2. Key Laboratory for Cloud Physics of China Meteorological Administration, Chinese Academy of Meteorological Sciences, China Meteorological Administration, Beijing 100081
The Tibetan Plateau (TP) is the largest plateau in the world with the most complex terrain and highest average elevation. The total area of TP is nearly 3 million square kilometers and the average elevation is 4000 m above sea level (ASL). The unique large topography of TP makes it a significant thermal–dynamic forcing on the atmosphere above, leading to frequent convective activities in summer (Jiang et al., 1996; Chen et al., 1999). Most of summertime precipitation over the TP shows convective property (Flohn, 1968). The development and eastward propagation of convective cloud systems over the TP exert significant impacts on disastrous weather events both locally and downstream (Yasunari and Miwa, 2006). In addition, convections over the TP are also important for the exchange of moisture, heat, and chemical species between the troposphere and stratosphere (Fu R. et al., 2006). Therefore, the study of convective clouds over the TP has always been a hot research topic worldwide.
Convective clouds play a critical role in the atmospheric circulation, energy balance, and hydrological cycle through their impacts on atmospheric radiative transfer and phase changes of hydrometeors. The radiative effect and microphysical properties of convective clouds are closely related to cloud structure, especially the vertical structure of clouds (Randall et al., 1989; Wang and Rossow, 1998). Cloud vertical structure actually reflects the thermal–dynamic processes inside the cloud as well as cloud and precipitation microphysics, and has distinct impacts on the formation and intensity of precipitation (Jakob and Klein, 1999). However, the lack of in-depth knowledge of cloud vertical structure is still one of the reasons for large uncertainties in numerical models at present. Thereby, more observations of the vertical structure of clouds are a prerequisite for further improvements on cloud simulation and precipitation forecast. Due to the influence of high topography, ground-based cloud observations in the TP are quite limited; instead, satellite remote sensing is an effective approach to obtain high-density spatial observations. At present, there are two ways to acquire vertical structures of clouds by satellite remote sensing. One is to detect vertical structure of cloud through active measurements by instruments onboard polar-orbiting satellites. CloudSat satellite (Stephens et al., 2002) and TRMM satellites (Kummerow. et al., 1998) are examples of such a type of satellites, which can provide vertical structures of radar reflectivity and microphysical parameters in the cloud. The other way is to detect cloud top by passive sensors onboard polar-orbiting satellites. Due to the low altitude of polar-orbiting satellite, the spatial resolution of the detected data is higher than that detected by geostationary satellites. For example, the Moderate Resolution Imaging Spectroradiometer (MODIS; Savtchenko et al., 2004) onboard Aqua satellite and the high-resolution radiometers onboard NOAA satellites can provide brightness temperature (TBB), cloud optical depth (COD) and particle effective radius (Re) at the cloud top. Specific retrieval technique and cloud microphysics analysis can be applied to these satellite remote sensing data to obtain microphysical structure of the cloud (Rosenfeld and Lensky, 1998; Rosenfeld and Woodley, 2000; Yuan and Li, 2010; Yuan et al., 2010; Dai et al., 2011).
Based on measurements of CloudSat and TRMM, numerous statistical analyses have been conducted to investigate vertical structure of convective clouds over the TP (Luo et al., 2011; Wang H. et al., 2011; Wang S. H. et al., 2011; Xu, 2013; Qie et al., 2014; Zhao et al., 2014; Fu et al., 2016; Liu and Chen, 2017). Among them, a few studies have revealed unique features of convective clouds over the TP. For example, compared to low-elevation regions, convective clouds over the TP have thinner vertical thickness and smaller depth of mixed-phase region. UtilizingCloudSat/CALIPSO data, Luo et al. (2011) found that deep convections over the TP are usually embedded in small-scale convective systems and shallower than convections over the southern slope of TP and the monsoon region in South Asia. However, the top layer of deep convective clouds over the TP is deeper than in other regions. Based on TRMM data, Xu (2013) proposed that deep convective precipitation over the TP heavily relies on mixed-phase precipitation processes, though the depth of mixed-phase region is smaller than that over basins, low-elevation plains, and oceans at the same latitude. Based on the TRMM data, Qie et al. (2014) compared structures and intensities of strong convections over the TP, the southern slope of TP, South Asia subcontinent, and the adjacent oceans. They found that the intensity of deep convection over the TP is relatively weak with lower vertical extension and smaller horizontal scale. In addition, several case studies have also been conducted to investigate vertical structure of typical convective cloud over the TP using the TRMM data (Fu et al., 2007; Li et al., 2012). For example, Li et al. (2012) analyzed the characteristics of a strong convective weather event over the TP, and showed that convective clouds over the TP were compressed and vertically affected by the topography and tropopause, but cloud water droplets, precipitable liquid water droplets and ice crystals were concentrated below 8-km height (ASL, the same hereafter), and the storm height was much lower than that over low-elevation plains. Unfortunately, due to low temporal resolution of polar-orbiting satellite measurements, it is hard to find a deep convection case detected by multiple polar-orbiting satellites simultaneously. Therefore, applications of polar-orbiting satellite measurements to case studies on vertical structure of convective clouds over the TP are still limited to single-type polar-orbiting satellite so far.
