The Upstream "Strong Signals" of the Water Vapor Transport over the Tibetan Plateau during a Heavy Rainfall Event in the Yangtze River Basin

The Tibetan Plateau (TP) is the highest plateau in the world with the most complex terrain. Its average elevation is about 4000 m above mean sea level. It makes up about 1/4 of China's and 1/6 of Asia's total land. (Tao, 1980) pointed out that China and some other countries that are influenced by monsoon systems are prone to flooding in the summer. (Tao, 1987) also evidenced that the TP, located in the north of the Indian monsoon area and west of the East Asia monsoon area, plays a key role in the modulation of weather and climate of China and East Asia, even in global climate. The impact of the TP's terrain on regional and global atmospheric circulation and hydrological cycles, particularly the Asian monsoon, remains a hot topics of debate within the scientific community (Wu and Zhang, 1998; Boos and Kuang, 2010).

In particular, the relationship between heavy rainfall in the Yangtze River Basin (YRB) and the dynamic and thermal effects of the TP upstream has long been a key focus for meteorologists and atmospheric scientists (e.g., Yanai et al., 1992; Wu and Zhang, 1998; Xu et al., 2008, 2013; Chen et al., 2013). A number of previous studies have shown that the dynamic effect of the TP has crucial influences on the downstream weather and climate over the Yangtze-Huai drainage areas (Chen et al., 1985; Yanai et al., 1992; Chen et al., 2012). (Chow et al., 2008) pointed out that the water vapor accounting for early summer rainfall in China is mainly transported via the southwest of China from the Indian monsoon. They suggested that a weak Indian monsoon will bring less rainfall in China, whereas a stronger Indian monsoon will cause flooding. (Xu et al., 2008) proposed a "world water tower" model for the special atmospheric circulation and water vapor cycle over the TP. During the heavy rainfall events over the YRB in 1998, the frequent development of convective clouds could be observed over the central and eastern TP. These convective clouds moved successively eastward in clusters. According to field experiments on the TP, it was found that the mesoscale convective clouds in the east of the TP moved eastward along the YRB (about 30^°N). Satellite data analysis also suggested that the immediate cause of the floods in 1998 was low-level lows or shear lines, and some of these low-level lows could be tracked back to the TP.

The weather systems related to the rainfall in the YRB include regional lows, tropical cyclones and frontal cyclones. The development of these systems involves multi-scale interactions under large-scale circulation conditions. However, the cause of heavy rainfall in East China, especially the water vapor transport and its corresponding circulation structure, has a close relationship with the lows over the TP. For example, (Li, 2012) demonstrated that both the plateau low vortex and the northwest low vortex can give rise to heavy rainfall in East China. The eastward movement of the plateau low vortex takes place mainly in July, while it is in May for the northwest low vortex. And there were more eastward moving low vortexes from the plateau in 1998, 1999 and 2003, resulting in severe flooding over the YRB. There is no doubt that the low vortex from the TP is one of the most important weather systems impacting upon summer rainfall in China. However, in view of the diverse explanations about the relationship between the heavy rainfall over the YRB and the dynamic and thermal effects of the TP, as well the debate in terms of how to connect the upstream dynamical and hydrological "strong signals" and the downstream heavy rainfall, we hope the results of this study will shed some light on this controversial topic.

The meteorological field data from the NCEP Final Operational Global Analysis dataset, with a resolution of 1^°× 1^° and 26 levels vertically extending from the surface to 10 hPa (see http://dss.ucar.edu/datasets/ds083.2/data/) are used in this study. We also use observed hourly precipitation data (July 2000) from 2400 gauge stations over China, which are from the National Meteorological Center of the China Meteorological Administration observation archives. The precipitation data are controlled for quality before being released. The low-level cloud-cover data from 753 observational stations over China for 1961-2010 are also adopted.

