Characteristics of Summer Convective Systems Initiated Over the Tibetan Plateau

OCTOBER 2008 YAODONG ET AL. 2679

Characteristics of Summer Convective Systems Initiated over the Tibetan Plateau. Part I: Origin, Track, Development, and Precipitation

LI YAODONG AND WANG YUN Institute of Atmospheric Physics, Chinese Academy of Sciences, and Beijing Aviation Meteorological Institute, Beijing, China

SONG YANG* College of Science, George Mason University, Fairfax, Virginia

HU LIANG AND GAO SHOUTING Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, China

RONG FU Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, China, and School of Earth and Atmospheric Sciences, Georgia Institute of Technology, Atlanta, Georgia

(Manuscript received 23 January 2007, in final form 10 February 2008)

ABSTRACT

Summer convective systems (CSs) initiated over the Tibetan Plateau identified by the International Satellite Cloud Climatology Project (ISCCP) deep convection database and associated Tropical Rainfall Measuring Mission (TRMM) precipitation for 1998–2001 have been analyzed for their basic characteristics in terms of initiation, distribution, trajectory, development, life cycle, convective intensity, and precipitation. Summer convective systems have a dominant center over the Hengduan Mountain and a secondary center over the Yaluzangbu River Valley. Precipitation associated with these CSs contributes more than 60% of total precipitation over the central-eastern area of the Tibetan Plateau and 30%–40% over the adjacent region to its southeast. The average CS life cycle is about 36 h; 85% of CSs disappear within 60 h of their initiation. About 50% of CSs do not move out of the Tibetan region, with the remainder split into eastward- and southward-moving components. These CSs moving out the Tibetan Plateau are generally larger, have longer life spans, and produce more rainfall than those staying inside the region. Convective system occurrences and associated rainfall present robust diurnal variations. The midafternoon maximum of CS initiation and associated rainfall over the plateau is mainly induced by solar heating linked to the unique Tibetan geography. The delayed afternoon–late night peak of rainfall from CSs propagating out of this region is a combined outcome of multiple mechanisms working together. Results suggest that interactions of summer Tibetan CSs with the orientation of the unique Tibetan geography and the surrounding atmospheric circulations are important for the development, intensification, propagation, and life span of these CSs.

1. Introduction (Reiter 1982), in terms of boundary layer, physical processes of the surface–air interactions, heating sources, The effect of the Tibetan Plateau on the atmospheric meteorological elements, synoptic weather systems, systems has been studied in detail for several decades general atmospheric circulations, and climate variations. Various methods have been utilized in the pro- * Current affiliation: I. M. Systems Group, and NOAA/ cesses, such as boundary layer observational analysis NESDIS/Center for Satellite Applications and Research (STAR), (Yanai and Li 1994; Gao et al. 2002), objective analyses Camp Springs, Maryland. of synoptic weather and general atmospheric circulation (Ye 1981; Yanai et al. 1992), statistical study of Corresponding author address: Dr. Song Yang, NASA GSFC, synoptic weather systems (Tao and Ding 1981; Leber et Code 613.1, Greenbelt, MD 20771. al. 1995), analysis of satellite measurements (Jin 1997; E-mail: [email protected].nasa.gov Ueno 1998), diagnostic studies of meteorological vari-

DOI: 10.1175/2008JAMC1695.1

JAMC1695 2680 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 47 ables and atmospheric energy (Dell’Osso and Chen by the horizontally inhomogeneous heating distribution 1986; Reiter 1987), numerical simulations (Shen et al. in the north of the heating source with westerly exis- 1986b), fluid dynamic experiments (Boyer and Chen tence. Wu and Zhang (1998) also suggest that the onset 1987), and theoretic studies of dynamics (Xie 1981; Le- of the Asian summer monsoon mainly depends on the roux 1993). The ultimate goal of these investigations is Tibet heating source. Furthermore, results based on nu- to describe the processes of thermodynamics and dy- merical model simulations by He et al. (1984), Wang et namics of the Tibetan Plateau, and to understand their al. (1984), and Qian et al. (1988) show that there would impact on global climate change. Results could lead to be no Tibetan high pressure center, no breakup of the a new conceptual model and a prediction theory for west Pacific subtropical high, and no tube of monsoon severe weather systems influenced by the plateau, with circulation without the Tibetan heating source. In ad- further understanding of the environmental conditions dition, there would be no convergence zone in eastern of the Tibetan synoptic systems and climate, and the Tibet during the time period of mei-yu. Therefore, the characteristics of structure and movement of Tibetan physical processes and atmospheric circulation of the synoptic systems as well as the Tibetan heating and plateau have important impacts on global climate, topographical forcing. Asian atmospheric circulation, abnormal weather phe- It is well known that the Tibetan Plateau is a source nomena, and the climatology of severe weather. of dynamic and thermodynamic turbulence (Zhao and Literature results have shown that strong convective Chen 2000). Many severe weather systems that im- weather systems (such as heavy rainfall, ice storms, tor- pacted China in the past are linked to the dynamic and nadoes, wind gusts, and downdraft flows) are mostly thermodynamic influences of the plateau (Tao and associated with mesoscale convective systems (MCSs) Ding 1981; Reiter 1982; Shen et al. 1986a,b). Tao et al. (McCollum et al. 1995; Gray 2001). Because of its (1980) indicate that heavy rainfall in China is mainly unique thermodynamic forcing, the Tibetan Plateau is from typhoon, midlatitude frontal systems, and cyclonic an active region of strong summer CSs (Shen et al. vortices that propagate eastward from Tibet. Jiang et al. 1986a,b; Wang et al. 1993). Flohn (1968) demonstrates (1996) demonstrate that the eastward propagation of 20–50 well-developed cumulus clouds per 10, 000 km2 convective cloud systems over the Tibet Plateau could over the region, which indicates frequent summer con- trigger heavy precipitation over the Yangtze River ba- vective activity. Fu et al. (2006) show that these strong sin. For example, the persistent extreme heavy rainfall convective systems play a central role in transporting and flooding over the Yangtze and Huihe River basins water vapor to the global stratosphere during summer in 1991 and the Yangtze River basin in 1998 were primarily caused by the cyclonic vortices that originated seasons. Thus, the characterization of these CSs is also from Tibet (Fang 1985; Li et al. 1989; Xiang and Jiang important in understanding the global stratosphere 1995; Shi et al. 2000; Jiang and Fan 2002). The surface composition and climate. heating over the Tibetan Plateau is an elevated heating Jiang and Fan (2002) indicate two maximum convec- source in the midtroposphere (Luo and Yanai 1983, tion centers separated around 95°E over the region 1984; Yanai et al. 1992; Yanai and Li 1994). Wu et al. (i.e., the southeast center and the relatively stronger (1997, 2002a,b), Wu and Liu (2000, 2003), and Liu et al. south center). Only a small portion of these CSs propa- (2001, 2002) describe in detail the characteristics of the gates eastward, influencing precipitation over the summer atmosphere general circulation forced by the Yangtze River basin. Although only a few eastward Tibetan heating, and propose a mechanism to explain CSs intensify as they propagate, they often trigger how the atmospheric circulation is influenced by the heavy rainfall events over the middle low segments of Tibetan heating based on the theory of thermodynamic the Yangtze River basin. (The eastward propagation of geostrophic adjustment. This mechanism is defined as Tibetan CSs or MCSs is for a system moving east of the the “sensible heat driven air pump.” They demonstrate Tibetan Plateau during its life cycle.) Zhuo et al. (2002) that positive vorticity would be generated below the point out that these CSs tend to propagate eastward altitude of the Tibetan heating source, where heating toward the mid-low segments of the Yangtze River ba- intensity increases with height. A negative vorticity sin, where the occurrence of heavy rainfall events were would be produced above the heating source, where the generally coincident with the arrivals of these CSs. The heating rate decreases with height. The low-level posi- eastward propagation of a CS appears to be in favor of tive vorticity will balance the upper-level negative vor- following large-scale dynamic circulation patterns: 1) ticity and the transport of negative vorticity from the the fields of isobar height, divergence, and vorticity at Tibetan lateral boundaries, and depletion from surface 400 hPa show a south–north orientation pattern; and 2) friction. A deep anticyclone vortex would be induced the atmospheric thermo-instability index and isobar

