtze River and in South China as a result of the high-altitude trough and the shear line at middle and low levels. Heavy rainstorms were experienced in Fujian, Guangdong, Guangxi, Jiangxi, Hunan, Zhejiang, and Jiangxi provinces, and 25 counties (cities) in Jiangxi Province experienced heavy rainfall on June 19, 2010. The number of stations recording heavy and torrential rain exceeded the historical record (23 counties in June 1998) for extreme amounts of rain in Jiangxi Province, which experienced the most severe rainstorm process since meteorological records began. The daily precipitation in Jinxian, Yujiang, Dongxiang, Zixi, and Nanchang was 329, 321, 328, 318, and 204 mm, respectively. Torrential rain also occurred on the same day in Shunchang (252 mm) and Wuyi Mountain (324 mm) in Fujian Province, with the daily precipitation exceeding the historical maximum record.
3. Large-Scale Circulation Features of Mei-Yu Front Rainstorms
Although there are some differences between the 20 cases of Mei-yu front precipitations in our study, our analysis is based on events occurring in a fixed area (the Yangtze-Huai River Basin) in a fixed time period (June–July). Therefore, the climatic regions (average areas of all 20 cases) and temporal means (time period contains every day in 20 cases) give meaningful statistics, which is the scientific basis for this study of the environmental characteristics of Mei-yu front rainstorms.
The environmental fields of 20 Mei-yu front precipitation events were analyzed using the composite technique, which is a statistical method. Figure 3(a) shows the average geopotential height and wind fields at 200 hPa. The South Asian high is clearly stronger than the summer climatological average , and the main body of the South Asian high lies in the region 30–115°E, the center is located on the south side of the Tibetan Plateau, and the ridge line is near 27°N. The westerly jet is maintained on the north side of the South Asian high at about 40°N and covers a wide area, with a central velocity >50 m/s. The Mei-yu front cloud belt is located in the anticyclonic circulation current on the northern edge of the South Asian high to the right of the westerly jet (not shown). The subtropical high is stable at 500 hPa (Figure 3(b)) in southeastern China. The ridge line is near 25°N and the western end of the 5880-gpm line is near 115°E. There is a clearly defined low-pressure trough over the Indian Peninsula and a weak ridge over the Qilian Mountains.
The 850 hPa mean wind field (Figure 3(c)) shows that a southwesterly jet in maintained on the southern side of the Mei-yu front cloud belt. The Somali cross-equatorial flow from the Arabian Sea through the Bay of Bengal reaches south of the Yangtze River and extends eastward to the eastern Japan Sea, forming a southwest-northeast band over China. The central wind speed of the jet reaches 20 m/s, which is stronger than the summer climatological average wind speeds in this region. The horizontal temperature gradient of the Mei-yu front is small (Figure 3(c)). At this stage, northeast China is dominated by a cold air mass at 850 hPa, and two high-temperature centers (>20°C) are located to the south of the Yangtze River and in northwest China. The isotherms of the Mei-yu front heavy rain area are sparse, indicating that there is no clear temperature gradient. This is because the air mass in the northern region sinks under the control of denaturing cold and high pressure, resulting in adiabatic warming, whereas the temperatures are decreased in the southern area where precipitation occurs as a result of the shielding effect of thicker clouds and the cooling effect caused by heavy precipitation .
The high- and low-level jets play an important role in Mei-yu front precipitation. The jets diverge at high levels and converge at low levels, which promotes the development and strengthening of vertical upward motion and provides the necessary power and water vapor conditions for heavy rainfall. Figure 3(b) shows that the most important sources of water vapor during the precipitation period of the Mei-yu front are the Bay of Bengal, the center of the Indochina Peninsula, and the South China Sea; the South China Sea contributes the largest amount of water vapor. Water vapor from the northwest Pacific Ocean is also imported into the region near the South China Sea. These two sources of water vapor form a significant water vapor convergence zone in the middle and lower reaches of the Yangtze River, as shown by the water vapor flux divergence field (Figure 3(c)). The divergence of the water vapor flux is clearly negative in the Mei-yu frontal area in the middle and lower reaches of the Yangtze River. This continuous convergence of water vapor throughout the lower troposphere favors the occurrence of heavy rainfall. In addition, the strong low-level southwesterly jet strengthens convergence at low levels, which increases the vertical velocity and provides sufficient water vapor for the heavy rainfall of the Mei-yu front (Figure 3(b)). The southeasterly jet on the south side of the South Asian high is much stronger and transports a large amount of water vapor.