2.2.1. Stream function and velocity potential analysis

Considering the dynamic impacts of bypassing and cross-mountain flow associated with the TP's terrain on water vapor transport, it is convenient to use the velocity potential and stream function to represent the atmospheric flow fields. The formulas for velocity potential and stream function are: \begin{eqnarray} \label{eq1} &&\frac{\partial v}{\partial x}-\frac{\partial u}{\partial y}=\zeta ;\quad (1) \\ \label{eq2} &&\frac{\partial u}{\partial x}+\frac{\partial v}{\partial y}=D ;(2) \\ \label{eq3} &&\nabla^2\phi=\zeta ; (3) \\ \label{eq4} &&-\nabla^2\chi=D . (4)\end{eqnarray} The u component represents the east-west component of the horizontal wind (x-direction), while the v component represents the north-south component (y-direction). ζ and D represent horizontal vorticity and divergence, respectively. The symbol ∇² is the Laplacian operator. Here, we solve the Poisson equation, Eq. (4), to obtain lines with a constant value of the stream function φ. Similarly, we solve the Poisson equation, Eq. (5), to obtain lines with a constant value of velocity potential χ.

2.2.2.Correlation vector analysis

In order to investigate the relationship between the formation and development of the heavy rainfall and the upstream water vapor flux, the method of correlation vector analysis is adopted. The correlation vector is defined as \begin{equation} \label{eq5} {R}(x,y)={R}_u (x,y)+{R}_v(x,y) , (5)\end{equation} where R is the correlation vector between the rainfall and total water vapor flux; and R_u(x,y) and R_v(x,y) are the zonal and meridional components of the correlation vector, respectively. Equation (6) can also be easily derived with the potential or stream function.

(Hu and Ding, 2003) showed the structure of the inflow and outflow of water vapor during heavy rainfall in the Yangtze-Huaihe River basin. It appeared that both the inflow and outflow occur at the lower levels. Specifically, the inflow is in the south and west of the rainfall region, while the outflow is in the east and north. Xu et al. (2002, 2012) suggested that the TP plays two roles in water vapor transport. First, as a heating source, the plateau acts as an air pump that attracts low-latitude warm and moist air coming up towards the TP. Second, due to the high-rise nature of the plateau, it blocks and deflects a large amount of water vapor to the east. Therefore, in this sense, the TP plays the role of a water vapor "re-channel station". (Xu and Chen, 2006) also characterized the convective clouds moving successively eastward from the TP and pointed out that these low-level clouds are the precursors of the convective systems responsible for the heavy rainfall over the YRB. During the second TP Atmospheric Scientific Experiment (TIPEX), the analysis results with temperature black body satellite data and water vapor cloud images showed that the formation and development of convective systems could account for the heavy rainfall in Wuhan in late July.

Figure 1 shows the correlation coefficient distribution of summertime rainfall during July 1961-2010 over the YRB and the low-level cloud cover for the same time period. It is notable that the areas of significant positive correlation (with a confidence level of 95%, or correlation coefficient R>0.28) represent a typical banded area extending from the TP to the YRB. These results reflect the close relationship between the banded structure features of the local or eastward-moving convective cloud activity with heavy rainfall within this area. The results also further verify the conclusion obtained in TIPEX that the clouds responsible for the heavy rainfall of the YRB are from the TP. And a key fact that the intense convective clouds originate from the TP is also proven.

Figure 1.  Correlations between average rainfall (units: mm) in July and low-level cloud cover for the period 1961-2010. Color shading indicates the topography (units: m). The red and blue solid circles indicate stations exceeding the 90% confidence level (R>0.23 or R<-0.23).

Among all heavy rainfall zones in China, there are two main zones with frequent extreme heavy rainfall events. One is the middle and lower reaches of the Yangtze River, and the other is North China (Zou et al., 1987; Wang et al., 1991). Considering the dynamic and thermal effects of the TP on convective systems and heavy rainfall downstream, we select a heavy rainfall event that occurred in the Yangtze-Huaihe River Basin during 11-15 July 2000 for this study. The corresponding accumulated precipitation from 2000 LST 11 July to 2000 LST 15 July is shown in Fig. 2.