Based on passive remoting sensing observations by polar-orbiting satellites, Rosenfeld and Lensky (1998) proposed a method to reveal characteristics of cloud microphysical structure using parameters like TBB and Re. Several studies have been conducted to investigate vertical structure of clouds using this method. For example, Dai et al. (2011) applied this method to NOAA satellite data and explored cloud microphysical property of a thunderstorm with light precipitation. They found that the main body of such a cloud is often located above the 0°C level and the precipitation is mainly dominated by cold cloud process; meanwhile, the growth process participated by ice-phase is the main growth mode of cloud and rain particles. Based on observations of MODIS onboard the Aqua satellite, Yuan et al. (2010) proposed a technique to estimate the glaciation temperature, which is the turning point from mixed-phase process to ice-phase process in the deep convective clouds.
For a long time, ground observations of vertical structure of convective clouds over the TP are lacking; however, the reliability of satellite measurements is subject to evaluation based on surface observations. During the field campaign of the Third Tibetan Plateau Atmospheric Science Experiment (TIPEX-III; Zhao et al., 2018), comprehensive measurements of clouds and precipitation were obtained by multiple types of radars at Naqu over the central TP during July–August 2014, among which vertically-pointing radars like KA-band millimeter cloud radar and C-band frequency-modulated continuous-wave radar (hereafter C-FMCW) can directly detect vertical structure of the entire convective clouds from the ground. With high spatial and temporal resolutions, the above two radars provide excellent observations of vertical structure of convective clouds over the TP. At present, based on these observations, several studies have revealed characteristics of vertical structure of clouds and diurnal variations of precipitation over the TP in summer (Liu et al., 2015; Chang and Guo, 2016; Zhao and Yuan, 2017).
In order to further investigate the vertical structure of deep convective clouds over the TP with multiple sources of satellite remote sensing and surface observations, a deep convective event that occurred at Naqu during 1300–1600 Beijing Time (BT) 9 July 2014 is selected for the present study. This case is well observed by the TRMM, CloudSat, and Aqua. More important, it took place during the TIPEX-III period and the entire process was well captured by two vertically-pointing radars (the KA-band millimeter cloud radar and the C-FMCW radar) deployed near Naqu weather station. Therefore, these space-based and ground-based instruments jointly provided complete observations of cloud vertical structure of this deep convective process. In the present study, the characteristics of vertical structure of deep convective clouds over the TP are analyzed in depth, and the satellite observations are compared with the ground-based radar observations to verify the reliability of the satellite data. Results of the present study will provide evidence for the evaluation of numerical simulation of vertical structure of deep convective clouds over the TP.2 Data and method 2.1 Observations
Satellite and surface observations on 9 July 2014 are used in the present study. In the following, details of the observations are introduced.
Four kinds of satellite observations (TRMM, CloudSat, Aqua and MTSAT-1R) are used. TRMM, CloudSat and Aqua are polar-orbiting satellites. They all passed over Naqu once on 9 July 2014. MTSAT-1R is geostationary satellite with temporal resolution of 30 minutes. Satellite retrieval products used in this study are listed below. (1) The 2A25 and 1B11 products are retrieved from measurements of the precipitation radar (PR) and microwave scanning radiometer (TMI) onboard TRMM. The PR operates at 13.8 GHz, providing information on a horizontal resolution of approximately 5 km, a total of 80 levels in the vertical with a vertical resolution of 250 m, and a detection range of 0–20 km. The 2A25 product provides three-dimensional distribution of radar reflectivity factor. The 1B11 product is calibrated microwave TBB retrieved from the TMI measurements. In this study, the 85-GHz polarization-corrected TBB extracted from the 1B11 product used. (2) The 2B-GEOPROF, 2B-CWC-RO, and 2B-CLDCLASS products are retrieved from observations of cloud profiling radar (CPR) onboard CloudSat. The CPR operates at 94 GHz, providing observations with a vertical resolution of 240 m at 125 vertical levels and a detection range of 0–30 km. The 2B-GEOPROF product contains vertical distributions of radar reflectivity factor along the orbital trajectory of CloudSat; the 2B-CWC-RO product includes liquid (ice) water content, liquid (ice) water number concentration, and effective radius of liquid (ice) water particles. In the 2B-CLDCLASS product, clouds are classified into eight types, i.e., cirrus (Ci), altostratus (As), altocumulus (Ac), stratus (St), stratocumulus (Sc), cumulus (Cu), nimbostratus (Ns) and deep convective clouds (Dc). (3) The MYD06_L2 product is retrieved from observations of MODIS onboard Aqua (Platnick et al., 2015). In this paper, cloud microphysical parameters such as TBB at 11-μm wavelength and COD and Re at 3.7-μm wavelength are used. The horizontal resolution of TBB is 5 km, and that of COD and Re is 1 km. (4) The MTSAT-1R product provides TBB observations on a 4-km horizontal resolution.