As a summary of the above analysis, the South Asian high and the high-level westerly jet cover a wide area during the precipitation stage of Mei-yu fronts. The locations of the low-level southwesterly jet on the north side of the subtropical high and the eastern Mei-yu front precipitation area are coincident with each other. The whole Mei-yu front area is located to the right of the upper jet stream and to the left of the low-level jet stream. The area of heavy rainfall is mainly located in the area of divergence to the right of the upper jet and the area of convergence of the lower jet, where the vertical updraft is stronger, which provides dynamic lifting conditions. It is consistent with previous studies on Mei-yu heavy rainfall .
4. Characteristics of Water Vapor Transport
The preceding analysis mainly focused on the Eulerian method for water vapor transport of the Mei-yu front precipitation. However, due to the instantaneous change in the atmospheric wind field, the change in the water vapor flux given by the Eulerian method also presents instantaneous characteristics with time. It therefore gives a simple water vapor transport path, but cannot quantitatively determine the relationship between the sources and sinks of the water vapor and the contribution of each water vapor source to the precipitation [36, 37].
4.1. Water Vapor Trajectory Simulation Program
The current work used the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT_4.9)  model running with the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP-NCAR) reanalysis data from May to July 2010–2015 to identify the origins of the air masses observed in these Mei-yu front precipitation events. The HYSPLIT_4.9 model assumes that when a particle passively follows the winds, the starting points do not necessarily have to be located on the data grid points as HYSPLIT uses a bilinear interpolation to calculate meteorological variables at times and locations between the standard times and grid points available in the gridded data sets. The final position is computed from the average velocity at the initial position (P) and first-guess position (P′)where the integration time step is variable: Δt < 0.75 Umax, Umax stands for one meteorological grid, here Δt = 6 hrs. In this article, three-dimensional back trajectories were calculated to track the sources of mid-level (1500 m above the ground) moisture. The variables included the geopotential height, temperature, zonal wind, meridional wind, specific humidity at 1000–300 hPa, and the vertical velocity of each level from 1000 to 100 hPa, and at 6 hourly temporal resolution and 2.5°latitude/longitude grid. The calculation method for the water vapor contribution has been described in detail by Jiang et al. .
The start and end dates of the 20 Mei-yu front precipitation processes (Table 1) determined the simulation periods. The simulated target area was divided into three areas according to the distribution of precipitation (Figure 2) and the differences in topography (Figure 4, where region I is the Tibetan Plateau, region II is the transition zone from the eastern plateau to the western Sichuan Basin, and region III is the middle and lower reaches of the Yangtze River). A total of 71, 35, and 74 points with the spatial resolution 1° × 2°(longitude/latitude grid) were identified as the starting points of the trajectories in these three regions. As most of the water vapor transport is concentrated in the mid- and lower troposphere, a height of 1500 m above the ground was selected as the initial height of the simulation. The position of air parcels was output every 6 hours, with 11 days backward trajectories at each grid point. We obtained the specific humidity at the corresponding position using physical variables such as relative humidity and temperature.