Unauthenticated | Downloaded 09/30/21 10:48 PM UTC OCTOBER 2008 YAODONG ET AL. 2681 height at 500 hPa has a west–east orientation distribu- (Rossow et al. 1996; Rossow and Schiffer 1999). The tion (Guo et al. 2003a,b, 2004; Fang et al. 2004a,b). DX dataset has 3-h temporal resolution and associated However, most previous investigations of the Ti- pixels at 30 km ϫ 30 km spatial resolution. Second, betan CS or MCS were case studies, focusing on the analyzing the DX database leads to CTD. This proce- vicinity of the Plateau. The characteristics of the gen- dure includes two steps (i.e., identification of cloud eral atmospheric circulations during CS propagation clusters and tracking analysis). The identification of and development, and the relationship between propa- cloud clusters from satellite images follows the method gation and environmental variables, remain unclear. proposed by Machado and Rossow (1993); that is, a CS Therefore, the basic characteristics of these CSs and is defined as adjacent cloud clusters where TB Ͻ 245°K, their propagation, development, life cycles, and impacts while a convective cluster (CC) is determined as adja- on surface rainfall are the main objectives of this study. cent cloud clusters where TB Ͻ 220°K. Thus, a CC is a We focus on the mean properties of CSs initiated over relatively stronger convective cell than a CS. These pro- the Tibetan Plateau in terms of intensity, size, propa- cedures are consistent with the published literature gation, and precipitation, by tracking all summer CSs (Clark 1983; Maddox 1980; Machado et al. 1998). In and associated surface rainfall from satellite measure- general, there are several CCs inside the coverage of a ments in 1998–2001. CS. The tracking analysis based on the method by Machado et al. (1998) has two substeps: 1) determination of the time interval between two continuous im- 2. Methodology and datasets ages; and 2) determination of the overlap area of a convective system in two continuous images by calcu- a. Datasets lating the ratio between all pixels in the overlapped There are only a few standard meteorological sta- region of two convective systems and the total pixels of tions over the Tibetan Plateau because of its harsh cli- these two convective systems. This process is conducted mate and unique geography features, and high-quality for all possible pairs of overlapped convective systems. traditional meteorological datasets with a long time se- The pair with the largest ratio is selected as the con- ries and large areal coverage are lacking for this region. vective system to be tracked. Two more tests are per- In addition, ground radar remote sensing cannot pro- formed to assess the quality of the tracking analysis. In vide good measurements over the remote mountain case of missing satellite images, the CS tracking contin- area. However, satellite remote sensing can overcome ues if the time separation is less than the time criterion those issues associated with the traditional observation estimated with the CS’s size (Machado et al. 1998). The techniques. The International Satellite Cloud Climatol- CS for this time step is then considered as missing so ogy Project (ISCCP) Convection Tracking Database that it would not be counted in the statistical analysis. (CTD) is utilized to investigate the summer CSs origi- Because the situation of missing images happens only nating from the Tibetan region. Since Tropical Rainfall occasionally, it would not significantly impact the re- Measuring Mission (TRMM) precipitation products are sults of this study. The CTD includes all convective to date one of the best rainfall datasets (Kummerow et systems with a minimum radius of 90 km (correspond- al. 2001; Yang et al. 2006a), the TRMM blended rainfall ing to a minimum of 30 pixels in the DX dataset) over (3B42) and the precipitation radar (PR)–based rainfall the global 70°S–70°N belt. The datasets are available at (2A25) rain products are used in this study (Iguchi et al. 0000, 0300, 0600, 0900, 1200, . . . , 2100 UTC. Detailed 2000; Meneghini et al. 2001; Huffman et al. 2001, 2007). documentation on the ISCCP dataset can be found on- CTD has 41 parameters related to the macrocharac- line (http://isccp.giss.nasa.gov/). teristics and cloud physics of CSs during their life The initial observations contributing to the CTD cycles. It systematically provides descriptions of CSs in dataset come from five geostationary satellites: the U.S. terms of propagation, track, speed, intensity, location of Geostationary Operational Environmental Satellite- heavy rainfall, internal cloud properties, and size. CTD (GOES) East (GOE), GOES-West (GOW), Japanese is constructed using the following procedures (Rossow Geostationary Meteorological Satellite (GMS), Europe et al. 1996): first, clear or cloudy pixels are identified Meteorological Satellite (MET), and India Satellite using the infrared (IR) and visible (VIS) observations. (INS). The VIS data are only applied during daytime. If The TRMM 3B42 rain product based on rain retriev- cloudy, the cloud-top brightness temperature (TB) and als from multiple satellite passive microwave and IR cloud optical depth (daytime only) are obtained. This measurements is available at 0000, 0300, 0600, information is used to build up the ISCCP DX database 0900, . . . , 2100 UTC at 0.25° ϫ 0.25° grid resolution

Unauthenticated | Downloaded 09/30/21 10:48 PM UTC 2682 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 47

FIG. 1. The selected large study area and the Tibetan region (dashed box). The dotted contours show the topography at 3000 and 5000 m.

(Huffman et al. 2001, 2007). This product ultimately areas and the evaporation effects associated with sur- utilizes high-accuracy rain estimates from satellite pas- face rain gauges. sive microwave observations [from TRMM, Special TRMM PR–based rain retrievals (Iguchi et al. 2000), Sensor Microwave Imager (SSM/I), Advanced Micro- designated as 2A25 in the TRMM data system, are ap- wave Scanning Radiometer for Earth Observing Sys- plied as alternative rain products to validate the per- tem (AMSR-E), and Advanced Microwave Sounding formance of 3B42 rain products. Wolff et al. (2005) Unit (AMSU-B)] and high temporal resolution from demonstrate that 2A25 rain rates are in good agree- satellite IR measurements. Any valid rain rates from ment with ground radar measurements at TRMM available passive microwave measurements within 90 ground validation sites. Yang et al. (2006a, 2008) and min of a designated data point are selected as the rain Yang and Smith (2008) show that 2A25 rain products rate for this point. If no passive microwave–based rain are very comparable with TRMM microwave and mi- rate is available, the rain rate from IR measurements crowave–precipitation radar combined rain products. adjusted with passive microwave–based rain rates is However, the sampling issue associated with TRMM used for this data point. The 3B42 monthly rain esti- PR is obvious for studying small spatial systems be- mates are finally adjusted by surface rain gauge mea- cause of its narrow swath. Both 3B42 and 2A25 rain surements. Published results demonstrate that the 3B42 estimates have relatively large errors at instantaneous rain products are highly comparable to any other sat- and pixel resolution. This error decreases with the in- ellite remotely sensed, surface gauge, and radar rain crease of spatial and temporal resolution. The version-6 products (Huffman et al. 2007; E. A. Smith et al. 2008, TRMM rain products have an error of about 20% (3%) unpublished manuscript). However, it was noted that at the 14 mm hϪ1 rain rate for the instantaneous 0.5° ϫ 3B42 rain locations could shift a little bit at storm edge 0.5° (monthly 2.5° ϫ 2.5°) grid scale (Olson et al. 2006; because of the IR rain estimates. In addition, E. A. Yang et al. 2006a). For June–August 1998–2001, ISCCP Smith et al. (2008, unpublished manuscript) indicate CTD, 3B42, and 2A25 rain datasets are used for this that the surface gauge adjustments might lead to the study. underestimation of 3B42 rainfall over the west U.S. The rectangular area (25°–40°N, 75°–105°E) is de- mountain regions during the warm season because of fined as the Tibetan domain for this study (Fig. 1). Any less reliable surface rain gauges over these mountain CS formed inside this domain is referred to as a Tibetan

Unauthenticated | Downloaded 09/30/21 10:48 PM UTC OCTOBER 2008 YAODONG ET AL. 2683

FIG. 2. (a) Spatial distribution on a 1.5° ϫ 1.5° grid of summer CS occurrences originating from the Tibetan region and (b) contributions of precipitation from CS generation in Tibetan region to total rainfall on a 0.25° ϫ 0.25° grid. The dashed box is the selected Tibetan region with dotted contours for 3000 and 5000 m of the topography.