We choose this event because it is a highly typical heavy rainfall case. The distribution of daily rainfall from during 11-15 July 2000 presents an obvious banded pattern. The rainfall started from the TP, moved eastward, and finally reached the middle and lower reaches of the YRB; and an obvious banded pattern for the rainfall was formed (figure omitted). Figure 2 also shows this typical heavy rainfall case was characterized by two maximum rainfall centers. One was located at the junction of Sichuan, Gansu, Chongqing and Shanxi provinces, and the other at that of Hubei, Henan, Jiangsu and Anhui provinces. Both maximum centers had accumulated rainfall of more than 50 mm. Compared with Fig. 1, we can see that the accumulated rainfall distribution of this case in Fig. 2 resembles the pattern of correlation between the heavy rainfall over the YRB and low-level cloud cover upstream. This further emphasizes the important role of the moving cloud system in heavy rainfall downstream.

The rainfall in the major regions of the YRB fluctuated during the period, with an increasing trend (Fig. 3), and reached its peak at 2000 LST 13 July. Figure 3 also depicts the features of the rainfall variation during this period, especially the continuous heavy rainfall in the latter days of this period. But how is the rainfall related to the eastward-moving system and associated water vapor transport? In the following sections, we continue to examine the dynamic structure and characteristics of the water vapor transport associated with the heavy rainfall over the YRB.

Figure 2.  Accumulated rainfall (units: mm) of day 5 (from 2000 LST 11 July to 2000 LST 15 July). The blue solid outline represents the key rainfall area selected in this study (29^°-35^°N, 104^°-122^°E).

Figure 3.  Time series of average rainfall (units: mm) in the key rainfall area (29^°-35^°N, 104^°-122^°E) from 2000 LST 5 July to 2000 LST 15 July (6-h interval). The red double-headed arrow indicates the heavy rainfall process, as shown in Fig. 2.

The dominant flows associated with the Indian monsoon and South China Sea (SCS) monsoon create the major water vapor transport channel during summer (Zhou et al., 2005). (Liu and Ding, 2009) suggested that the monsoon flow from the Indian Ocean and the Pacific have significant impacts on summer rainfall over China. The water vapor source for the rainfall in East China is mainly the SCS, followed by the Bay of Bengal (BOB) and West Pacific (Jin, 1981; Shen and Huang, 1981; Chen, 1982). (Xu and Chen, 2006) further revealed that the "big-triangle" area surrounding the TP is the key region of water vapor transport resulting in the rainfall in the middle and lower reaches of YRB. The water vapor channels in the south and east of the TP play a critical role in the rainfall of the YRB. By analyzing the heavy rainfall over the YRB in 1998, (Jian and Luo, 2001) argued that the variation of the strength of the meridional monsoon flow follows the diurnal cycle of the upward motion of the air over the TP. The upward motion is stronger at night, which indicates the monsoon meridional flow is stronger and more water vapor is transported.

Knowing that a high correlation exists between the rainfall over the YRB and low-level convective activities, we further examine the structure of water vapor transport using correlation vector analysis. In order to analyze the evolution of the water vapor transport prior to and after this event, lagged correlations are calculated from 2000 LST 5 July to 2000 LST 15 July (with a temporal resolution of 6 hours). Figure 4a shows the lagged correlation vectors between the rainfall in the key regions of the YRB and the total water vapor flux two days prior to the rainfall. It clearly shows the confluence of two water vapor channels from the SCS and the BOB, respectively. The water vapor transport comes from the BOB, turns to the YRB after passing the southeast corner of the TP, which approximately coincides with the maximum rainfall on the same day. Figure 4b shows the lagged correlation vectors one day prior to the rainfall. Compared with Fig. 4a, we can see that the major rainband moves eastward, and the eastern edge of the rainband even extends to the middle and lower reaches of the Yangtze River. The simultaneous correlation vectors indicate that the center of maximum rainfall was located in the middle and lower reaches of the YRB at that time; the water vapor was transported northward from the BOB and turned to the middle and lower reaches of the Yangtze River along the northeastern edge of the TP. The water vapor starting from the SCS continuously transported the moisture to the YRB. The lagged and simultaneous correlation analysis clearly show the convergence/confluence zone of the water vapor transport, which coincides with the eastward migration of the maximum rainfall.