4.2. Water Vapor Tracking of Mei-Yu Front Rainstorms
We can clearly identify the distribution of specific humidity by statistical analysis of the simulated backward trajectory of air parcels in the three regions during Mei-yu front rainstorm periods. Figure 5 shows the spatial distribution of air parcels and the water vapor carried by them in the three regions from the Tibetan Plateau to the middle and lower reaches of the Yangtze River during 20 Mei-yu front rainstorm events. On the day before the target area is reached (day −1), the air parcels and specific humidity are concentrated in or near each target area, but there is a significant difference in height and temperature in the different target areas. The large value contours representing the number of air parcels over the Tibetan Plateau incline westward (Figure 5(a)), indicating that the westerly wind has a dominant role. The height of the air parcels over the plateau and the region to the north are above 600 hPa, whereas they are below 800 hPa to the south of the plateau. The area of high humidity in region I inclines to the southeast (exceeds 3000 g/kg) (Figure 5(b)) because the air parcels from the south carry more water vapor. The temperature of the air parcels over the central and northwestern plateau is < 0°C, whereas that of the air parcels to the southeast of the plateau is >15°C. The height of the air parcels gradually increases as the temperature decreases from the southeast to northwest over the Tibetan Plateau.
There are some differences between regions I and II. The large number of air parcels and high humidity in the transition area from the Tibetan Plateau to the Sichuan Basin are located to the southwest (Figures 5(c) and 5(d)), indicating that a southwesterly wind prevailed during Mei-yu front precipitation in this area, with most water vapor from this direction. The shading in Figure 5(c) shows that the southwesterly airflow is low in height and high in temperature, whereas the northwesterly airflow through the plateau is generally above 600 hPa and the temperature is <0°C. The highest water vapor center is located to the southwest of the Tibetan Plateau, with a specific humidity larger than 4000 g/kg. The air parcels in region III on day −1 are distributed over a larger range than in regions I and II, and the water vapor extends into the South China Sea and the western Pacific region, indicating a higher wind speed in the southeasterly direction. The largest number of air parcels is located to the southwest, and the isoline of humidity is further south than that of the air parcels because the southwesterly airflow carried more specific humidity. Moreover, the large value center of specific humidity in region III is more than 6000 g/kg, which is significantly larger than that in regions I and II, due to the larger amount of moisture carried by the airflows. The height (temperature) increases (decreases) from the southeast to northwest.
The air parcels and specific humidity were tracked over much longer distances six days before reaching the target area (Figure 6). From region I (the Tibetan Plateau), they can be traced southward to the Arabian Sea and the Bay of Bengal, eastward to the middle reaches of the Yangtze River, northward to Siberia, and westward to the Atlantic Ocean. The highest number of air parcels and the highest humidity are found near the Pamir Plateau, west of the Tibetan Plateau (Figures 6(a) and 6(b)). The air parcels are found at higher altitudes and lower temperatures westward from the Tibetan Plateau. The air parcels to the east and south of the Tibetan Plateau are below 900 hPa at temperatures >20°C. The large numbers of air parcels and high humidity in region II show a zonal distribution located in the Sri Lanka-Bay Mounds-northern Indochina Peninsula region. Smaller numbers can be tracked eastward to the coast and westward to Eurasia. The southward range of air parcels and specific humidity from region III (the middle and lower reaches of the Yangtze River) is clearly larger than that of regions I and II and extends into the equatorial Indian Ocean and the western Pacific Ocean. The area with the highest number of air parcels and the highest humidity is located near the Bay of Bengal and the Indochina Peninsula.
Figure 7 shows the spatial distribution of air parcels and specific humidity tracked 11 days before 20 Mei-yu front rainstorm events. The air parcels from all three regions can be traced south of the equatorial Indian Ocean and those from region III have been pushed southward to Australia and eastward across 180°E to the mid-Pacific Ocean (Figures 7(e) and 7(f)). By contrast, the air parcels from region I barely reached the Pacific Ocean (Figure 7(a)). The maximum number of air parcels and highest humidity in regions I and II are mainly found near the Maldives islands in the northern Indian Ocean, whereas the distribution of the maximum number of air parcels and highest humidity in region II are more concentrated (Figures 7(c) and 7(d)). The maximum values in region III are located in the north Indian Ocean, and the large value region extends to the western Pacific Ocean, indicating that the southeasterly wind is as strong as the southwesterly wind. After 11 days of backward tracking of these three regions, few air parcels are distributed over the Tibetan Plateau, indicating that the plateau provides little water vapor. According to the height and temperature distribution of the air parcels (shading, Figure 7), most of the air parcels over the Indian Ocean, the Bay of Bengal, the South China Sea, and the Pacific Ocean are below 800 hPa and the temperatures are >10°C. By contrast, the air parcels over Eurasia and the North Atlantic Ocean are mainly located above 700 hPa and the temperatures are <0°C—that is, the air masses from the southeasterly ocean are low altitude, warm, and humid, whereas those from the northwest mainland are high altitude, dry, and cold.
