1. Introduction
Cumulus (Cu) play an important role in the earth’s radiation budget (Berg et al. 2013). It is suggested to be a leading factor in determining the cloud–climate feedback in climate models (Bony and Dufresne 2005; Bony et al. 2004). Shallow convection also mixes the boundary air with the free tropospheric air, contributing to the upward transport of heat, moisture, and momentum from the surface (Powell and Houze 2013; Xu and Rutledge 2014).
The Tibetan Plateau (TP; defined here as the Himalaya region where altitude is higher than 2 km) can significantly influence the atmospheric circulation in Asia and even over the globe. It is often considered as an atmospheric heat source in the summer because of its much higher surface temperature relative to surrounding regions at the same altitudes (Wu et al. 2015). It is also a crucial moisture source or a transfer station of moisture for East Asia (Zhang et al. 2013). The convection that is responsible for the Cu provides a mechanism for the TP surface heat to be redistributed in the atmosphere, which can impact the atmospheric circulation regionally and globally.
Satellite observations have shown that Cu prevails over the TP in the summer (Wang et al. 2014). To further study the role of Cu over the TP, first we should know the distribution and cloud characteristics of Cu, and we should know the connection between Cu and the unique topographic forcing and thermodynamical environment of the TP. Because of the small spatial scale of Cu and the lack of high-resolution observation, few studies have examined the distribution and variability of Cu over the TP and their physical causes. The purpose of this study is to characterize the climatology and variability of Cu over the TP and explain the physical mechanisms that cause them. The fine-resolution cloud-type product of CloudSat (Stephens et al. 2008) and Cloud–Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO; Winker et al. 2009) will be used in this study. Since the environmental conditions over the TP are very different from those in many other regions around the world where Cu prevails, this study could also provide a unique test case of shallow convection in climate models that can complement previously available cases used by the Global Energy and Water Cycle Experiment (GEWEX) Cloud System Study program (Stevens et al. 2001; Brown et al. 2002; vanZantan et al. 2011) and over other continents (e.g., Zhang and Klein 2013; Vogelmann et al. 2015).
This paper is organized as follows. Section 2 gives a description of the data and methodology. Section 3 presents the horizontal and vertical distribution of Cu over the TP, along with discussions of their formation mechanism and diurnal variation. The last section contains a summary and discussion.
2. Data and methodology
CloudSat and CALIPSO profile data from 15 June 2006 to 31 December 2010 are used in this study (all data can be downloaded at http://cloudsat.atmos.colostate.edu/data). The cloud profiling radar (CPR) on the CloudSat operates at 94 GHz. CPR can penetrate optically thick clouds to detect multilayer cloud systems, but its long wavelength limits its capability to detect midlevel supercooled water clouds with relatively small water droplets or cold ice clouds with low concentrations of small ice crystals. CPR measurements of clouds are also often contaminated by precipitation. Lidar can detect optically thin cirrus clouds that are missed by the radar, but lidar signals are strongly attenuated by the optically thick clouds (Wang et al. 2012). Combining lidar and radar measurements can provide better cloud detection and characterization because of their unique complementary capabilities. So the combined CloudSat CPR and CALIPSO lidar product (2B-CLDCLASS-lidar; Sassen and Wang 2008, 2012) is used in this study as the primary dataset and the CloudSat CPR product (2B-CLDCLASS; Wang and Sassen 2007) is used as supplement when no lidar data are available, which includes the year 2006 and 19.7% of all cases in years 2007–10.