Surface observations used in this paper include the following: (1) hourly surface precipitation collected at 34 automatic weather stations in the central TP (30°–34°N, 88°–94°E) on 9 July 2014 and surface precipitation at 10-min intervals at Naqu weather station; (2) soundings (31.48°N, 92.06°E; 4508-m ASL) at 0800 and 2000 BT 9 July 2014 at Naqu weather station; (3) C-band Doppler weather radar (31.48°N, 92.06°E; 4526-m ASL) observations at Naqu; (4) observations of KA-band millimeter wave cloud radar (31.48°N, 92.01°E; 4507-m ASL), C-FMCW (31.48°N, 92.06°E; 4507-m ASL), and disdrometer (31.48°N, 92.06°E; 4507-m ASL) deployed at Naqu during the TIPEX-III. Table 1 lists the instruments, locations, and observation parameters and their resolutions. The geographical distribution of surface observing facilities is shown in Fig. 1.
|Instrument||Location||Observation parameters and resolutions|
|KA-band millimeter wave cloud radar||31.48°N,
4507 m ASL
|Operating frequency: 33.44 GHz; observations: echo intensity, radial velocity, linear depolarization ratio, and power spectral density; vertical detection range: 0.12–15 km; vertical resolution: 30 m; and temporal resolution: 0.85 s|
4507 m ASL
|Operating frequency: 5530 ± 3 MHz; observations: echo intensity, radial
velocity, velocity spectral width, and echo power; vertical detection range:
0.02–15 km; vertical resolution: 30 m; and temporal resolution: 2–3 s
|C-band Doppler weather radar||31.48°N,
4526 m ASL
|Conventional C-band operational radar, providing echo intensity, radial
velocity and velocity spectral width; temporal resolution: 6 minutes
4507 m ASL
|Wavelength: 650 nm; frequency: 50 kHz; transmit power: 3 m W; radius
ranges: 0.2–5 mm for liquid particles, 0.2–25 mm for solid particles;
types of particles in total: 1024 ; and temporal resolution: 1 minute
4508 m ASL
|L-band, providing elements of temperature, pressure, humidity, winds, etc. at temporal resolution of 12 h (0800 and 2000 BT); extra observations at 1400 BT were conducted during aircraft observations in the TIPEX-III of 2014|
Radar reflectivity data from the C-band Doppler weather radar deployed at Naqu are remapped to gridded data with horizontal resolution of 1 km and vertical resolution of 0.5 km. The data cover an area of 400 km × 400 km and 0.5–20 km in the vertical. The REORDER software package (Oye and Case, 1995) developed by the NCAR is used for radar data interpolation, which adopts a closest point weighting function. The effective radius used in the interpolation is 1 km for radar range, 1° for azimuth angle, and 2° for elevation angle. This method has been applied in previous research (Wang et al., 2014). In the present study, gridded observations of C-band Doppler weather radar are used to calculate composite radar reflectivity, whose horizontal distribution is used to demonstrate the echo evolution during the deep convective process.
The bright band is determined based on vertical distribution of C-FMCW radar echoes. For each observed echo profile, the maximum and minimum vertical gradients of radar reflectivity over 0.2–2 km above the C-FMCW radar are calculated, and their corresponding heights (H1 and H2, H1 < H2) are determined. The maximum curvature heights are then determined from H1 downward 300 m and H2 upward 300 m, respectively, and the maximum curvature heights are the bottom and top heights of the bright band. In order to ensure echoes inside the bright band maintain certain levels of intensity, it is required that the echo intensity in the bright band must be greater than –10 dBZ and the maximum value is not less than 5 dBZ.
Following the approach proposed by Rosenfeld and Lensky (1998) for analysis of vertical structure of cloud microphysics based on satellite data, cloud microphysical parameters (TBB, Re) under various cloud top heights are used to investigate Re changes with the cloud top height (represented by TBB). The result is displayed by the so-called T–Re relation, which simplifies microphysical processes in cloud into five main processes, i.e., diffusional droplet growth, coalescence droplet growth, rainout, mixed-phase precipitation, and glaciation. Cloud microphysical parameters (TBB, COD, and Re) retrieved from the Aqua MODIS data are used to obtain the T–Re relation, which is then applied to analysis of vertical structure of cloud microphysics. Similar to the method proposed by Yuan et al. (2010), deep convection pixels are defined as the pixels satisfying TBB < 260 K and COD > 30 in the MODIS observations.
The method to determine deep strong convection and deep weak convection is a from Pan and Fu (2015) and Fu et al. (2016). Based on the TRMM observations, deep strong convection is determined if the storm height (the first layer at which the echo detected by PR exceeds 17 dBZ and echoes in three consecutive layers all exceed 17 dBZ) is above 7.5 km and reflectivity of at least one layer echo detected by PR exceeds 39 dBZ; deep weak convection is determined if the storm height is above 7.5 km while the reflectivities of whole layer echoes are all smaller than 39 dBZ.3 Evolution of the deep convective clouds
Since TRMM, CloudSat, and Aqua detected deep convective clouds at different times during 1300–1600 BT, it is necessary to first understand the evolution and development of the deep convective clouds and then determine which stages of the convective process these polar-orbiting satellites observations corresponded to. In the following, the evolution of the deep convective cloud near Naqu is analyzed based on TBB, station observations of precipitation, and radar echoes, respectively.
Figure 2 presents the evolution of TBB retrieved from MTSAT-1R. Two northeast–southwest oriented deep convective clouds were identified: one was located near Naqu weather station with a horizontally dense structure, and the other was located about 100 km to the west of Naqu weather station, with a horizontally loose structure. The one near Naqu experienced a complete process of formation, development, and decay during 1300–1600 BT: a convective cell formed at 1301 BT, reached its mature stage at 1432 BT, continued to expand horizontally after 1501 BT, weakened since 1601 BT, moved out of the Naqu weather station, and started to dissipate at 1701 BT. The one to the west of Naqu started 1–2 h later than that near Naqu, and developed during the entire period of 1432–1701 BT.