The tracing and analysis of the air parcels and specific humidity during the Mei-yu front precipitation show that the sources of water vapor are mainly distributed over the Indian Ocean, the Bay of Bengal, the South China Sea, the western Pacific Ocean, and Eurasia.
4.3. Contribution of Water Vapor Transport to Mei-Yu Front Rainstorms
The back trajectories of air parcels during Mei-yu front precipitation appeared messy, so we clustered all the trajectories. In order to identify the most significant patterns starting from a wide number of calculated trajectories, suitable criteria to gather them around representative modes have been borrowed from clustering analysis techniques. In this article, a clustering algorithm  has been selected in the HYSPLIT model due to good performance. Minimization of the differences among air mass trajectories in a cluster, and maximization of the differences between clusters, was achieved by minimizing the Euclidean distances between the corresponding coordinates of the individual trajectories (considering the full length of each 11-day backward trajectory).
4.3.1. Air Parcel Channels for Water Vapor
Five water vapor delivery channels for each of the three regions were calculated using the cluster method (Figure 8). Among the air masses reaching region I, channel 1 originates from the North Atlantic Ocean and reaches the Tibetan Plateau after passing over Europe and the Ural River. This channel has the longest transport distance and consists of high-level northwesterly flows, accounting for 3% of the total trajectories. Northern channel 2 originates west of the Altai Mountains and approaches the Tibetan Plateau from the northern side. It has the shortest transport distance and consists of a northerly airstream, accounting for 21% of the total trajectories. Central Asian channel 3 originates near the Caspian Sea and crosses the Pamir Mountains to reach the western part of the Tibetan Plateau with a short transport distance. Channel 3 accounts for 52% of the total number of trajectories because of the dominance of westerlies over the Tibetan Plateau. Channel 4 originates from the western Indian Ocean around the Seychelles and passes through Somalia and the Indian Peninsula before reaching the southern Tibetan Plateau. The path of this channel is close to the Somali jet and accounts for 8% of the total trajectories. Channel 5 originates near Sri Lanka, crosses the Bay of Bengal, and makes landfall in the lower reaches of the Ganges River. It delivers water vapor to the southern Tibetan Plateau, accounting for 16% of the total trajectories.
There are significant differences between regions II and I in the sources and proportions of water vapor transport (Figure 8). The transitional zone from the eastern Tibetan Plateau to the Sichuan Basin (region II) is complex, with a drop in elevation >1000 m. Western channel 1 originates from the North Atlantic Ocean and only accounts for 1% of the total trajectories. Channel 2 originates from the southern end of the Ural Mountains in Central Asia, passing through Balkhash Lake and Xinjiang before arriving in region II. This channel consists of a northwesterly airstream, accounting for 14% of the total trajectories. Eastern channel 3 has the shortest transport distance and shows local effects, accounting for 20% of the total trajectories. Both channels 4 and 5 are from the south, originating in the middle of the Indian Ocean and the eastern Bay of Bengal, respectively. They enter region II after crossing the Indian Peninsula, accounting for 40 and 25% of the total trajectories, respectively. They are the main transport channels for water vapor.
The middle and lower reaches of the Yangtze River (region III) only have one cold air transport path, which is from the northwest and originates near the Ural Mountains before passing through the Altai Mountains and Inner Mongolia into region III, accounting for 12% of the total trajectories (channel 1, Figure 8). Channel 2 from the east is very short and delivers local sources of water vapor, accounting for 16% of the total trajectories. Channel 3 consists of a southwesterly airflow, originating from the Indian Ocean before passing through the Bay of Bengal and the Indochina Peninsula and arriving in the middle and lower reaches of the Yangtze River, accounting for the largest proportion of the total trajectories (37%). Channel 4 is the South China Sea transport path, which originates from the Malay Peninsula, accounting for 27% of the total trajectories. Channel 5 is the only western Pacific channel that consists of a southeasterly airflow. It is transported northwestward through the South China Sea into the lower reaches of the Yangtze River. This channel mainly reflects the transport of water vapor in the southeast airstream near the western Pacific subtropical high.