The 2B-CLDCLASS-lidar and 2B-CLDCLASS products are the level-2 cloud scenario classification data with cloud-type information. There are approximate 37 080 rays (profiles) along each satellite orbit with 125 bins (boxes) (the top height is 30 km) in the vertical for each ray. Only the lowest 80 bins (the top height is 19.2 km) were extracted in our dataset because nearly all clouds exist within these levels. The sampling resolution is 1.3 km in cross track, 1.1 km in along track, and 0.24 km in the vertical for CPR. CALIPSO lidar has better horizontal resolution (0.33 km) and vertical resolution (30 m) than that of CPR. In the combined product, lidar-attenuated backscattering coefficients at different horizontal and vertical resolutions are averaged to the CPR footprint. The box of 1.3-km wide by 1.1-km long by 0.24 km in height with significant cloud mask values equal to or larger than 20 (range is 0–40) is determined as cloudy in this study. The cloud is classified into eight types: high clouds (HC), altostratus (As), altocumulus (Ac), nimbostratus (Ns), stratocumulus (Sc), stratus (St), cumulus (Cu), and deep convection (DC). It has up to 10 cloud layers in one profile, and provides information of cloud type, cloud cover (clouds can occupy part of a box, but the amount generally is 0 or 100%), and cloud-base and cloud-top heights in each cloud layer (Sassen and Wang 2012; Subrahmanyam and Kumar 2013).
The Cu, representing cumulus congestus and fair-weather cumulus in the 2B-CLDCLASS-lidar product, is defined as isolated clouds with base height 0–3 km above the ground, horizontal scale 1–10 km, vertical extent shallow or moderate, liquid water path larger than zero, and with or without drizzle and snow (Wang and Sassen 2007). In the classification algorithm, the inputs include cloud properties from CloudSat and CALIPSO lidar, MODIS information, ECMWF temperature profiles, and other information such as topography. A combined rule-based and fuzzy logic–based approach is used to classify clouds. A brief description is as follows: first, the combined radar and lidar cloud masks are used to find a cloud cluster according to their persistence in the horizontal and vertical directions. A minimum horizontal extent for a cluster is required to identify horizontally broken, but vertically similar cloud fields. Once a cloud cluster is found, cloud height, temperature, and maximum radar reflectivity, as well as the occurrence of precipitation, are derived. Then the cluster mean properties as well as spatial inhomogeneities in cloud-top height and lidar and radar maximum signals are used as input to a fuzzy classifier to classify the cluster into cloud types with an assigned confidence level. The membership functions for the fuzzy sets can be seen in the document by Wang et al. (2012). Land and ocean differences, as well as polar, tropics, and midlatitude differences are also considered.
Because the CPR surface echo contaminates the first three or four bins (~1.0 km) above the surface, small Cu (e.g., fair-weather Cu) are likely to be underrepresented in the CloudSat data near the surface. Many warm Cu also lack hydrometeors of sufficient size to enable millimeter-wave radar detection (Sassen and Wang 2008). Product uncertainties are extensively discussed in Kahn et al. (2008) and in the algorithm theoretical basis online document. An evaluation of the appropriateness of the cloud classification algorithm specifically for clouds over the TP is beyond the scope of this paper due to lack of surface observations. However, the uniqueness of Cu in the satellite data is consistent with visual observations in this region.
CloudSat and CALIPSO have at most 14 orbits each day around the globe. However, only two orbits pass through the TP each day at two fixed local solar times: one at approximately 1300 during the daytime and another at approximately 0200 at night. The A-train satellites pass through the same region only about every 14 days and may never pass some regions. To increase samples in specific regions and explore the climatic characteristics of clouds, the orbit data are spatially and temporally aggregated to monthly gridded data for this study. Monthly data are constructed by merging all profiles in one month in a given grid box. The daytime profiles and nighttime profiles are separately aggregated.