CS. A larger area (Equator–40°N, 70°–140°E) is also the region. The distribution sheds light on the propa- used for studying the life cycle and propagation of sum- gation of these CSs and their impacts on surface rain- mer Tibetan CSs. fall, which is particularly useful in rainfall prediction over the Tibetan and surrounding regions. b. Methodology Figure 2a shows the spatial distribution of summer 1998–2001 Tibetan CS occurrences at 1.5° ϫ 1.5° grid Because of the different spatial resolutions of ISCCP resolution identified with CTD. It is evident that the CTD variables and TRMM 3B42 rainfall, they must be CSs originating from the Tibetan region moved over reconstructed to have consistent analyses of rainfall and east China, South Asia, and the Bay of Bengal. Three other properties. The 3B42 rain rates are selected to salient features are clearly seen from Fig. 2a. First, two match the size of any tracked CS during its life cycle. maximum centers are located inside the Tibetan region. Since both CTD variables and 3B42 rain rates are at the This feature is expected since all of these CSs originat- same 3-h interval temporal scale, a spatial matching ed in the Tibetan region, which would lead to a high procedure is applied. First, the size and coverage of any occurrence of CSs over the plateau. The dominant tracked CS at a designated time are determined by an maximum in central-east/southern Tibet and the sec- ellipse equation based on its central position, eccentric- ondary center in central-southern Tibet have never ity e, semimajor axis a, semiminor axis b, angle ␪ be- been revealed in detail before. Second, the obvious tween the semimajor axis and the northward direction, broad eastward orientation of the relatively large am- and radius (r) of a circle with an equivalent area of this plitude of CS frequencies indicates the eastward move- oval. The equations controlling the ellipse shape are ment of the CSs, which affects the rainfall over the a ϭ r/͌e and b ϭ r͌e. Second, the 3B42 rain rates Yangtze River basin, east China, and even the Korean located in the oval area are matched to the pixels of this Peninsula and Japan Sea. Third, the southward orien- CS to have a spatial distribution of the CS surface rain- tation of large amplitudes of CS frequencies shows a fall. The CS center position is used for statistical analy- southward propagation of CS activities that mainly im- sis of CS initiation, tracking, and intensity. These pro- pact southwest China, the Indo-China Peninsula, and cedures are followed for all tracked CSs. the Bay of Bengal. It is also evident that the southward movements are first concentrated in the southeast val- 3. Climatology of summer Tibetan CSs ley of the Tibetan Plateau, and then propagate with a southwest direction into the Bay of Bengal. This fea- a. Spatial distribution of CS frequency ture demonstrates that the Himalaya Mountains pri- A total of 643 CS life cycles originating from the marily block CSs direct southward propagation. In ad- Tibetan region are identified during the boreal sum- dition, there are rarely CS occurrences over the south- mers of 1998–2001. The spatial distribution of these CS east coast of China where the west Pacific subtropical occurrences is virtually considered as climatology over high persists in summer. This suggests that the propa-

Unauthenticated | Downloaded 09/30/21 10:48 PM UTC 2684 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 47

FIG. 3. (a) Evolution of mean characteristics of summer CSs originating from the Tibetan Plateau as a function of life cycle: (top to bottom) TBmin, maximum radius (Rcs), strong convection coverage (Fc), radius of maximum CC (Rcc-max), and CC counts (Ccc). The dotted, dashed, dashed–dotted, and solid lines are for CSs propagating eastward, westward, not out of Tibet, and overall mean, respectively. (b) As in (a), but for the parabolic fitted results. gation of summer Tibetan CSs is closely influenced by located at the central-eastern area of the plateau. In the large-scale environmental circulations and the addition, the Sichuan basin and the upper-middle seg- blockage of the huge geographic Himalaya. ment of the Yangtze River basin have relatively large The summer Tibetan CSs are categorized into three magnitudes (30%–40%). The contributions over other different groups: 1) CSs that stay inside the Tibetan areas are about 10%–30%. Results suggest that rainfall Plateau during their life cycles; 2) eastward-propa- from Tibetan CSs is an important contributor to total gating CSs: they stay mainly to the north side of a precipitation over the Tibetan Plateau and its surround- southeast diagonal from the bottom right corner of the ing regions. This is consistent with published results Tibetan region when moving out of the plateau; 3) from Zhong et al. (1994), Yu (2001), and Zhang et al. south-propagating CSs: they stay mostly to the south (2001). side of this diagonal during their life cycles. As demon- b. Characteristics of Tibetan CSs strated in the following sections, the first, second, and third category of summer Tibetan CSs is actually the Five parameters extracted from the ISCCP datasets relatively short-lived local, long-lived eastward, and are utilized to present the characteristics of the summer southward-moving CS, respectively. More discussions Tibetan CSs: 1) Rcs—CS radius (size of CS); 2) TBmin— on the categorized CS movements can be found in sec- CS minimum cloud-top brightness temperature (inten- tion 4a. sity of convection); 3) Fc—percentage of strong convec- Ͻ The contribution of rainfall from summer Tibetan tive fraction defined by the ratio of CC areas (TBmin Ͻ CSs to total precipitation is one of the important indices 220 K) to the CS coverage (TBmin 245 K) (fraction of showing the impact of the Tibetan CSs on surface pre- relatively stronger convection); 4) Rcc-max—maxi- cipitation. Figure 2b presents a spatial distribution of um CC radius (size of the largest CC inside a CS); and rainfall contributions (%) from Tibetan CSs to total 5) Ccc—CC count (number of CCs inside a CS). precipitation. It is evident that the maximum center Figure 3 displays the evolution of the averaged life with amplitudes Ͼ60% (up to a maximum of 76%) is cycles of categorized summer Tibetan CSs during 1998–

Unauthenticated | Downloaded 09/30/21 10:48 PM UTC OCTOBER 2008 YAODONG ET AL. 2685

2001. These CSs are averaged at a 3-h temporal scale from their initiations over the plateau for different cat- egories so that the evolution of the mean life cycles for the categorized CSs can be displayed. This approach demonstrates how the different categorized CSs evolve in terms of variability, intensity, size, propagation, and life span. It is obvious that the life span of CSs staying inside the Tibetan Plateau is only up to 66 h, while those CSs moving eastward and southward have a maximum life span of 192 and 288 h, respectively.