Hourly precipitation collected at automatic weather stations near Naqu can well represent the distribution of precipitation caused by the deep convective process (Fig. 3a–c). Precipitation near Naqu was small (less than 1 mm h–1) during 1300–1400 BT, reached the maximum of up to 5 mm during 1400–1500 BT, and became weak during 1500–1600 BT (less than 2 mm h–1). Precipitation associated with the deep convective cloud to the west of Naqu started at 1500 BT, and it was less than 1mm h–1 during 1500–1600 BT. Apparently, more precipitation was induced by the deep convective cloud near Naqu, which is consistent with that observed by MTSAT-1R (Fig. 2). During 1300–1600 BT, precipitation occurred at Naqu with an accumulative amount of 6 mm. The largest 10-min accumulative precipitation occurred during 1400–1430 BT. Precipitation started to decrease after 1430 BT and ended after 1530 BT (Fig. 3d).
Due to the limitation of detection range of the C-band weather radar at Naqu, only the deep convective cloud near Naqu weather station could be observed. At around 1300 BT, a northeast–southwest oriented rainbelt appeared 30 km to the west of Naqu, which contained 3–4 convective cells and a wide stratiform precipitation region to the west and northwest of the convective cells (Fig. 4a). The convective cells were intensified while the rainbelt moving from west to east, forming a northeast–southwest oriented convective rainbelt at around 1330 BT with the central maximum echo intensity greater than 40 dBZ (Fig. 4b). At around 1400 BT, the convective rainbelt passed over the C-band weather radar at Naqu, and the convective intensity started to decrease. By 1500 BT, the main body of the convective rainbelt near Naqu radar significantly weakened, and stratiform precipitation area to the northeast of Naqu weather station expanded. After 1600 BT, the rainbelt completely moved out of the detection range of the C-band weather radar at Naqu.
The deep convective cloud near Naqu weather station was observed by polar-orbiting satellites TRMM, CloudSat, and Aqua at 1418, 1452, and 1520 BT, respectively (Fig. 5). From TBB distribution from MTSAT-1R (Fig. 2), it can be found that the TRMM PR and Aquasimultaneously observed the deep convective clouds above Naqu and to its west (Figs. 2b, 5a, 5b, 5e); however, CloudSat could only observe the deep convective cloud to the west of Naqu (Figs. 2c, 5c). The above analysis of the evolution of TBB, surface precipitation and radar echoes indicates that the deep convective cloud near Naqu observed by the TRMM PR was on its mature stage (Figs. 2b, 3b, 4d), while the one to the west of Naqu was at its early stage (Fig. 2b). The deep convective cloud near Naqu observed by AquaMODIS was on its decaying stage (Figs. 2d, 3c, 4e–f), while that to the west of Naqu was at its mature stage (Fig. 2d). Based on these satellite observations, the overall features of deep convective clouds near Naqu and to its west are as follows.
During the mature stage of the deep convective cloud above Naqu weather station, the TRMM observations show an organized rainbelt extending about 100 km along the northeast–southwest direction with four embedded rainfall centers. The maximum precipitation intensity was greater than 20 mm h–1 (Fig. 5a), and the minimum 85-GHz polarization-corrected TBB was 180–200 K, indicating large content of ice particles in the convective cloud (Fig. 5b). During the decaying stage of the convective cloud, the Aqua MODIS observations showed correspondingly lower TBB and larger COD (TBB < 220 K, COD > 30; Figs. 5e, f), implying that the deep convective cloud was associated with low cloud top temperature, was thick, and was cold.
During the early stage of the deep convective cloud located 100 km to the west of Naqu weather station, TRMM observed a rainbelt of about 200 km long extending the northeast–southwest. Precipitation along this rainbelt was weak and not well organized, the maximum precipitation intensity in the center of the rainbelt did not exceed 20 mm h–1, and the minimum 85 GHz polarization-corrected TBB was 200–220 K. At the development stage of the deep convective cloud, the CloudSat CPR observed echoes of convective cloud that reached high altitude (Fig. 5c) with four embedded deep convective cells. All the four convective cells remained upright, accompanied by cirrus clouds at the upper level and altocumulus at the middle and lower levels (Fig. 5d). At the mature stage of the convective cloud, the Aqua MODIS observations also indicate that the convective cloud exhibited deep cold cloud features (Figs. 5e, f).4 Vertical structure of radar echoes of convective clouds
Vertical structure of radar echoes of the convective clouds was well observed by TRMM, CloudSat and ground-based vertically-pointing radars. Figure 6 displays cross-sections of echoes across the strong echo centers near Naqu weather station (Figs. 6b, d) and to its west (Figs. 6c, e) from the TRMM PR observations. The horizontal scale of convective cells in the two rainbelts was about 10–20 km, and their vertical scale was above 15 km, indicating that convective clouds over the TP had small horizontal scale but could reach high altitude vertically. The convective cell embedded in the strong echo belt of the convective cloud near Naqu weather station could reach up to about 10–15 km, the top height of 35 dBZ was about 8–11 km, and the vertical thickness of the layer of larger than 35 dBZ was 3–6 km (Figs. 6c, e). But the vertical extend of convection was relatively thin due to the high terrain elevation over the TP. The convective system to the west of Naqu weather station consisted of 3–4 scattered strong convective cores, which were weaker than that over Naqu weather station. The maximum echo intensity was less than 40 dBZ, but the echo top could reach above 10 km (Figs. 6b, d). According to the definition of deep convection strength, it can be deduced that the convective system above/near Naqu weather station was a deep strong convective system while the one to the west was a deep weak one.