Because the profile number is grid-size dependent, four sets of grids at the resolutions of 0.5° × 0.5°, 1° × 1°, 1.5° × 1.5°, and 2° × 2° are tested to see which size is better based on the amount of profile samples in the grid boxes and uniformity of the profile numbers across different grid boxes. Figure 1a–d show the spatial distribution of the number of profiles for these four sets of grids in July 2007. We take the 1° × 1° grid size as the desired resolution because its grid boxes have more samples than those of the 0.5° × 0.5° grid and its spatial distribution is more uniform than that of the 1.5° × 1.5° grid because of the uneven orbit number (see Fig. 1c) over the TP. The 1° × 1° grid size also matches the resolution of many contemporary climate models, so that data can be compared with model results without interpolation. For the vertical resolution, 0.24 km is selected as in the original data. Monthly three-dimensional cloud data from June 2006 to December 2010 are thus constructed in this work. Unless otherwise specified, the result in this study is for June–August (JJA) averaged from 2006 to 2010.
The number density of profiles with grid sizes of (a) 0.5° × 0.5°, (b) 1.0° × 1.0°, (c) 1.5° × 1.5°, and (d) 2.0° × 2.0° in July 2007.
Citation: Journal of Climate 29, 3; 10.1175/JCLI-D-15-0492.1
The frequency of each cloud type is calculated at every vertical level as the ratio of cloud-type number to the total profile number in given grid, and the total cloud frequency is calculated as the ratio of cloudy profile number to the total profile number. Although clouds can occupy a partial box, cloud cover is generally 100% when clouds exist. So cloud frequency is very similar to cloud cover. Hereafter, cloud cover and cloud frequency are used interchangeably, with cloud cover referring to cloud spatial distribution, and frequency referring to temporal occurrence. The two- and three-dimensional cloud distributions are derived from the gridded data while statistics of seasonal variation and diurnal variations are computed from the original profile data.
Reanalysis data from ERA-Interim (Dee et al. 2011) with 1° × 1° resolution from June 2006 to December 2010 are also used in this study to describe the large-scale atmospheric circulation. ERA-Interim data are 6 hourly (0000, 0600, 1200, and 1800 UTC each day) data. To match the CloudSat transit time over the TP, only 0600 UTC (1400 local time) and 1800 UTC (0200 local time) data are extracted to construct the monthly atmospheric fields.
3. Total clouds and cumulus over the Tibetan Plateau
a. Cloud types and total amount
Cloud frequency over the TP is 73.6% in JJA. The five dominant cloud types in terms of occurrence frequency are Sc, As, Cu, HC, and Ac, each with cloud frequency in the range between 15% and 21% (Fig. 2a).
(a) Cloud frequencies and (b) cloud-event frequencies over TP in JJA. Total cloud frequencies (Cld) are represented by black histograms. Frequencies of eight cloud types are represented by colored histograms with their individual abbreviations on the top.
Citation: Journal of Climate 29, 3; 10.1175/JCLI-D-15-0492.1
The typical size of cumulus is 1–4 km in diameter, while the sizes of stratiform clouds can be on the order of 50–1000 km. Because of this difference in the horizontal sizes, as well as the difference in the cloud lifetimes, we also examined the cloud events in addition to the cloud frequency from profile numbers. A cloud event is defined as follows: in 45 consecutive profiles along the satellite track (~50 km, the same as minimum stratiform cloud scale defined in 2B-CLDCLASS), if a specific cloud type occurs at any level, it is defined as one event of this cloud type. For example, if there are one or more profiles of Cu among 45 consecutive profiles, the 45 profiles are considered to have one Cu event. On the other hand, even if all 45 consecutive profiles contain stratus clouds, they are counted as one event. From the model-grid point of view, the number difference between Cu frequency and Cu event frequency also give us evidence of how often the subgrid clouds should be considered.
The cloud-event frequency thus defined is shown in Fig. 2b for different cloud types. The Cu cloud–event frequency is 54% with 23 520 events among a total of 43 409 events during our study period. This can be compared to the cloud profile frequency 18% of 250 853 Cu profiles among the total of 1 431 472 profiles over the TP in JJA from 2006 to 2010. Figure 2b shows that Cu is the dominant cloud event over the TP in JJA. Over half of the time, Cu is present within a scene of about 50 km.