TBmin decreases significantly and Rcs increases consid- erably in the first 3 h after the formation of a CS, while both Rcc-max and Ccc increase significantly. In general, TBmin of an eastward-moving CS continually decreases in the first 36-h life cycle, sustains the minimum cloud temperature for the next 36 h, and then gradually rises, while the variations of Rcs, Rcc-max, and Ccc are almost the opposite. However, the percentage of relatively stronger convective areas decreases significantly during the first 24 h and then gradually keeps decreasing. The southward-moving CSs have similar features to the eastward-moving CSs for the first 48 h; however, they continually intensify as TBmin decreases and Rcs /Rcc-max and Ccc /Fc increases. The similarities between the eastward- and southward-moving CSs during the first 2-day life cycle are possibly due to the fact that they are still strongly influenced by the thermodynamical forces on FIG. 4. Mean properties of summer Tibetan CSs at five selected ϭ ϭ 1 their way out of the Tibetan region, while dissimilarities life stages: 1 initiation, 2 ⁄2 time between the initiation and the mature stage, 3 ϭ mature, 4 ϭ 1⁄2 time between mature and are possibly caused by interactions with different envi- ϭ dissipation, and 5 dissipation: (top to bottom) TBmin, maximum ronmental circulations during further development and radius (Rcs), strong convection coverage (Fc), radius of maximum propagation after moving out of the region. In addition, CC (Rcc-max), and CC counts (Ccc). the southward-moving CS has a stronger intensity and a longer life cycle. The detailed physical mechanisms and associated environmental circulations for these two the Tibetan Plateau are equally important as the south- kinds of CSs are beyond the scope of this paper, but will ward CSs; however, it shows a bias toward the south- be the subject of additional research. ward-moving CSs after that early stage when influences It is also noticed that the curve for all CS cases is near from the southward-moving CSs increase. At the late the mean value of the three categorized CSs at the first stage of the CS life cycle, the all-CS curve is equivalent 84 h of the CS life cycle, and the curve starts to bias to the southward-moving CS curve because the contri- toward the time series of the southward-moving CS butions are entirely from southward-moving CSs. thereafter. The following three reasons could possibly The CSs show a strong diurnal variation during their explain this feature: 1) the curve for all cases is not a life cycles (detailed analysis on diurnal behavior is in weighted average of curves from the three categorized section 3c). The mean diurnal amplitudes of CSs stay- CSs so that CSs with more populations would have ing inside the region are much larger than those moving more impacts on the final mean curve; 2) the CSs mov- out. This suggests that CSs staying inside the Tibetan ing southward are stronger and last longer than the region are mostly locally forced thermodynamic con- other two categorized CSs; and 3) there is a relatively vection. large frequency of the southward-moving CS popula- For comparison, the mean characteristics of summer tions whose life cycles are longer than 84 h (more dis- Tibetan CSs at five different stages of life cycles are cussion can be found in the description of Fig. 12). shown in Fig. 4. The five stages are defined as (i) ini- Therefore, the curve for all cases is not biased toward tiation, (ii) 1⁄2 time between the initiation and the ma- the southward-moving CSs at the early stage when con- ture stage, (iii) mature, (iv) 1⁄2 time between mature and tributions from CSs moving eastward and staying inside dissipation, and (v) dissipation. The initiation stage is

Unauthenticated | Downloaded 09/30/21 10:48 PM UTC 2686 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 47

TABLE 1. The characteristics—means and std devs—of CSs originating from the Tibetan Plateau, in terms of radius (Rcs), minimum cloud-top temperature (TBmin), coverage of convective portion (Fc), maximum radius of cloud clusters (Rcc-max), and CC count (Ccc).

Rcs TBmin Fc Rcc-max Parameter (km) (K) (%) (km) Ccc Mean 339.7 203.4 22.5 119.4 13.0 Std dev 193.5 10.6 15.1 79.0 14.3 when a CS begins to be tracked. The mature stage is when the CS TBmin is at its lowest magnitude; while the dissipation stage is the latest time of a tracked CS (similar features are found with defining the mature stage as FIG. 5. Diurnal cycle of initiation of summer Tibetan CSs in Rcs at its largest magnitude). It is evident that the CS 1998–2001. TBmin is higher at its initial and dissipation stages, as is expected. The CS TBmin at the intermediate stages is 2006, 2008; Yang et al. 2006b, 2008). The initiation of between the lowest temperature at the mature stage the summer Tibetan CS is influenced by the diurnal and the initialization and dissipation stages. The corre- cycle of solar radiation and Tibetan thermodynamics. sponding change of CS size is the opposite of the varia- Case studies indicate that these CSs have a similar dition of CS TBmin. The total number of CCs and the urnal cycle with a frequency peak around 1800 local largest CC radius follow the same pattern. The fraction solar time (LST) (Zhu and Chen 2003); however, these of relatively stronger convection area shows a different case studies are not able to provide an accurate diurnal feature for the southward-moving CSs. Results demon- description of the Tibetan CS. Figure 5 exhibits an strate that the approach shown in Fig. 4 presents the overall diurnal variation in the generation of summer composite features of CSs at five different stages of life Tibetan CSs in 1998–2001. A rapid increase of CS ini- cycles; however, it does not exhibit the evolution of the tiation around 1200 LST is clearly evident, with a maxi- categorized Tibetan CSs. mum around 1500 LST. A total of 260 CSs occurred The statistical mean and standard deviations of those around 1500 LST, which accounts for about 40% of all variables shown in Fig. 3 are listed in Table 1. It is CSs. A relatively weak secondary maximum is evident evident that the averaged Rcs is about 340 km with around 0600 LST. These results demonstrate that sum- TBmin around 203 K. The relatively small TBmin stan- mer Tibetan CSs have a strong diurnal variability. dard deviation shows that CS cloud-top height does not Figure 6 presents spatial distributions of the diurnal change dramatically, while the relatively large standard frequency of summer CS initiation over the Tibet Pla- deviations of Rcs (193 km) and Rcc-max (79 km) indicate teau. It is evident that the diurnal cycle of CS initiation that CS sizes vary substantially. The mean CC count has significant regional characteristics. The initiation of (Ccc) of 13 with a large standard deviation (14) demon- the summer CS increase when the sun rises, with a strates that the summer CS has a considerable CC maximum around 1500 LST (when the surface heating variation at different stages of its life cycle. In addition, is at peak). It also has a relatively large center over the the relatively large Rcc-max with respect to Rcs shows central-eastern Tibet region in the midafternoon. A that there are many small-sized CCs inside s. It suggests clear secondary maximum of the CS initiation is over that one or more CCs favor more intensification than the central-eastern region at 0600 LST. Thus, it appears others inside a CS, indicating inhomogeneous develop- that CSs favor the central-eastern region as their start- ment of CCs inside CSs. Therefore, these results hint at ing location. an asymmetrical development process during CS life Precipitation is one of the important indices in de- cycles. The small Fc suggests that the relatively deep scribing characteristics of the Tibetan CS. Figure 7a convection activities are not active inside summer Ti- exhibits the diurnal variations of total precipitation and betan CSs. CS-only precipitation over the region for four summers (1998–2001). The total precipitation refers to rain from c. Diurnal cycle all raining systems, while the CS-only precipitation in- Diurnal variation is one of the most important phe- dicates rain only from Tibetan CSs. It illustrates that nomena of atmospheric variables (e.g., Yang and Smith the summer total precipitation and CS-only-induced

Unauthenticated | Downloaded 09/30/21 10:48 PM UTC OCTOBER 2008 YAODONG ET AL. 2687

FIG. 6. Spatial distributions on a 1.5° ϫ 1.5° grid of diurnal frequency of summer Tibetan CS initiation. The number in the upper left corner of each plot is the LST. The dotted contours show the topography at 3000 and 5000 m. rainfall have consistent prominent diurnal properties. Tibetan Plateau and surrounding regions. Results illus- The total precipitation has a dominant peak around trate that the CS-only-induced rainfall has a major im- 1800 LST and a secondary maximum around 0300 pact on the diurnal variation of total precipitation. LST, while CS rain has a dominant peak around 0000– Figure 7b illustrates the diurnal variations of rainfall 0300 LST and a secondary maximum around 1800 LST. from the categorized summer Tibetan CSs over their The relatively larger late-night CS precipitation indi- life cycles. The midafternoon and late-night rainfall cates in general that the intensity of the summer Ti- peaks are obvious for CSs that do not move out of the betan CS is stronger in late night than in mid–late af- region, while these CSs moving out of the region have ternoon, suggesting that these CSs have a late-night a dominant late night peak. It is interesting to see that intensification process. In addition, it is understandable the CSs moving southward have a clear secondary rain- that the diurnal amplitude of CS rainfall is much stron- fall peak. However, CSs propagating eastward do not ger than that of total precipitation. We have also shown have this feature; instead, a phase of the weak rainfall in Fig. 2b that the CS-induced precipitation contribu- maximum is delayed a few hours. tion to total precipitation is very important over the These results clearly suggest possibly different