Figures 7a–c and 8a–c show the time–height distributions of reflectivity, radial velocity, and velocity spectral width of the deep convective cloud above Naqu weather station during 1330–1600 BT 9 July 2014 observed by the C-FMCW and KA-band millimeter cloud radar, respectively. These two ground-based vertically-pointing radars both observed the complete evolution process of this deep convective cloud; however, the time of the C-FMCW observation was 10 minutes later than that of the cloud radar observation. This is attributed to two main reasons: firstly, the C-FMCW was located 5 km to the east of the cloud radar, so the deep convective cloud moved from west to east and passed over the cloud radar first before it passed over the C-FMCW; secondly, the capability of C-FMCW to detect incipient convective cloud is not as good as that of cloud radar. Based on their observations, characteristics of vertical structure of deep convective cloud near Naqu weather station are as follows.
(1) The deep convective cloud near Naqu weather station is thick, with the tops of cloud and precipitation both reaching 16 km ASL and over. The C-FMCW observations show that the 15 dBZ-echo top could be up to about 16 km, which is consistent with the TRMM observations (Fig. 6c); cloud radar observations indicate that the echo top of –35 dBZ could reach up to 17 km ASL, which is consistent with the top of deep convective cloud to the west of Naqu weather station observed by the CloudSat (Fig. 5c). During 1342–1405 BT, the KA-band millimeter wave cloud radar observed that the cloud bottom was just 2 km above the ground or close to ground, while the tops of echoes of –30, 0, and 10 dBZ were above 16, 13, and 12 km ASL, respectively. According to Luo et al. (2011), based on the CloudSat observations, a deep convective core is defined if the cloud bottom is below 3 km above the ground and the echo tops of –30, 0, and 10 dBZ reach above 12, 11, and 9 km ASL, respectively. Thus, it can be inferred that the convective cloud near Naqu weather station contains deep convective cores.
(2) Radar reflectivity factor of this deep convective cloud increased downward below the zero-degree level, indicating that ice-phase particles grew through deposition and aggregation during the falling process. Furthermore, the radar reflectivity factor increased rapidly with the value of 35–40 dBZ above the zero-degree level to greater than 45 dBZ below the zero-degree level (Fig. 7a), suggesting nonnegligible effect of melting of solid particles after they fell below the zero-degree level. During 1403–1413 BT, radar reflectivity of C-FMCW radar was larger than 45 dBZ near surface (within 1 km above ground), implying that there were large precipitation particles near the ground. This explained the phenomenon that graupel particles with a 4-mm diameter at around 1400 BT fell down to the ground in the ground observation record. Moreover, surface observations support the above observations, as the disdrometer deployed near the C-FMCW radar also observed large precipitation particles with a maximum diameter of 5 mm and precipitation intensity up to 25 mm h–1 (Fig. 7d), and surface rain gauge also observed maximum 10-min accumulative precipitation of 1.4 mm (Fig. 3d).
(3) In the decaying stage of convection (after 1430 BT), an obvious bright band appeared at around 5.5 km ASL (i.e., 1 km above the ground level) (Figs. 7a, 8a). The bright band was below the zero-degree level with a thickness of 200–400 m (Fig. 7a). The linear depolarization ratio (LDR) observed by the KA-band millimeter wave cloud radar revealed some information of particles phase near the bright band: within the bright band, LDR was mostly larger than –22, indicating existence of wet snow and graupel particles within the bright band (Fig. 8d); LDR was smaller than –26 below the bright band, suggesting that raindrops were dominant there (Figs. 8b, d).
The vertical profiles of radar reflectivity detected by TRMM PR and C-FMCW at 19 seconds past 1418 BT 9 July 2014 are shown in Fig. 9. Firstly, it can be found that they were consistent in the upper layer of 11–13 km ASL. Since the PR performs measurement from the top downward, its measurements at upper levels were relatively accurate. Therefore, the above results indicate that the observations of the C-FMCW radar were reliable at upper levels. Secondly, below 11 km ASL, the echoes observed by C-FMCW were much weaker than those by the PR, and their differences increased with height decreasing. The maximum difference of 7 dBZ appeared at 6 km ASL. At lower levels, the PR observations were prone to systematic errors due to influences of attenuation and/or ground clutter, while C-FMCW measures from the bottom upward, and its observations in the lower troposphere were more accurate compared to that of PR. Therefore, apparently the PR overestimated echo intensity below 11 km ASL. Feng et al. (2001) compared vertical distributions of radar reflectivity observed by ground-based X-band Doppler radar and TRMM PR. They found that the echo intensity observed by the TRMM PR was about 23 dB larger than that by the X-band Doppler radar. The above differences indicate that the systematic error of radar reflectivity factor in the PR observations should be further calibrated and examined.5 Vertical structure of convective cloud microphysics
Figure 10 displays vertical distributions of cloud microphysical parameters extracted from the CloudSat 2B-CWC-RO product. It is found that the convective cloud contained a very small amount of liquid water and was mainly presented as ice-phase cloud, which could be attributed to the high elevation of the TP. Zhao et al. (2014) also proposed that central TP is featured with large values of ice water path and dominated by ice-phase clouds. Overall, CloudSat observed four deep convective cores over 30°–34°N. The ice water content of these deep convective cores were more than 210 mg m–3, the ice particle number concentration exceeded 120 L–1, and the ice effective radius was larger than 60 μm in the layer between 8–16 km ASL, indicating abundant ice water content in the deep convective cores over the TP. However, despite the fact that ice water content was consistently larger than 210 mg m–3 in the layer between 8 and 16 km, the ice particle number concentration and effective radius were quite different between the upper and lower levels. For example, above (below) 10 km ASL, the ice particle number concentration basically was larger (smaller) than 280 (200) L–1, and the effective radius of ice particles was smaller (larger) than 120 μm, suggesting that small-size ice particles were abundant above 10 km ASL, while large ice crystals were concentrated below 10 km. This result is consistent with the conclusion of Li et al. (2012) that precipitable ice particles were largely concentrated at low levels based on the TRMM data.