We next examine the spatial distribution of clouds over the TP and its surrounding area by using the gridded data. The total cloud cover is shown in Fig. 3a. Cloud amount over the TP as outlined by the 2-km topography height is more than the surrounding regions except for the southwest corner (30°–35°N, 65°–70°E) of the TP. There is a maximum belt of cloud amount along the southern steep slope of TP where the sky is overcast almost all the time in JJA.
JJA-averaged spatial distribution of (a) total cloud cover derived from CloudSat data, (b) wind vectors at 10-m height above the ground, (c) vertical velocity at 500 hPa, and (d) relative humidity at 2-m height above the ground derived from ERA-Interim data. The thick red contour in all panels denotes the 2-km topography height referred to as the TP. Red contours in (b) are topography height with a 0.5-km interval. Reference vector in (b) is 3 m s−1. Negative (positive) values in (c) denote upward (downward) motion.
Citation: Journal of Climate 29, 3; 10.1175/JCLI-D-15-0492.1
In the summer, the TP is in the confluence zone of both the Indian summer monsoon and the midlatitude circulations. The TP as a heat source is at least partially responsible for the convergence of surface winds toward the plateau. Near-surface winds are directed toward the plateau from all directions (Fig. 3b), associated with which are the broad upward motion over the TP and maximum upward motion near the steep slopes (Fig. 3c).
The southerly surface flow converging toward the TP also transports moisture to the plateau. The upward motion and moisture provide the condition of overall larger cloud amount over the TP than its surrounding regions. Figure 3d shows the air relative humidity at 2-m height. Except for the northern slope and the southwest corner of the TP where air originates from dry regions or has downhill motion, the surface relative humidity over the TP is comparable to that in eastern China where the topography is much lower. The surface relative humidity is much larger over the southern TP. The spatial distribution of the surface relative humidity correlates well with the spatial variation of the total cloud amount over the TP. This implies that surface-based processes play a major role in explaining the total cloud distribution over the TP.
b. Cumulus clouds
The topographic map of the TP is shown in Fig. 4a and the spatial distribution of the cumulus cloud cover in the summer is shown in Fig. 4b. A larger amount is found over the TP than in its surrounding regions. The maximum in the western TP is much larger than the maximum in the south. The south slope has relatively smaller Cu than in the western TP is because more deep convection (not shown) occurs in this region in the summer. The contrast of cloud amount between the TP and the surrounding regions is much more distinct for Cu that occurs alone (here “alone” means that Cu is not accompanied by other cloud types) (Fig. 4c). As mentioned before, the average Cu frequency over the TP is 18% when Cu occurs. The frequency is 10% when Cu occurs alone. It means that in 55% of the cases Cu occurs alone over the TP whereas in 45% of the cases it is accompanied by other cloud types over the surrounding regions of the TP.
(a) The topography of the Tibetan Plateau. Spatial distribution of cloud cover for (b) Cu and (c) Cu alone in JJA. (d) Spatial distribution of Cu cloud cover in DJF. In (c), the red star labeled “A” indicates representative location in eastern China and the red star labeled “B” indicates representative location in the middle of the TP. The gray boxes in (b),(c), and (d) are missing data.
Citation: Journal of Climate 29, 3; 10.1175/JCLI-D-15-0492.1
If the cloud fraction of Cu in December, January, and February (DJF) (Fig. 4d) is examined, it is found the cloud amount in the west and south is only a little smaller than that in the summer in same regions, but there is nearly no Cu over the northern TP in winter. This implies that surface wind convergence and upward motion forced by topography are also important for Cu incidence in the western and southern TP by providing the necessary moisture and trigger of convection.