Unauthenticated | Downloaded 09/30/21 10:48 PM UTC 2688 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 47

afternoon. The nighttime intensification process of the rainfall diurnal cycle described in the “static radiation– convection” and “dynamic radiation–convection” mechanisms proposed by Ramage (1971) and Gray and Jacobson (1977) could also explain that nighttime fa- vors the development of the Tibetan CS. Huang et al. (1986) demonstrate that the heavy rainfall activities over southeast China are often accompanied by the late night peak of the summer low-level jet diurnal cycle. This could also link to the nighttime intensification process of the Tibetan CS propagating eastward and southward. To independently check the 3B42 rainfall diurnal cycle, the TRMM PR–based rain datasets (2A25) over the Tibetan region for four summers (1998–2001) are also utilized in the diurnal analysis (Fig. 7c). It is clear that the total precipitation has a dominant afternoon peak at 1500–1800 LST with a secondary late night maximum at 0000–0600 LST. Further analysis of convective and stratiform rain shows that convective rain has only one prominent afternoon peak forced by the surface solar radiative heating, while stratiform rain has a relatively weak diurnal amplitude with a maximum at 0300–0600 LST and a secondary peak at 1500–1800 LST. The diurnal variations of the total and convective precipitation are very close, and the late-night peak of stratiform rain is primarily responsible for the weak late-evening maximum of the total precipitation. Yang and Smith (2008) and Yang et al. (2008) document in detail a global spatial distribution of the multiple modes of the rainfall diurnal cycle and the modulation of stratiform rain on the secondary peak. These facts demonstrate that the afternoon variability of rainfall in the Tibetan region is mainly controlled by convective ac- FIG. 7. (a) Diurnal variability of the CS-only precipitation and tivities forced by the surface solar heating, while strati- total precipitation over the Tibetan region based on four-summer form rain leads to the secondary late-night peak. Com- TRMM blended rain dataset (3B42). (b) Diurnal variation of pre- parisons of the total precipitation over the Tibetan Pla- cipitation from the categorized summer Tibetan CSs during their life cycles. (c) Diurnal cycle of the categorized rainfall from teau in Figs. 7a,c show that diurnal amplitude is slightly TRMM PR rain dataset (2A25) for same four summers; the all, smaller from 2A25 than from 3B42, and there is also a convective, and stratiform precipitation from 2A25 are indicated very small difference in the diurnal phase between by T, C, and S, respectively. them. The 2A25 slight underestimation is due to its physical assumptions in the algorithm, which have been mechanisms for precipitation from summer Tibetan discussed by many investigators (Wolff et al. 2005; CSs. The midafternoon peak of CS rainfall suggests the Yang et al. 2006a; Kummerow et al. 2001). The very forced surface heating, whereas the late night peak and small difference in the diurnal phase might be caused the delayed afternoon maximum might be the com- by different instrument observations and sampling. bined outcome of four different mechanisms. The first However, the similarities between these two diurnal is the so-called mobile terrain-forced precipitating sys- cycles are the dominant features, indicating a reliability tem (MTFPS) mechanism summarized by Yang and of the TRMM 3B42 rain datasets applied in this study. Smith (2006). The mobility of the Tibet CS extends the A phase shift is obvious between diurnal cycles of the rainfall maximum from afternoon into late afternoon– summer Tibetan CS occurrences and the associated late night when the CS normally originates in the mid- precipitation in Figs. 5, 7a. The phase of the afternoon

Unauthenticated | Downloaded 09/30/21 10:48 PM UTC OCTOBER 2008 YAODONG ET AL. 2689

TABLE 2. Differences of CS characteristics—means and std 4. CS life cycle and development devs—during daytime and nighttime. a. Initiation of CSs Rcs TBmin Fc Rcc-max Time Parameter (km) (K) (%) (km) Ccc Figure 8 presents a spatial distribution of the summer Nighttime Mean 359.8 201.6 24.0 127.0 13.4 Tibetan CS initiations. The prominent feature is the Std dev 195.0 10.2 14.2 71.0 13.8 two local maximum centers near (30°N, 100°E) and Daytime Mean 318.2 205.2 20.9 111.0 12.7 (29°N, 92°E). This pattern is closely linked to activities Std dev 189.5 10.6 15.8 86.0 14.8 of the south–north-oriented shear line over the Tibetan Plateau (Gao et al. 1984). The east maximum center is much stronger than the west center. This differs from rain maximum is about 3 h behind the CS occur- findings by Jiang and Fan (2002) in which the west rences because of the time needed for precipitation in- maximum center is stronger. Such a difference could be tensification processes associated with the CS develop- due to the different datasets involved in these two stud- ments. ies. The CS formed in the Tibetan region with a mini- A comparison of the summer Tibetan CS properties mum radius of 55 km and life cycle of 3 h were included during daytime (0600–1800 LST) and nighttime (1800– in Jiang and Fan (2002), while we focus on relatively 0600 LST) is further conducted to explain the rainfall larger CSs that have a minimum radius of 90 km and diurnal cycle discussed above. Table 2 lists the statistics 6-h life spans. They also indicate that CSs that originat- of five variables representing the CS properties. They ed in the western center rarely move out of the plateau, show that CSs have in general larger Rcs, Fc, Rcc-max, which illustrates that these CSs have small sizes and Ccc, and lower TBmin with associated smaller standard short life spans, while those from the eastern center deviations during nighttime than daytime, indicating often propagate out of the region. Based on this study, that summer Tibetan CSs generally have a nighttime CSs staying inside the Tibetan region have smaller and intensification process that could partially explain the shorter life spans than those moving out; thus, more dominant late-night peak of precipitation. small-size and short-lived CSs in the Jiang and Fan

FIG. 8. Spatial distribution on a 1.5° ϫ 1.5° grid of Tibetan CS initiations based on four summers (1998–2001). The dotted contours show the topography at 3000 and 5000 m.

Unauthenticated | Downloaded 09/30/21 10:48 PM UTC 2690 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 47

FIG. 9. As in Fig. 8, but by month: (top to bottom) June–August.

(2002) study could shift the maximum center to the that these mobile Tibetan CSs will generally impact western region. In addition, there is a clear temporal rainfall in the south and east vicinity of the Tibetan variation in the regions where CSs formed (Fig. 9). The Plateau. The eastward-moving CSs sometimes could af- coverage and frequency of the summer Tibetan CS ini- fect rainfall over east China and even possibly the Ko- tiation increase from June to August. A slight north- rea–Japan region, while the southward-moving CSs ward shift of the frequency maximum centers is also could reach to the Bay of Bengal near the Thailand visible. This is possibly due to the northward migration coast and occasionally to the Indian Ocean. It is rare to of maximum solar radiation. find any summer Tibetan CS reaching the southeast Figure 10 shows the trajectories of all CSs that moved China coast area because of obstruction of the persis- out of the Tibetan Plateau with eastward (orange lines) tent summer west Pacific subtropical high. and southward (blue lines) propagations. It is evident The frequency statistics of the three categorized sum-

Unauthenticated | Downloaded 09/30/21 10:48 PM UTC OCTOBER 2008 YAODONG ET AL. 2691

TABLE 3. Annual variation of different CSs originating from the Tibetan Plateau.