By use of the retrieval method in Rosenfeld and Lensky (1998), vertical microphysical structures of deep convective clouds over the TP were analyzed based on the Aqua MODIS data. The deep convective pixels to the west of Naqu weather station (30°–34°N, 88°–91.6°E) represented a deep weak convective cloud in mature stage and the deep convective pixels near Naqu weather station (30°–34°N, 91.6°–94°E) represented a deep strong convective cloud in decaying stage, are selected for the present study. The T–Re curves were obtained by analyzing 15,049 samples of deep weak convective pixels and 10,928 samples of deep strong convective pixels, respectively. The T–Re curves and vertical distributions of numbers of samples of deep convection pixels are displayed in Fig. 11. It can be seen that the T–Re curves of the deep weak convective cloud in mature stage and the deep strong convective cloud in decaying stage shared some common features as described below. (1) Microphysical processes in deep strong and deep weak convective clouds both contained growth of mixed-phase and glaciation. The glaciation temperature was about –41°C for the deep weak convective cloud to the west of Naqu and –45°C for the deep strong convective cloud near Naqu, which were lower than the glaciation temperature of –37.5°C proposed by Rosenfeld and Woodley (2000). Since Yuan et al. (2010) pointed that the glaciation temperature was affected by geographical location and environmental conditions, the glaciation temperature of deep convective clouds over the TP might be related to the high elevation of TP. (2) Re in mixed-phase process increased rapidly with height, which involved two growing processes: one was mainly the rimming process below the level of –25°C (near Naqu weather station) or –29°C (to the west of Naqu) due to abundant supercooled water content; and the other was mainly aggregation and deposition processes above that level due to low supercooled water content. The latter process was accompanied by faster increase in particle effective radius.6 Conclusions
Characteristics of vertical structures of deep convective clouds over Naqu in the central TP during 1300–1600 BT 9 July 2014 in the TIPEX-III are analyzed in this paper based on multi-source satellite data from TRMM, CloudSat, and Aqua and radar data from ground-based vertically pointing radars (C-FMCW and KA-band millimeter wave cloud radar). Conclusions are as follows.
(1) The deep convection over Naqu consisted of two northeast–southwest oriented deep convective clouds: the deep strong convective cloud over/near Naqu weather station and the deep weak convective cloud located about 100 km to the west of Naqu weather station.
(2) Vertical distributions of radar reflectivity observed by TRMM PR and CloudSat indicated that the deep strong and weak convective clouds were both comprised of multiple isolated convective cells. For the convective cells, their horizontal scale were about 10–20 km, top heights were 15 km ASL, and thickness was around 10 km, suggesting that convective clouds over the TP had small horizontal scale but could reach high altitude in the vertical. The top height of 35 dBZ of the deep strong convective cloud was about 8–11 km, and the vertical thickness of the layer of over 35 dBZ was 3–6 km.
(3) The ground-based vertically-pointing radar observed the complete evolution process of the deep strong convective cloud. In the mature stage of the deep strong convective cloud, radar reflectivity factor increased downward below the zero-degree level, indicating that ice-phase particles grew through deposition and aggregation during the falling process. Furthermore, the radar reflectivity factor increased rapidly, suggesting nonnegligible effect of melting of solid particles after they fell below the zero-degree level. In the decaying stage of deep strong convective cloud, an obvious bright band appeared at around 5.5 km ASL (i.e., 1 km above the ground level), and the bright band was below the zero-degree level with a thickness of 200–400 m. The bright band of stratiform cloud over the TP was observed by the C-FMCW for the first time, which could not be detected by Doppler radar in the past Tibetan Plateau atmospheric science experiments (Uyeda et al., 2001) and the TRMM PR (Fu Y. F. et al., 2006).
(4) Comparison of radar reflectivity factors observed by the TRMM PR and C-FMCW indicates that they are consistent in the upper layer of 11–13 km ASL. Below 11 km ASL, radar reflectivity factors observed by the C-FMCW were smaller than by PR, and the maximum difference of 7 dBZ appeared at 6 km ASL, indicating that the PR overestimated radar reflectivity at lower levels. This is the first comparison between observations of TRMM and ground-based vertically pointing radars in the TP, which provides reference for the quality evaluation of TRMM PR data over the TP in the future.
(5) The deep convective cloud to the west of Naqu weather station observed by CloudSat was mainly ice-phased cloud. A huge amount of small ice particles was concentrated above 10 km ASL, while a relatively small amount of large-size ice crystals was concentrated below 10 km ASL.