Figure 4.  The (a) two-day lagged, (b) one-day lagged and (c) simultaneous correlations between the total water vapor flux (units: g m^-1 s^-1) and the rainfall in the key region, represented as correlation vectors (arrows). Color-shaded areas on land represent the corresponding daily rainfall (units: mm) measured at 2000 LST (a) 11 July, (b) 12 July and (c) 13 July 2000. The blue solid lines denote the confidence level of ≥99%. The red circle indicates the precipitation area.

From the perspective of eddy transport, (Gao and Zhai, 1993) demonstrated that the eddy transport of water vapor is in the same direction as the gradient of the water vapor content. (Guan et al., 2011) suggested that extreme rainfall in the middle and lower reaches of the Yangtze River is related to anomalous local circulation systems. When extreme rainfall occurs, there is usually a cyclonic circulation in the middle-lower troposphere over the middle and lower reaches of the Yangtze River and an anticyclonic circulation to the south. We calculate the two-day and one-day lagged correlation, as well as the simultaneous correlation, between the rainfall in the major regions and the velocity potential of water vapor flux, and obtained the corresponding water vapor flux structure (Fig. 5). The two-day lagged correlation map (Fig. 5a) shows a clear water vapor convergence area in the central and eastern TP (Zone A) at 500 hPa, which coincides with the east-west orientated rainband that starts from the central and eastern TP and extends to the YRB. Figure 5b shows that, for the one-day lagged correlation, the area with water vapor convergence in Fig. 5a has moved to the upper and middle reaches of the Yangtze River (Zone B), and coincides with the heavy rainfall in Hubei Province. Figure 5c shows the simultaneous correlation, from which it can be seen that the area with water vapor flux convergence has moved to the middle and lower reaches of the Yangtze River. The heavy rainfall area covers the whole of the middle and lower reaches of the Yangtze River and expands to Anhui, south of Henan and Jiangsu provinces. By analyzing the structure of water vapor transport during and prior to the heavy rainfall (Figs. 5a-c),we find there is also an eastward-migrated water vapor convergence region corresponding to the eastward migration of the heavy rainfall.

Figure 5.  As in Fig. 4, but for the correlations between the rainfall (units: mm) and the potential function of water vapor flux (units: g cm^-1 s^-1 hPa^-1) at 500 hPa. The blue solid lines denote the confidence level of ≥90%. The red circle indicates the precipitation.

The correlations between the rainfall and velocity potential of water vapor flux at 200 hPa and 500 hPa two days prior to and after the rainfall are also compared, to further examine the relationship between them (Fig. 6). The difference between the two-day and one-day lagged correlations shows a clear divergence center of velocity potential of water vapor flux at 200 hPa (Fig. 6a), and a convergence center at 500 hPa (Fig. 6b), in the central and eastern TP. This "lower-level convergence and upper-level divergence" coupled system moved eastward and reached the middle and lower reaches of the Yangtze River, which can be easily identified from the difference between the one-day lagged and simultaneous correlations (Figs. 6c and d).

The dynamic and thermal effects of the TP usually result in the formation of a low pressure system over the TP. These low pressure systems, called Plateau lows, evolve into severe weather systems and deliver abundant rainfall in downstream regions. (Yu, 2008) demonstrated the tremendous impacts of the eastward-moving plateau vortexes on the rainfall over the Yangtze River and reaches of the Yellow River-Huaihe River. (Yang et al., 2001) demonstrated that there was a close relationship between the eastward-moving plateau vortexes and the extreme heavy rainfall over the YRB in 1998.

Given the "lower level convergence-upper level divergence" coupled structure of the eastward-moving plateau vortex, we calculate the correlations between the rainfall in the major regions of the YRB and the component of the velocity potential function.