Move out of Move out of Year Tibet eastward Tibet southward Stay in Tibet 1998 35 40 93 1999 39 45 77 2000 48 32 73 2001 38 27 96 Total 160 144 339

meteorological variables. Figure 11 presents the (a) PDF and (b) CDF of the selected 643 summer Tibetan CSs in 1998–2001. It shows that both the PDF and CDF have a similar distribution pattern for each year, except for 1998 in which short-lived CSs dominated. Overall, FIG. 10. Trajectories of summer Tibetan CSs moving out of the the percentage of CS life spans in 12, 24, 36, and 48 h region in 1998–2001. The blue lines mark CSs propagating south- are about 23%, 26%, 16%, and 10%, respectively. The ward while the orange lines mark CSs moving out in an easterly mean CDF curve indicates that almost 50% of the sum- direction. The dashed box is the selected Tibetan region. mer Tibetan CSs disappear in 24 h, with 75%, 85%, and 95% disappearing in 48, 60, and 108 h, respectively. mer Tibet CSs are listed in Table 3, illustrating that 339 Those features explain why over 50% of CSs did not CSs (52.7% of total CS) did not move out of the Ti- move out of the region, because of their short life betan region, while 160 CSs (24.9%) moved out in an cycles. The occasional long life spans over 7 days might easterly direction and 144 CSs (22.4%) out toward the be because they merge into another system, or are sim- south. Therefore, more than one-half of the summer ply due to uncertainties in the ISCCP CTD (Machado CSs die inside the region, while the remaining CSs et al. 1998). virtually split between southerly and easterly propa- The PDF of the three categorized summer Tibetan gations. There are slight annual variations of CS oc- CSs (Fig. 12a) show that CSs staying inside the plateau currences; however, these changes are small and not are dominated by short life cycles, suggesting that the significant because of only four summer datasets. Nev- localized system mainly depends on thermodynamical ertheless, it suggests a stable annual Tibet thermody- forcing. The east-propagating CSs have relatively more namical forcing. percentages than the south-propagating CSs for life spans of 48–72 h, indicating interactions with the sur- b. Life cycle of CSs rounding atmospheric circulations. It is also apparent The probability density function (PDF) and cumula- that there are more south-propagating CSs for life tive distribution function (CDF) are the commonly cycles longer than 84 h. Similar patterns exist for used parameters in illustrating statistical properties of the PDF of precipitation associated with these three-

FIG. 11. (a) PDF and (b) CDF of summer CSs originating from the Tibetan Plateau in 1998–2001.

Unauthenticated | Downloaded 09/30/21 10:48 PM UTC

Fig 10 live 4/C 2692 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 47

mm hϪ1, while those staying in the region are 291.5 km, 204.7 K, 21.4%, 100.2 km, 9.7, and 0.45 mm hϪ1, respec-

tively. They show a 50% increase for Rcs and Rcc-max, 5% rise in relatively stronger convection, and 100%

more Ccc as well as a 45% increase of rain intensity, from CSs staying in the region to those moving out of the area. Results illustrate that CSs moving out of the Tibetan region have more rainfall and longer life cycles when compared with those staying inside the plateau. These also suggest influences of easterly-moving CSs on the upper-middle segments of the Yangtze River basin, and potential impacts on faraway regions from the southerly-propagating CSs. These results are consistent with the findings by Yu and Gao (2006) that 93% of the Tibetan vortexes moving out of the Tibetan region produce more rainfall than those staying inside the region. Those CSs moving out of the plateau appear to have favorable environmental circulations that ex- tend their life cycles. It also indicates that the surrounding atmospheric conditions play an important role in the development of the summer Tibetan CS.

5. Discussion and conclusions All CSs originating from the Tibetan Plateau defined by the ISCCP deep convection database along with associated TRMM precipitation datasets for four summers (June–August 1998–2001) have been analyzed to characterize their occurrence frequency, spatial distribution, development, life cycle, track, and precipitation. There are three types of summer CSs that have their initiation over the Tibetan Plateau. The first one does not move out of the plateau and dominates the CS FIG. 12. (a) PDF of summer Tibetan CS life cycles for those population with 53% of frequency. The second cat- moving easterly and southerly out of the region, and those staying inside. (b) As in (a), but for associated surface rainfall. egory, counted at 25% of the CS population, propa- gates eastward out of the Tibetan Plateau and impacts primarily precipitation over the upper-middle segment categorized CSs (Fig. 12b). The statistics of the six pa- of the Yangtze River basin. The third type is CSs mov- rameters describing the clear differences between CSs ing southward out of the region, which accounts for staying inside and moving out of the plateau are listed about 22% of all Tibetan CSs. This type of CS produces in Table 4. The mean Rcs,TBmin, Fc, Rcc-max, Ccc, and heavy rainfall in the southwest of China, Thailand, and rain rate for CSs moving out of the Tibetan region is even to the east of the Bay of Bengal. The CS rainfall 429.7 km, 199 K, 25.9%, 150.7 km, 18.5, and 0.65 contribution to total precipitation is up to 76% over the

TABLE 4. Comparison of summer CS characteristics—means and std devs— between those moving out of the Tibetan region and those staying inside.

Ϫ1 CS category Parameter Rcs (km) TBmin (K) Fc (%) Rcc-max (km) Ccc Rain rate (mm h ) Move out Mean 429.7 199.0 25.9 150.7 18.5 0.65 Std dev 224.6 9.5 13.6 82.9 18.5 0.44 Not out Mean 291.5 204.7 21.4 100.2 9.7 0.45 Std dev 142.2 9.9 15.3 60.3 9.4 0.38

Unauthenticated | Downloaded 09/30/21 10:48 PM UTC OCTOBER 2008 YAODONG ET AL. 2693 central-eastern area of the Tibetan Plateau, and 30%– CS precipitation in late night also suggests a nighttime 40% over the east-southern adjacent regions, showing intensification process during the development of the important impacts of summer Tibetan CSs on regional summer Tibetan CSs, indicating that the mechanism for total precipitation. the CS development is different from its initiation. The two maximum initiation centers of summer Ti- betan CSs are located to the south of 33°N, one over Acknowledgments. The authors thank ISCCP for dis- the Yaluzangbu River valley to the west of 95°E and tribution of the convective tracking database and another over the Hengduan Mountain valley to the east TRMM Data Information System (TSDIS) at NASA of 95°E. The east center is much stronger, and moves Goddard Space Flight Center (GSFC) for providing slightly northward from June to August because of the 3B42 and 2A25 rainfall datasets. They are grateful for northward migration of the sun. the constructive comments from three anonymous re- The summer Tibetan CS has a mean life span of viewers to improve the quality of this paper. The au- about 36 h. Approximately 85% of the Tibetan CSs thors are supported by the National Natural Science disappear within 60 h of their initiations. The south- Foundation of China with two grants (40428002 and ward-moving CS has the longest life cycle, while those 40633018), except Song Yang who was supported by CSs staying inside the Tibetan Plateau have the shortest the NASA TRMM and Global Precipitation Measure- life cycle. The relatively stronger convection area is ment (GPM) projects. Rong Fu is also supported by the only about 22% of the CS coverage. Imbedded CCs NASA Aura project. A careful editing of the English in clearly show the CS initiation–die processes and have the manuscript by Karen Mitchell at NASA GSFC is an inhomogeneous distribution. Results also demon- appreciated. strate that the more intense a Tibetan CS is, the farther away the maximum CC is from its center, indicating an REFERENCES asymmetric development process during the CS life Boyer, D. L., and R.-R. Chen, 1987: Laboratory simulation of cycle. In addition, CSs staying inside the Tibetan Pla- mechanical effects of mountains on the general circulation of teau are generally smaller, have shorter life spans, and the Northern Hemisphere: Uniform shear background flow. J. Atmos. Sci., 44, 3552–3574. produce less rainfall than those moving out of this re- Clark, D., 1983: The GOES user’s guide. NOAA/NESDIS, 163 gion. The differences between these two types of CSs pp. suggest that the former is mainly forced by the surface Dell’Osso, L., and S.-J. Chen, 1986: Numerical experiments on the heating while the latter is due to interactions with the genesis of vortices over the Qinghai-Tibetan Plateau. Dyn. surrounding atmospheric circulations during their life Meteor. Oceanogr., 38, 236–250. Fang, Z., 1985: The preliminary study of medium-scale cloud clus- cycles. The detailed physical mechanisms behind these ters over the Changjiang basin in summer. Adv. Atmos. Sci., CSs are the subject of a follow-up study. 2, 334–340. Initiation of CSs occur mainly in the afternoon with a ——, H. Lin, L. Wu, J. Jiang, Q. Huang, and Z. Guo, 2004a: maximum around 1500 LST. The maximum frequency Visualization and characteristics of MCSs moving out Ti- in midafternoon is about 6 times the minimum fre- betan Plateau. J. Inst. Surv. Map., 21, 61–65. ——, L. Wu, H. Lin, J. Jiang, and Z. Guo, 2004b: Correlation quency at midnight, demonstrating that surface solar analysis on the factors impacting MCSs movement and pro- heating is the primary force for initiation of the summer mulgation with focus on spatial data exploration. J. Trop. Tibetan CS. The CS development and associated rain- Meteor., 20, 600–604. fall also have obvious diurnal cycles. It appears that Flohn, H., 1968: Contributions to a meteorology of the Tibetan stratiform rainfall has a distinct late night peak and Highlands. Colorado State University Atmos. Sci. Paper 130, 120 pp. midafternoon maximum while convective rainfall Fu, R., and Coauthors, 2006: Convective transport over the Ti- shows a major midafternoon peak, indicating possibly betan Plateau—A short-circuit of water vapor and polluted multiple mechanisms controlling the diurnal cycle of air to the global stratosphere. Proc. Natl. Acad. Sci. USA, rainfall over the Tibetan Plateau. The afternoon rain- 103, 2006. fall peak corresponds to the convection forced by the Gao, Y., S. Huang, and J. Lu, 1984: Climatology of Tibetan Pla- teau. Science Press, 294 pp. surface solar radiative heating. The late night maximum Gao, Z., Q. Xu, L. Lu, and Y. Cheng, 2002: Measurements of is linked to “static radiation–convection” and/or “dy- turbulence transfer in the near-surface layer over the south- namic radiation–convection” mechanisms as well as the eastern Tibetan Plateau. Bound.-Layer Meteor., 102, 281– low-level jet diurnal cycle. In addition, CSs initiated in 300. midafternoon at one location could bring delayed mid- Gray, M. E., 2001: The impact of mesoscale convective-system potential-vorticity anomalies on numerical-weather- afternoon–late night maximum rainfall to another area. prediction forecasts. Quart. J. Roy. Meteor. Soc., 127, 73–88. This process is summarized as the MTFPS mechanism Gray, W. M., and R. W. Jacobson Jr., 1977: Diurnal variation of by Yang and Smith (2006). The prominent peak of the deep cumulus convection. Mon. Wea. Rev., 105, 1171–1188.