The black curve in Fig. 5a shows the vertical profile of Tq at representative locations in eastern China labeled as “A” in Fig. 4c at 28.5°N, 115.5°E. The corresponding vertical profile of Tqs is shown as the red curve in Fig. 5a. Their corresponding standard deviations with regard to time are indicated by the dashed lines. The variable Tq of an air parcel rising from the surface follows the black vertical straight line in Figs. 5a and 5b. The two intercepting points of this straight line with the red curve represent the levels of free convection and neutral buoyancy at the top of the convection. The altitude range where the black vertical line is at the right of the red curve is the range of convective instability. Figure 5a indicates that the atmosphere in eastern China is convectively unstable, with the level of free convection at 3.4 km and the cloud top over 10 km. This is consistent with the dominance of deep convective clouds in this region in the summer (not shown).
(a) Vertical profiles of Tq and Tqs at location A in Fig. 4c derived from ERA-Interim data in JJA. Black solid curve is Tq and red solid curve is Tqs. The corresponding standard deviations are dashed curves with short horizontal lines. Straight vertical line represents the trajectory of an undiluted air parcel–conserving surface Tq. The first (lowest) intercepting point marked by a black asterisk on the vertical line with the red curve represents the level of free convection, the second (highest) intercepting point represents the neutral buoyancy height. (b) As in (a), but for the location B in Fig. 4c. (c) As in (b), but the temperature above the TP surface is replaced by the temperature at the same altitude in eastern China. (d) As in (b), but the specific humidity above the TP surface is replaced by the specific humidity at the same altitude in eastern China.
Citation: Journal of Climate 29, 3; 10.1175/JCLI-D-15-0492.1
We now turn to a representative location in the middle of the TP, location “B” in Fig. 4c at 33.5°N, 93.5°E, to investigate why Cu is more favored there. The profiles of Tq and Tqs over the TP are shown in Fig. 5b. It is seen that above the altitude of about 8 km, the atmospheric temperature (represented by the red curve of Tqs) is not very different from that in eastern China. Below 8 km above the TP surface, however, air temperature is higher than that in eastern China. This leads to a larger vertical curvature in the Tqs over the TP. Likewise, above 8 km, Tq over the TP is not very different from that in eastern China. Below 8 km above the TP surface, Tq is much larger than that in eastern China at the same altitude. The difference between Tqs and Tq over the TP is much smaller than that in eastern China. This difference represents the relative humidity of the atmosphere. The larger surface Tq is, therefore, due to both higher temperature and larger specific humidity over the TP. The vertical shapes of these profiles in Fig. 5b relative to those in Fig. 5a indicate two unique atmospheric conditions over the TP for cumulus. First, the atmospheric instability over the TP is weaker than that in eastern China because the environmental atmospheric temperature is warmer below 8 km above the TP surface than that in eastern China at the same altitude due to the mixing in the atmospheric boundary layer. Second, the levels of free convection and convective top above the ground are lower over the TP than those in eastern China. The lower altitude of the level of free convectivion makes the shallow convection easily triggered by the rugged TP terrain. The lower convective top limits the occurrence of deep convection.
The above two conditions are all related to the higher surface temperature and humidity near the TP surface than those at the same altitude in eastern China. To understand the relative roles of temperature and humidity, in Figs. 5c and 5d, we plot the same profiles as in Fig. 5b except that the temperature and humidity over the TP is replaced, respectively, by using the temperature and humidity at the same altitude in eastern China. Figures 5c and 5d show that without the joint contributions of temperature and humidity, there would be no convective instability over the TP. Humidity plays a larger role since without the contribution of water vapor (Fig. 5d), the atmosphere would be very stable. Therefore, it is the combined impact of TP temperature and humidity that created the atmospheric vertical instability condition that favors shallow convection. The large-scale surface wind convergence toward the TP provided the necessary dynamic lifting to trigger the convection, as shown in Figs. 3b and 3c.
The Tq and Tqs at the two selected locations represent the broad difference of atmospheric conditions between the TP and its surrounding regions. This is shown in Figs. 6a and 6b, respectively, for Tq and Tqs at 500 hPa. Larger values of Tq and Tqs are seen over the TP than in other regions. The large Tq makes the atmosphere over the TP convectively unstable, while large Tqs makes the atmosphere only weakly unstable with a shallow convective layer with convective clouds. In Figs. 6c and 6d, we use zonal-mean T and q to plot the Tq, it is seen that both temperature and water vapor contribute to the maximum Tq over the TP, but moisture contributes more. This is consistent with the results described previously at the two selected locations.