(6) The T–Re curves retrieved from the Aqua MODIS indicated that the microphysical processes in both the deep strong and deep weak convective clouds included the mixed-phase process, during which Re increased with height, and the glaciated process, in which Re decreased with height. Re in mixed-phase process increased rapidly with height, which involved two growing processes: one was mainly rimming process below the level of –25°C (over Naqu weather station) or –29°C (to the west of Naqu weather station) due to abundant supercooled water content; the other was mainly aggregation and deposition processes above that level due to low supercooled water content. The latter process was accompanied by faster increase in particle effective radius.
Acknowledgments. We acknowledge the team of the Third Tibetan Plateau Atmospheric Science Experiment for providing the surface observations of cloud and precipitation. Many thanks go to Professor Zhijin Hu, Professor Xiaofeng Lou, and Dr. Jing Duan for offering great suggestions and comments for the present study.
|Chang, Y., and X. L. Guo, 2016: Characteristics of convective cloud and precipitation during summer time at Naqu over Tibetan Plateau. Chinese Sci. Bull., 61, 1706–1720. DOI:10.1360/N972015-01292|
|Chen, L. X., Y. K. Song, J. P. Liu, et al., 1999: On the diurnal variation of convection over Qinghai–Tibet Plateau during summer as revealed from meteorological satellite data. Acta Meteor. Sinica, 57, 549–560. DOI:10.11676/qxxb1999.053|
|Dai, J., X. Yu, G. H. Liu, et al., 2011: Satellite retrieval analysis on microphysical property of thunderstorm with light precipitation over the Qinghai–Tibet Plateau. Plateau Meteor., 30, 288–298.|
|Feng, J. M., L. P. Liu, Z. J. Wang, et al., 2001: Comparison of cloud observed by ground based Doppler radar with TRMM PR in Qinghai–Tibet Plateau, China. Plateau Meteor., 20, 345–353.|
|Flohn, H., 1968: Contributions to a meteorology of the Tibetan highlands. Atmospheric science paper, No. 130, Colorado State University, Fort Collins, 120 pp.|
|Fu, R., Y. L. Hu, J. S. Wright, et al., 2006: Short circuit of water vapor and polluted air to the global stratosphere by convective transport over the Tibetan Plateau. Proc. Natl. Acad. Sci. USA, 103, 5664–5669. DOI:10.1073/pnas.0601584103|
|Fu, Y. F., G. S. Liu, G. X. Wu, et al., 2006: Tower mast of precipitation over the central Tibetan Plateau summer. Geophys. Res. Lett, 33, L05802. DOI:10.1029/2005GL024713|
|Fu, Y. F., H. T. Li, and Y. Zi, 2007: Case study of precipitation cloud structure viewed by TRMM satellite in a valley of the Tibetan Plateau. Plateau Meteor, 26(1), 98–106.|
|Fu, Y. F., X. Pan, G. S. Liu, et al., 2016: Characteristics of precipitation based on cloud brightness temperatures and storm tops in summer Tibetan Plateau. Chinese J. Atmos. Sci., 40, 102–120. DOI:10.3878/j.issn.1006-9895.1507.15165|
|Jakob, C., and S. A. Klein, 1999: The role of vertically varying cloud fraction in the parametrization of microphysical processes in the ECMWF model. Quart. J. Roy. Meteor. Soc, 125, 941–965. DOI:10.1002/qj.49712555510|
|Jiang, J. X., X. K. Xiang, M. Z. Fan, 1996: The spatial and temporal distributions of severe mesoscale convective systems over Tibetan Plateau in summer. J. Appl. Meteor. Sci., 7, 473–478.|
|Kummerow, C., W. Barnes, T. Kozu, et al., 1998: The tropical rainfall measuring mission (TRMM) sensor package. J. Atmos. Oceanic Technol., 15, 809–817. DOI:10.1175/1520-0426(1998)015<0809:ttrmmt>2.0.co;2|
|Li, D., A. J. Bai, and S. J. Huang, 2012: Characteristic analysis of a severe convective weather over Tibetan Plateau based on TRMM data. Plateau Meteor., 31, 304–311.|
|Liu, J. J., and B. D. Chen, 2017: Cloud occurrence frequency and structure over the Qinghai–Tibetan plateau from CloudSat observation . Plateau Meteor., 36, 632–642. DOI:10.7522/j.issn.1000-0534.2017.00028|
|Liu, L. P., J. F. Zheng, Z. Ruan, et al., 2015: Comprehensive radar observations of clouds and precipitation over the Tibetan Plateau and preliminary analysis of cloud properties. J. Meteor. Res., 29, 546–561. DOI:10.1007/s13351-015-4208-6|
|Luo, Y. L., R. H. Zhang, W. M. Qian, et al., 2011: Intercomparison of deep convection over the Tibetan Plateau–Asian monsoon region and subtropical North America in boreal summer using CloudSat/CALIPOS data . J. Climate, 24, 2164–2177. DOI:10.1175/2010JCLI4032.1|