Unauthenticated | Downloaded 09/30/21 10:48 PM UTC 2694 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 47

Guo, Z., H. Lin, X. Dai, J. Jiang, and J. Wu, 2003a: Mining east- ——, and ——, 1984: The large-scale circulation and heat sources ward-moving MCSs features over the Tibetan Plateau. J. over the Tibetan Plateau and surrounding areas during the Geo-Inf. Sci., 1, 5–10. early summer of 1979. Part II: Heat and moisture budgets. ——, ——, J. Jiang, Q. Huang, and Z. Fang, 2003b: The features Mon. Wea. Rev., 112, 966–989. of MCS and their eastward moving and propagation over the Machado, L. A. T., and W. B. Rossow, 1993: Structural charac- Tibetan Plateau. J. Remote Sens., 7, 350–357. teristics and radiative properties of tropical cloud clusters. ——, X. Dai, H. Lin, J. Jiang, and Q. Huang, 2004: Abstraction Mon. Wea. Rev., 121, 3234–3260. the environmental physical field values of influencing MCSs ——, ——, R. Guedes, and A. Walker, 1998: Life cycle variations movement. J. East China Normal Univ. (Nat. Sci.), 1, 67–72. of mesoscale convective systems over the Americas. Mon. He, J., L. Chen, and W. Li, 1984: Preliminary results from numeri- Wea. Rev., 126, 1630–1654. cal simulations of atmospheric heating sources and Tibetan Maddox, R., 1980: Mesoscale convective complexes. Bull. Amer. Plateau topography on summer monsoon circulations. Col- Meteor. Soc., 61, 1374–1387. lection of Scientific Papers from Tibet Meteorological Experi- McCollum, D. M., R. A. Maddox, and K. W. Howard, 1995: Case ment, Z. Shen, D. Wen, and S. Pan, Eds., Vol. 1, Science study of a severe mesoscale convective system in central Ari- Press, 324–332. zona. Wea. Forecasting, 10, 643–665. Huang, S., Z. Li, C. Bao, and Z. Yu, 1986: Pre-Raining Season Meneghini, R., J. A. Jones, T. Iguchi, K. Okamoto, and J. Kwiat- Heavy Rainfall. Guangdong Science and Technology Press, kowski, 2001: Statistical methods of estimating average rain- 237 pp. fall over large space–timescales using data from the TRMM Huffman, G. J., R. F. Adler, M. Morrissey, D. T. Bolvin, S. Curtis, precipitation radar. J. Appl. Meteor., 40, 568–585. R. Joyce, B. McGavock, and J. Susskind, 2001: Global pre- Olson, W. S., and Coauthors, 2006: Precipitation and latent heat- cipitation at one-degree daily resolution from multisatellite ing distributions from satellite passive microwave radiom- observations. J. Hydrometeor., 2, 36–50. etry. Part I: Improved method and uncertainties. J. Appl. ——, and Coauthors, 2007: The TRMM Multisatellite Precipita- Meteor. Climatol., 45, 702–720. tion Analysis (TMPA): Quasi-global, multiyear, combined- Qian, Y., H. Yan, Q. Wang, and A. Wang, 1988: Numerical Study sensor precipitation estimates at fine scales. J. Hydrometeor, of Topography Impacts on Planetary Atmosphere. Science 8, 38–55. Press, 207 pp. Iguchi, T., T. Kozu, R. Meneghini, J. Awaka, and K. Okamoto, Ramage, C. S., 1971: Monsoon Meteorology. Academic Press, 295 2000: Rain-profiling algorithm for the TRMM precipitation pp. radar. J. Appl. Meteor., 39, 2038–2052. Reiter, E. R., 1982: Where we are and where we are going in Jiang, J., and M. Fan, 2002: Convective clouds and mesoscale mountain meteorology. Bull. Amer. Meteor. Soc., 63, 1114– convective systems over the Tibetan Plateau in summer. Chi- 1122. nese J. Atmos. Sci., 26, 263–270. ——, 1987: Tibet revisited: TIPMEX-86. Bull. Amer. Meteor. Soc., ——, X. Xiang, and M. Fan, 1996: The spatial and temporal dis- 68, 607–615. tribtions of sever mesoscale convective systems over Tibetan Rossow, W. B., and R. A. Schiffer, 1999: Advances in understand- Plateau in summer. Chinese J. Appl. Meteor., 7, 473–478. ing clouds from ISCCP. Bull. Amer. Meteor. Soc., 80, 2261– Jin, Y.-Q., 1997: Radiative transfer of snowpack/vegetation 2287. canopy at the SSM/I channels and satellite data analysis. Re- mote Sens. Environ., 61, 55–63. ——, A. W. Walker, D. Beuschel, and M. Roiter, 1996: Interna- Kummerow, C., and Coauthors, 2001: The evolution of the God- tional Satellite Cloud Climatology Project (ISCCP) descrip- dard Profiling Algorithm (GPROF) for rainfall estimation tion of new cloud datasets. World Climate Research Pro- from passive microwave sensors. J. Appl. Meteor., 40, 1801– gramme Tech. Doc. WMO/TD 737, ICSU and WMO, 115 pp. 1820. Shen, R., E. R. Reiter, and J. F. Bresch, 1986a: A simplified hy- Leber, D., F. Holawe, and H. Häusler, 1995: Climatic classifica- drodynamic mesoscale model suitable for use over high pla- tion of the Tibet autonomous region using multivariate sta- teau regions. Meteor. Atmos. Phys., 34, 251–296. tistical methods. GeoJournal, 37, 451–472. ——, ——, and ——, 1986b: Numerical simulation of the devel- Leroux, M., 1993: The Mobile Polar High: A new concept explain- opment of vortices over the Qinghai-Xizang (Tibet) Plateau. ing present mechanisms of meridional air-mass and energy Meteor. Atmos. Phys., 35, 70–95. exchanges and global propagation of palaeoclimatic changes. Shi, C., J. Jiang, and Z. Fang, 2000: A study on the features of Global Planet. Change, 7, 69–93. severe convection cloud clusters causing serious flooding Li, Y., Q. Wang, X. Zheng, W. Guo, and W. Wang, 1989: Study of over Changjiang River Basin in 1998. Climate Environ. Res., mesoscale convective complex over southwest China. Sci. At- 5, 279–286. mos. Sin., 13, 417–422. Tao, S.-Y., and Y.-H. Ding, 1981: Observational evidence of the Liu, X., G. Wu, W. Li, and Y. Liu, 2001: Diabatic heating over the influence of the Qinghai-Xizang (Tibet) plateau on the oc- Tibetan Plateau and the thermal adaptation of the large-scale currence of heavy rain and severe convective storms in flow field. Adv. Nature Sci., 11, 33–39. China. Bull. Amer. Meteor. Soc., 62, 23–30. ——, ——, and Y. Liu, 2002: Diabatic heating over the Tibetan ——, and Coauthors, 1980: The Heavy Rainfalls in China. Science Plateau and the seasonal variation of the Asian circulation Press, 224 pp. and summer monsoon onset. Chinese J. Atmos. Sci., 26, 782– Ueno, K., 1998: Characteristics of plateau-scale precipitation in 793. Tibet estimated by satellite data during 1993 monsoon sea- Luo, H., and M. Yanai, 1983: The large-scale circulation and heat son. J. Meteor. Soc. Japan, 76, 533–549. sources over the Tibetan Plateau and surrounding areas dur- Wang, A., H. Guo, Q. Wang, and J. Zhang, 1984: Numerical study ing the early summer of 1979. Part I: Precipitation and kine- of impacts of the early summer heating forcing on the mean matic analyses. Mon. Wea. Rev., 111, 922–944. atmospheric circulations over east Asia. Collection of Scien-