Spatial distribution of daytime-averaged (a) Tq and (b) Tqs at 500 hPa in JJA derived from ERA-Interim data. (c) As in (a), but the temperature is replaced by the zonal-mean temperature in Eqs. (1) and (2). (d) As in (a), but the specific humidity is replaced by the zonal-mean specific humidity in Eq. (1).
Citation: Journal of Climate 29, 3; 10.1175/JCLI-D-15-0492.1
c. Properties of cumulus clouds over the Tibet Plateau
The Cu cloud-base height relative to the surface over the TP is about 1.0 km, slightly lower than those in its surrounding regions, as shown in Fig. 7a. The cloud-base height is the lowest in the western and southern TP. On the other hand, the cloud-top height of Cu relative to the surface is much higher over the TP than those in most of its surrounding regions (Fig. 7b). As a result, Cu over the TP is generally thicker than that in its surrounding regions.
Spatial distribution of (a) Cu cloud-base height and (b) Cu cloud-top height derived from CloudSat data in JJA, (c) LCL calculated by Eq. (3), and (d) PBL height in JJA derived from ERA-Interim data. The heights in all panels are relative heights to the surface. The gray boxes in (a) and (b) are missing data.
Citation: Journal of Climate 29, 3; 10.1175/JCLI-D-15-0492.1
The planetary boundary layer (PBL) height rising above the LCL may also be a mechanism of Cu. When comparing the PBL height (Fig. 7d) to the LCL over the TP by using the ERA-Interim data, it is found that the PBL height is generally a little bit lower than the LCL; therefore, the shallow convection is primarily due to convective instability. However, it is actually the tails of the distribution of LCL and the PBL height that are important in the initiation of shallow cumulus. Because the time resolution of ERA-Interim data is too coarse to reflect this variability, the relationship between PBL and the trigger of Cu warrants further study.
The contrasts of the cloud-base and cloud-top heights over the TP with those in the surrounding regions can be more clearly seen in the vertical cross sections of Cu cloud frequency at two representative latitudes 28.5° and 35.5°N (Figs. 8a and 8b). The Cu over the TP is thicker than in other regions. The cloud-top level over the TP is consistent with the height of the level of neutral buoyancy in Fig. 5b. Therefore, Cu is the dominant form of convection over the TP. In the surrounding regions such as eastern China (Fig. 5a), shallow convection occurs when deep convection is suppressed by large-scale subsidence under relative dry conditions, which correspond to lower Cu tops and higher Cu bases.
Vertical distribution of Cu frequency at the latitude cross sections of (a) 28.5°N and (b) 35.5°N. The hatched area represents topography.
Citation: Journal of Climate 29, 3; 10.1175/JCLI-D-15-0492.1
d. Diurnal variation
When the Cu daytime profiles and nighttime profiles are separated, we found 58 038 total profiles of Cu over the TP during the daytime and 38 684 profiles at night in JJA from 2006 to 2010 from the original profile dataset. Cloud frequency during the daytime is much larger than at night. If only the cases of Cu alone are considered, the profile number during the daytime is 34 524 versus 18 277 at night. Figures 9a and 9b show the spatial distribution of cloud frequency of Cu during the daytime and at night, respectively. Even though the satellite has different trajectories during the daytime and at night, so the frequency difference between daytime and nighttime for specific locations cannot be calculated, the spatial distribution patterns still reveal much larger cloud frequency during the daytime than at night. Surface heating from solar radiation, therefore, plays an important role in the Cu formation.