We calculated the evolution of height, temperature, and humidity in each channel during the transport of water vapor. For region I (the Tibetan Plateau; Figure 9(a)), both the west Indian Ocean path (channel 4) and the Bay of Bengal path (channel 5) originate from the low-latitude ocean surface at about 1000 hPa. The height increases rapidly in the 48 h before Mei-yu front rainstorms. The initial specific humidity of these two airflows are high (14.5 g/kg and 18.2 g/kg for channels 4 and 5, respectively; Figure 9(b)) because of the strong evaporation in the tropical ocean surface. They clearly decrease when they pass over the Indian Peninsula in the 48 h, which may be due to the rising topography and precipitation. The north path (channel 2) of region I is the shortest path with an initial height of 900 hPa and a specific humidity of 8.5 g/kg. The moisture decreases through the desert area of Xinjiang, down to 4.0 g/kg, as the airflow rises to the Tibetan Plateau. The initial height of the Central Asia path (channel 3) is 800 hPa with a specific humidity of 4.8 g/kg. The final humidity of this channel is 3.3 g/kg, lower than that of channels 4 and 5. The initial height of the North Atlantic path (channel 1) is near to 500 hPa with 0.7 g/kg humidity. It is mainly composed of a high-level westerly jet, which stays above 600 hPa.
In region II, the height and specific humidity of the North Atlantic path (channel 1), Indian Ocean path (channel 4), and southern path (channel 5) are almost the same as in region I (Figures 9(c) and 9(d)). The Central Asia path (channel 2) has an initial height of 800 hPa and a specific humidity of 4.9 g/kg. The height of this channel increases continuously, but the moisture does not change significantly until it arrives at region II. Although the transmission distance of the local path (channel 3) is short, there are clear changes in height. The initial height is below 900 hPa and eventually reaches 560 hPa with the moisture gradually decreasing. In region III, the middle and lower reaches of the Yangtze River, the Indian Ocean path (channel 3), and the South China Sea path (channel 4) originate from the ocean surface at 950 hPa. The specific humidity first increases due to the evaporation from warm tropical oceans but decreases at last locations. The local path (channel 2) and the western Pacific path (channel 5) have different initial locations, but their initial heights are both 850 hPa and their humidity is almost the same at about 10 g/kg. The northwest path (channel 1) consists of dry and cool air from the mid-troposphere over Eurasia and has a specific humidity of 3.1 g/kg and 7.6 g/kg at initial and final locations, respectively (Figures 9(e) and 9(f)).
4.3.2. Water Vapor Contribution for Mei-Yu Front Rainstorms
The ratio of each channel in Figure 8 is the percentage of trajectory clustering of air parcels. The fractional contribution of water vapor (Qcon) carried by each considered channel is calculated as follows:where q is the specific humidity carried by each individual trajectory, m is the number of clustered trajectories for each channel, and n is the total number of computed trajectories. Water vapor is lost or replenished as a result of precipitation or evaporation during transportation, so the contribution of water vapor at the initial (source) and final locations of the channel is calculated simultaneously (Table 2). The largest contribution of water vapor in region I (the Tibetan Plateau) is from the Central Asian path (channel 3). It originates from the dry and cold air mass over the Eurasian continent with a water vapor contribution of 41.5% at the initial location, which increases to 47.6% after transportation. Channels 4 and 5 originate from warm and wet air masses, with initial contributions of 11.7 and 23.3%, respectively. Water vapor is lost during transportation and decreases to 9.4 and 19.6%, respectively.