|Oye, D., and M. Case, 1995: REORDER: A Program for Gridding Radar Data—Installation and Use Manual for the UNIX Version. NCAR/ATD. Boulder, CO, USA, 30 pp.|
|Pan, X., and Y. F. Fu, 2015: Analysis on climatological characteristics of deep and shallow precipitation cloud in summer over Qinghai–Tibet Plateau. Plateau Meteor., 34, 1191–1203. DOI:10.7522/j.issn.1000-0534.2014.00112|
|Platnick, S., M. D. King, K. G. Meyer, et al., 2015: MODIS Cloud Optical Properties: User Guide for the Collection 6 Level-2 MOD06/MYD06 Product and Associated Level-3 Datasets. Version 1.0, NASA Goddard Space Flight Center, Greenbelt, MD, USA, 141 pp,|
|Qie, X. S., X. K. Wu, T. Yuan, et al., 2014: Comprehensive pattern of deep convective systems over the Tibetan Plateau–South Asian monsoon region based on TRMM data. J. Climate, 27, 6612–6626. DOI:10.1175/JCLI-D-14-00076.1|
|Randall, D. A., D. A. Dazlich, T. G. Corsetti, 1989: Interactions among radiation, convection, and large-scale dynamics in a general circulation model. J. Atmos. Sci., 46, 1943–1970. DOI:10.1175/1520-0469(1989)046<1943:IARCAL>2.0.CO;2|
|Rosenfeld, D., and I. M. Lensky, 1998: Satellite-based insights into precipitation formation processes in continental and maritime convective clouds. Bull. Amer. Meteor. Soc., 79, 2457–2476. DOI:10.1175/1520-0477(1998)079<2457:SBIIPF>2.0.CO;2|
|Rosenfeld, D., and W. L. Woodley, 2000: Deep convective clouds with sustained supercooled liquid water down to –37.5°C. Nature, 440–442. DOI:10.1038/35013030|
|Savtchenko, A., D. Ouzounov, S. Ahmad, et al., 2004: Terra and Aqua MODIS products available from NASA GES DAAC . Adv. Space Res., 34, 710–714. DOI:10.1016/j.asr.2004.03.012|
|Stephens, G. L., D. G. Vane, R. J. Boain, et al., 2002: The CloudSat Mission and the A-Train: A new dimension of space-based observations of clouds and precipitation . Bull. Amer. Meteor. Soc., 83, 1771–1790. DOI:10.1I75/BAMS-83-I2-I77I|
|Uyeda, H., H. Yamada, J. Horikomi, et al., 2001: Characteristics of convective clouds observed by a Doppler radar at Naqu on Tibetan Plateau during the GAME-Tibet IOP. J. Meteor. Soc. Japan, 79, 463–474. DOI:10.2151/jmsj.79.463|
|Wang, H., Y. L. Luo, R. H. Zhang, 2011: Analyzing seasonal variation of clouds over the Asian monsoon regions and the Tibetan Plateau region using CloudSat/CALIPSO data . Chinese J. Atmos. Sci., 35, 1117–1131.|
|Wang, H., Y. L. Luo, B. J. D. Jou, 2014: Initiation, maintenance, and properties of convection in an extreme rainfall event during SCMREX: Observational analysis. J. Geophys. Res. Atmos., 119, 13206–13232. DOI:10.1002/2014JD022339|
|Wang, J. H., and W. B. Rossow, 1998: Effects of cloud vertical structure on atmospheric circulation in the GISS GCM. J. Climate, 11, 3010–3029. DOI:10.1175/1520-0442(1998)011<3010:eocvso>2.0.co;2|
|Wang, S. H., Z. G. Han, Z. G. Yao, et al., 2011: An analysis of cloud types and macroscopic characteristics over China and its neighborhood based on the CloudSat data . Acta Meteor. Sinica, 69, 883–899.|
|Xu, W. X., 2013: Precipitation and convective characteristics of summer deep convection over East Asia observed by TRMM. Mon. Wea. Rev., 141, 1577–1592. DOI:10.1175/MWR-D-12-00177.1|
|Yasunari, T., and T. Miwa, 2006: Convective cloud systems over the Tibetan Plateau and their impact on meso-scale disturbances in the Meiyu/Baiu frontal zone—a case study in 1998. J. Meteor. Soc. Japan, 84, 783–803. DOI:10.2151/jmsj.84.783|
|Yuan, T. L., and Z. Q. Li, 2010: General macro- and microphysical properties of deep convective clouds as observed by MODIS. J. Climate, 23, 3457–3473. DOI:10.1175/2009jcli3136.1|
|Yuan, T. L., J. V. Martins, Z. Q. Li, et al., 2010: Estimating glaciation temperature of deep convective clouds with remote sensing data. Geophys. Res. Lett., 37, L08808. DOI:10.1029/2010GL042753|
|Zhao, P., and Y. Yuan, 2017: Characteristics of a plateau vortex precipitation event on 14 July 2014. J. Appl. Meteor. Sci., 28, 532–543. DOI:10.11898/1001-7313.20170502|
|Zhao, P., X. D. Xu, F. Chen, et al., 2018: The Third Atmospheric Scientific Experiment for understanding the earth–atmosphere coupled system over the Tibetan Plateau and its effects. Bull. Amer. Meteor. Soc., 99, 757–776. DOI:10.1175/BAMS-D-16-0050.1|
|Zhao, Y. F., D. H. Wang, and J. F. Yin, 2014: A study on cloud microphysical characteristics over the Tibetan Plateau using CloudSat data . J. Trop. Meteor., 30, 239–248. DOI:10.3969/j.issn.1004-4965.2014.02.005|