Spatial distribution of Cu cloud frequency during the (a) daytime and (b) nighttime. (c) Frequency profiles of Cu alone over the TP during the daytime (black solid line) and nighttime (blue dashed line). The gray boxes in (a),(b) are missing data.
Citation: Journal of Climate 29, 3; 10.1175/JCLI-D-15-0492.1
Figure 9c shows the composite vertical profile of occurrence frequency of Cu alone over the TP during daytime and nighttime. Not only is the Cu frequency larger during the daytime, the peak altitude is also higher. The maximum occurrence frequency is 9% at about 1.2-km height during the daytime, while it is only 5% at night, and the most frequent height is a little bit lower. The frequencies above 3 km become much smaller. The daytime Cu cloud base is at 1.07 km relative to the surface, whereas it is at 0.92 km at night. All these are consistent with the surface thermodynamic control of the vertical structure of the atmospheric conditions over the TP.
Even during the summer nights, the temperature near the TP surface is larger than its surrounding areas at the same altitude. Therefore, the moist static energy over the TP is still larger than in surrounding areas at the same altitude at night. The wind convergence from all directions and mountain breeze can also provide the necessary trigger of Cu formation in the western and southern TP.
4. Conclusions and discussion
The cumulus (Cu) transports heat and moisture from the boundary layer to the free atmosphere. We analyzed the climatic characteristics of Cu over the Tibetan Plateau based on the CloudSat–CALIPSO products. The conclusions are as follows:
The occurrence frequency of Cu over the TP is 18% and of Cu alone, it is 10% in JJA. The cloud-event frequency, defined as Cu occurrence within 50 km, is 54%. These frequencies are much larger than the corresponding values in the surrounding regions.
The ubiquitous Cu over the northern TP is caused by the higher air temperature and larger relative humidity above the TP surface than those in the surrounding regions at the same altitude. The favorable condition of shallow convection over the TP can be explained by using the vertical structures of the atmospheric moist static energy that is characterized by weak instability, a shallow layer of instability, and lower altitudes for the level of free convection. In addition, large-scale surface convergence of winds provides the necessary triggering of convection.
The frequency of Cu over the TP exhibits distinctive diurnal variations, with maxima during the daytime, which is consistent with the surface thermodynamic control of Cu.
In the summer, the Tibetan Plateau is a heat source of the atmosphere. The surface heat from solar radiation is ultimately transported from the surface to the free atmosphere through shallow convection. Results presented in this study could offer an observational test of climate models by next simulating shallow convection. The fidelity of a model in simulating cumulus will be important for it to correctly simulate the vertical redistribution of energy and moisture (vanZanten et al. 2011) and, therefore, the summer circulation in Asia. Because of the large frequency of Cu occurrence over the TP, cumulus also directly modulates the surface radiation budget by shielding the surface from solar radiation, which impacts the energy balance of glaciers that supply water sources to many rivers in Asia.
Acknowledgments
We thank the three anonymous reviewers whose comments have helped us to improve the paper. This research is supported by the Major National Basic Research Program of China (973 Program) on Global Change under Grant 2010CB951800, the China R&D Special Fund for Public Welfare Industry (meteorology: GYHY 201306068), the National Natural Science Foundation of China (Grant 41475069), and the State Scholarship Fund from China Scholarship Council. Additional support is provided by the Biological and Environmental Research Division in the Office of Sciences of the U.S. Department of Energy (DOE), and by the National Science Foundation, to Stony Brook University.
REFERENCES
Arakawa, A., and W. H. Schubert, 1974: Interaction of a cumulus cloud ensemble with the large-scale environment, Part I. J. Atmos. Sci., 31, 674–701, doi:10.1175/1520-0469(1974)031<0674:IOACCE>2.0.CO;2.
Berg, L. K., W. I. Gustafson, E. I. Kassianov, and L. Deng, 2013: Evaluation of a modified scheme for shallow convection: Implementation of CuP and case studies. Mon. Wea. Rev., 141, 134–147, doi:10.1175/MWR-D-12-00136.1.