The Indian Ocean is the most important source of moisture for region II. The contribution of water vapor from channel 4 is 42.9% at the initial location, and the final contribution is slightly decreased to 42.2%. The second largest source is channel 5 from the Bay of Bengal, with initial and final contributions of 27.2 and 26%, respectively. The contributions show no obvious change between the initial and final positions in the local path (channel 3), which makes a contribution of 20%. The largest contribution of water vapor in region III also comes from the Indian Ocean (channel 3) and is 38.6 and 40.8% at the initial and final locations, respectively. The second largest water vapor source is the South China Sea (channel 4), with a contribution of 27.2% at the final position. The western Pacific Ocean is an important source of moisture for region III, with a contribution of 8%, but has a negligible contribution to regions I and II.
5. Discussion and Conclusions
Continuous precipitation systems move eastward from the Tibetan Plateau to the middle and lower reaches of the Yangtze River under Mei-yu front rainstorm conditions. Based on station observations and reanalysis data, we selected 20 typical Mei-yu front precipitation events during the time period 2010–2015 and analyzed the characteristics of their circulation. We used the Lagrangian trajectory model HYSPLIT_4.9 to simulate the movement of air parcels in Mei-yu front precipitation events along the Tibetan Plateau to the middle and lower reaches of the Yangtze River. We then determined the quantitative contribution of different water vapor transport paths and the distribution of sources.
The Mei-yu front precipitation affected a wide area, reaching thousands of kilometers across the Tibetan Plateau to the middle and lower reaches of the Yangtze River. The amount of precipitation gradually increased eastward from the Tibetan Plateau. The center of maximum precipitation on the Tibetan Plateau is located in the Shannan area to the west of Nyingchi and reached 25 mm/day. The center of maximum precipitation in the region from the eastern Tibetan Plateau to the Sichuan Basin is located in the western part of the basin and had a local maximum of 50 mm/day, which is the result of the torrential rains that may be caused by the eastward shift in the plateau shear line [19, 25]. The intensity of precipitation reaches 50 mm/day in most areas along the Yangtze River, and precipitation in southern Anhui and southern Jiangsu in the middle and lower reaches of the Yangtze River exceeds 100 mm/day. There is a large area of precipitation in the northeastern Sichuan Basin. Fu et al. showed that the low vortex formed in the southwest of the Sichuan Basin can cause a series of rainstorms when it moves eastward along the Mei-yu front .
The synthetic analysis of the atmospheric circulation of Mei-yu front precipitation shows that the South Asian high and the high-level westerly jet have a large range and strong intensity. The western Pacific subtropical high exerts a positive influence on precipitation because the 5880-gpm contour extends to 115°E. The locations of the southwesterly jet under the subtropical high and the Mei-yu front are coincident. The area of heavy rainfall is located to the right of the area of divergence of the high-level jet and the convergence zone of the low-level jet, where the strong vertical upward flow provides the lifting conditions required for heavy rainfall.
The sources of water vapor for Mei-yu front precipitation are distributed in the Indian Ocean, the Bay of Bengal, the South China Sea, and the western Pacific Ocean. The largest contribution to rainfall over the Tibetan Plateau is from the North Atlantic Ocean and Central Asia; air parcel trajectories are 52% of the total trajectories, but water vapor only accounts for 41.5% (47.6%) of the total at the initial (final) location because the air mass is dry and cold. The most important source of water vapor from the eastern Tibetan Plateau to the Sichuan Basin is the Indian Ocean, which contributes 40% (air parcel trajectory), 42.9% (initial water vapor), and 42.2% (final water vapor) of the total. The warm and humid Indian Ocean makes the largest contribution to the middle and lower reaches of the Yangtze River, contributing 37% of the air parcel trajectories and 38.6% of the initial water vapor contribution. After supplementation and dissipation during long-distance transportation, the final water vapor contribution is 40.8%.
The data used to support the findings of this study may be available from the corresponding author upon request.
Conflicts of Interest
The authors declare that there are no conflicts of interest regarding the publication of this paper.
The authors acknowledge the NOAA/Air Resources Laboratory for providing the HYSPLIT model and ECMWF for providing atmospheric reanalysis data. This work was supported by the National Natural Science Foundation of China (grant nos. 91637211 and 41620104009), the National Key Research and Development Program of China (grant no. 2018YFC1507202), and the Central Asia Atmospheric Science Research Foundation (CAAS201803).