Cross-section maps (along 33^°N) of correlation vectors and vertical speed are plotted and shown in Figs. 7a-d. According to the cross section of the correlation vectors of the potential function, the area with upward motion (rectangular region in Fig. 7a) was initially located in the central and eastern TP (98^°-105^°E) 48 hours prior to the rainfall in the major regions of the YRB, and the upward movement center was located in the same area. At 36 hours prior to the rainfall, that area with significant upward motion had moved to somewhere in the east of the TP (rectangular region in Fig. 7b). The correlation between the rainfall and vertical speed field shows a positive center in this area between 500 hPa and 250 hPa. At 24 hours prior to the rainfall, the upward motion area had partially moved out of the TP to cover an area of the plains. At 12 hours prior to the rainfall, the upward motion had reached the middle and lower reaches of the Yangtze River. The results shown in Fig. 7 may verify that the accompanying significant upward motion was the dynamical cause of the heavy rainfall in the middle and lower reaches of the Yangtze River. The above cross-section maps of the velocity potential clearly depict the upward motion corresponding to the "low level convergence-upper level divergence" coupled structure, as well as the dynamic mechanism of generation, development and eastward movement of the convective systems.

Figure 6.  Differences in the correlations between the rainfall (units: mm) in the key rainfall region of the YRB and the potential function of water vapor flux (units: g cm^-1 s^-1 hPa^-1) from 2000 LST 5 July to 2000 LST 15 July 2000. Differences in the correlations between the two-day and one-day advanced correlations at (a) 200 hPa and (c) 500 hPa, and between the one-day advanced and simultaneous correlations at (b) 200 hPa and (d) 500 hPa. The blue lines show the difference field of the correlation between two consecutive days. The solid (dashed) lines indicate positive (negative) correlation. The red circle indicates the precipitation area.

Figure 7.  Correlations between the rainfall (units: mm) in the key rainfall region of the YRB and the components (u, -ω) of the potential function, as well as the cross-section maps of correlations with vertical movement (-ω) (along 33^°N). Panels (a-d) are the cross-section maps of 48-hour, 36-hour, 24-hour and 12-hour advanced correlations, respectively. The blue solid rectangle represents the eastern rainfall region focused upon in this study.

In this study, by analyzing the three-dimensional structure of the water vapor transport in the TP prior to a widespread heavy rainfall event over the YRB, we identified the "strong signals" of the dynamical and hydrological features of the synoptic systems in the TP. The results obtained in the study may shed some light on understanding the relationship between the heavy rainfall over the YRB and the water vapor transport and eastward-moving systems. The following are the main conclusions:

(1) The convective clouds over the TP play a key role in the occurrence and development of heavy rainfall over the YRB. The results from this study reveal a close relationship between the heavy rainfall over the YRB and the eastward-moving low-level clouds from the TP. The pattern of the low-level cloud cover coincides with the distribution of the accumulated rainfall over the YRB, which also confirms its role.

(2) Lagged correlation analysis shows the existence of a "convergence" area of water vapor transport and its eastward movement coincides with the corresponding maximum rainfall over the YRB.

(3) A "lower-upper level" coupled structure, represented by the velocity potential of water vapor flux, originating from the TP, also plays a key role in rainfall over the YRB. This specific structural configuration is critical in leading to the occurrence of heavy rainfall downstream.

(4) The structure of the velocity potential/stream function and the eastward migration of the strong upward motion area is also one of the upstream "strong signals" of heavy rainfall. The three-dimensional water vapor transport (flux, vorticity and divergence) prior to the occurrence of a heavy rainfall event in the middle and lower reaches of the Yangtze River.

(5) By adopting the method of three-dimensional correlation analysis between the heavy rainfall and the velocity potential/stream function of water vapor flux, it is possible to clearly reproduce the scenario that the eastward movement of water vapor transport eventually results in heavy rainfall in the middle and lower reaches of the Yangtze River. The analytical method of the velocity potential/stream function used in this paper could be adapted to a trace analysis of the generation, development and eastward movement of other heavy rainfall or mesoscale convective systems.