Unauthenticated | Downloaded 09/30/21 10:48 PM UTC OCTOBER 2008 YAODONG ET AL. 2695

tific Papers from Tibet Meteorological Experiment, Z. Shen, ——, W. Olson, J. J. Wang, T. L. Bell, E. A. Smith, and C. D. D. Wen, and S. Pan, Eds., Vol. 2, Science Press, 273–280. Kummerow, 2006a: Precipitation and latent heating distribu- Wang, W., Y.-H. Kuo, and T. T. Warner, 1993: A diabatically tions from satellite passive microwave radiometry. Part II: driven mesoscale vortex in the lee of the Tibetan Plateau. Evaluation of estimates using independent data. J. Appl. Me- Mon. Wea. Rev., 121, 2542–2561. teor., 45, 721–739. Wolff, D. B., D. A. Marks, E. Amitai, D. S. Silberstein, B. L. ——, E. A. Smith, and K.-S. Kuo, 2006b: Diurnal variability of Fisher, A. Tokay, J. Wang, and J. L. Pippitt, 2005: Ground precipitation from TRMM measurements. Remote Sensing validation for the Tropical Rainfall Measuring Mission and Modeling of the Atmosphere, Oceans and Interactions, T. (TRMM). J. Atmos. Oceanic Technol., 22, 365–380. N. Krishnamurti, B. N. Goswami, and T. Iwasaki, Eds., In- Wu, G., and Y. Zhang, 1998: Thermal and mechanical forcing of ternational Society for Optical Engineering (SPIE Proceed- the Tibetan Plateau and the Asian monsoon onset. Part 1: ings, Vol. 6404), 64040I, doi:10.1117/12.696275. Situation of the onset. Sci. Atmos. Sin., 22, 825–837. ——, K.-S. Kuo, and E. A. Smith, 2008: Persistent nature of sec- ——, and Y. Liu, 2000: Thermal adaptation, overshooting, disper- ondary diurnal modes of precipitation over oceanic and con- sion, and subtropical anticyclone. Part I: Thermal adaptation tinental regimes. J. Climate, 21, 4115–4131. and overshooting. Chinese J. Atmos. Sci., 24, 433–446. Ye, D., 1981: Some characteristics of the summer circulation over ——, and ——, 2003: Summertime quadruplet heating pattern in the Qinghai-Xizang (Tibet) Plateau and its neighborhood. the subtropics and the associated atmospheric circulation. Bull. Amer. Meteor. Soc., 62, 14–19. Geophy. Res. Lett., 30, 1201, doi:10.1029/2002GL016209. Yibing, X., 1981: Stable and unstable planetary waves. Bull. Amer. ——, W. Li, H. Guo, H. Liu, Z. Xue, and Z. Wang, 1997: The Meteor. Soc., 62, 36. Tibet sensible heating bump and Asia monsoon. Collection of Yu, S., 2001: Primary study of the impacts of Tibetan Plateau Papers for Memory of J.-Z. Zhao, D.-Z. Ye, Ed., Science weather systems on the major flooding peaks over the Press, 116–126. Yangtze River basin in 1998: Genesis mechanism of 1998 ——, X. Liu, Q. Zhang, Y. Qian, J. Mao, Y. Liu, and W. Li, 2002a: heavy rain in the Yangtze River and Nunjiang River valley Progress in the study of the climate impacts of the elevated and its prediction. Y. Qi, Q. Yi, and M. Cheng, Eds., China heating over the Tibetan Plateau. Climatic Environ. Res., 7, Meteorology Press, 359–364. 184–201. ——, and W. Gao, 2006: Observational analysis on the movement ——, L. Sun, and Y. Liu, 2002b: Impacts of land surface processes of vortices before/ after moving out the Tibetan Plateau (in on summer climate. Selected Papers of the Fourth Conference Chinese). Acta Meteor. Sin., 64, 392–399. on East Asia and Western Pacific Meteorology and Climate, C. Chang, G. Wu, and B. Jou, Eds., World Scientific, 64–76. Zhang, S., S. Tao, and Q. Zhang, 2001: Meteorological and hy- Xiang, X., and J. Jiang, 1995: Mesoscale convective complexes drological characteristics of severe flooding in China during Chinese J. Appl. Meteor., over the southern China mainland. Chinese J. Appl. Meteor., the summer of 1998. 12, 442–457. 6, 9–17. Zhao, P., and L. Chen, 2000: Study on climatic features of surface Yanai, M., and C. Li, 1994: Mechanism of heating and the bound- turbulent heat exchange coefficients and surface thermal ary layer over the Tibetan Plateau. Mon. Wea. Rev., 122, sources over the Qinghai-Xizang Plateau. Acta Meteor. Sin., 305–323. 14, 13–29. ——, ——, and Z. Song, 1992: Seasonal heating of the Tibetan Zhong, X., S. Yang, and Y. Zhu, 1994: The precipitation charac- Plateau and its effects on the evolution of the Asian summer teristics of mesoscale convective complexes over the eastern monsoon. J. Meteor. Soc. Japan, 70, 319–351. region of the Qinghai-Xizang Plateau. Plateau Meteor., 13 Yang, S., and E. A. Smith, 2006: Mechanisms for diurnal variabil- (2), 113–121. ity of global tropical rainfall observed from TRMM. J. Cli- Zhu, G. F., and S. J. Chen, 2003: Analysis and comparison of me- mate, 19, 5190–5226. soscale convective systems over the Qinghai-Xizang (Ti- ——, and ——, 2008: Convective–stratiform precipitation vari- betan) Plateau. Adv. Atmos. Sci., 20, 311–322. ability at seasonal scale from 8 yr of TRMM observations: Zhuo, G., X. Xu, and L. Chen, 2002: Instability of eastward move- Implications for multiple modes of diurnal variability. J. Cli- ment and development of convective cloud clusters over Ti- mate, 21, 4087–4114. betan Plateau. Chinese J. Appl. Meteor., 13, 448–456.

Unauthenticated | Downloaded 09/30/21 10:48 PM UTC