X. Zhang, S. Tao, and S. Zhang, “Three types of heavy rainstorms associated with the meiyu front,” Chinese Journal of Atmospheric Sciences, vol. 28, no. 2, pp. 187–205, 2004, in Chinese.View at: Google Scholar
K. Higashi, Y. Kiyohara, M. D. Yamanaka, Y. Shibagaki, M. Kusuda, and T. Fujii, “Multiscale features of line-shaped precipitation system generation in Central Japan during late baiu season,” Journal of the Meteorological Society of Japan, vol. 88, no. 6, pp. 909–930, 2010.View at: Publisher Site | Google Scholar
Y. Luo and Y. Chen, “Investigation of the predictability and physical mechanisms of an extreme-rainfall-producing mesoscale convective system along the meiyu front in East China: an ensemble approach,” Journal of Geophysical Research: Atmospheres, vol. 120, no. 20, pp. 10593–10610, 2015.View at: Publisher Site | Google Scholar
D. Y. Qin, J. X. Jiang, and Z. Y. Fang, “The characteristics of water vapor plume in the heavy rain events during 21-24 June 2002,” Journal of Meteorological Research, vol. 62, no. 3, pp. 329–337, 2004.View at: Google Scholar
H. Yang, Z. Jiang, and Z. Liu, “Analysis of climatic characteristics of water vapor transport based on the Lagrangian method: a comparison between meiyu in the Yangtze-Huaihe River region and the North Huaihe River rainy season,” Chinese Journal of Atmospheric Sciences, vol. 38, no. 5, pp. 965–973, 2014, in Chinese.View at: Google Scholar
Q. Moteki, H. Uyeda, T. Maesaka, T. Shinoda, M. Yoshizaki, and T. Kato, “Structure and development of two merged rainbands observed over the east China sea during X-BAIU-99 Part I: meso-.BETA.-scale structure and development processes,” Journal of the Meteorological Society of Japan, vol. 82, no. 1, pp. 19–44, 2004.View at: Publisher Site | Google Scholar
X. Cui, S. Gao, H. Zhang, and S. Hao, “A diagnostic analysis of the simulated structure of a meiyu front system in 1999,” Journal of Meterological Research, vol. 23, no. 1, pp. 43–52, 2008.View at: Google Scholar
R. R. Draxler and G. D. Hess, “An overview of HYSPLIT_4 modeling system for trajectories dispersion and deposition,” Australian Meteorological Magazine, vol. 47, no. 1, pp. 295–308, 1998.View at: Google Scholar
Y. Luo, Y. Wang, H. Wang, Y. Zheng, and H. Morrison, “Modeling convective-stratiform precipitation processes on a mei-yu front with the weather research and forecasting model: comparison with observations and sensitivity to cloud microphysics parameterizations,” Journal of Geophysical Research, vol. 115, no. D18, 2010.View at: Publisher Site | Google Scholar
P. Berrisford, D. Dee, P. Poli et al., “The ERA-interim archive version 2.0,” ECMWF, Reading, UK, 2011, Technical 349 Report.View at: Google Scholar
Q. Zhu, J. Lin, S. Shou et al., Principle and Approach of the Synoptic, China Meteorological Press, Beijing, China, 2000, in Chinese.
A. Drumond, R. Nieto, and L. Gimeno, “On the contribution of the tropical western hemisphere warn pool source of moisture to the Northern Hemisphere precipitation through a Lagrangian approach,” Journal of Geophysical Research: Atmospheres, vol. 116, no. D21, 2011.View at: Publisher Site | Google Scholar
S. R. Dorling, T. D. Davies, and C. E. Pierce, “Cluster analysis: a technique for estimating the synoptic meteorological controls on air and precipitation chemistry-method and applications,” Atmospheric Environment. Part A. General Topics, vol. 26, no. 14, pp. 2575–2581, 1992.View at: Publisher Site | Google Scholar