Bony, S., and J. L. Dufresne, 2005: Marine boundary layer clouds at the heart of tropical cloud feedback uncertainties in climate models. Geophys. Res. Lett., 32, L20806, doi:10.1029/2005GL023851.
Bony, S., J. L. Dufresne, H. Le Treut, J.-J. Morcrette, and C. Senior, 2004: On dynamic and thermodynamic components of cloud changes. Climate Dyn., 22, 71–86, doi:10.1007/s00382-003-0369-6.
Brown, A., and Coauthors, 2002: Large-eddy simulation of the diurnal cycle of shallow cumulus convection over land. Quart. J. Roy. Meteor. Soc., 128, 1075–1093, doi:10.1256/003590002320373210.
Dee, D. P., and Coauthors, 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation systems. Quart. J. Roy. Meteor. Soc., 137, 553–597, doi:10.1002/qj.828.
Kahn, B. H., and Coauthors, 2008: Cloud type comparisons of AIRS, CloudSat, and CALIPSO cloud height and amount. Atmos. Chem. Phys., 8, 1231–1248, doi:10.5194/acp-8-1231-2008.
Lawrence, M. G., 2005: The relationship between relative humidity and the dew point temperature in moist air: A simple conversion and applications. Bull. Amer. Meteor. Soc., 86, 225–233, doi:10.1175/BAMS-86-2-225.
Powell, S. W., and R. A. Houze Jr., 2013: The cloud population and onset of the Madden–Julian oscillation over the Indian Ocean during DYNAMO-AMIE. J. Geophys. Res. Atmos., 118, 11 979–11 995, doi:10.1002/2013JD020421.
Sassen, K., and Z. Wang, 2008: Classifying clouds around the globe with the CloudSat radar: 1-year of results. Geophys. Res. Lett., 35, L04805, doi:10.1029/2007GL032591.
Sassen, K., and Z. Wang, 2012: The clouds of the middle troposphere: Composition, radiative impact, and global distribution. Surv. Geophys., 33, 677–691, doi:10.1007/s10712-011-9163-x.
Stephens, G. L., and Coauthors, 2008: CloudSat mission: Performance and early science after the first year of operation. J. Geophys. Res., 113, D00A18, doi:10.1029/2008JD009982.
Stevens, B., and Coauthors, 2001: Simulations of trade wind cumuli under a strong inversion. J. Atmos. Sci., 58, 1870–1891, doi:10.1175/1520-0469(2001)058<1870:SOTWCU>2.0.CO;2.
Subrahmanyam, K. V., and K. K. Kumar, 2013: CloudSat observations of cloud-type distribution over the Indian summer monsoon region. Ann. Geophys., 31, 1155–1162, doi:10.5194/angeo-31-1155-2013.
vanZanten, M. C., and Coauthors, 2011: Controls on precipitation and cloudiness in simulations of trade-wind cumulus as observed during RICO. J. Adv. Model. Earth Syst., 3, M06001, doi:10.1029/2011MS000056.
Vogelmann, A. M., and Coauthors, 2015: RACORO continental boundary layer cloud investigations: 1. Case study development and ensemble large-scale forcings. J. Geophys. Res. Atmos., 120, 5962–5992, doi:10.1002/2014JD022713.
Wang, M. Y., J. Gu, R. Yang, L. Zeng, and S. Wang, 2014: Comparison of cloud type and frequency over China from surface, FY-2E, and CloudSat observations. Remote Sensing of the Atmosphere, Clouds, and Precipitation, E. Im, S. Yang, and P. Zhang, Eds., International Society for Optical Engineering (SPIE Proceedings, Vol. 9259), doi:10.1117/12.2069110.
Wang, Z., and K. Sassen, 2007: Level 2 cloud scenario classification product process description and interface control document, version 5.0. Colorado State University, 50 pp. [Available online at http://www.cloudsat.cira.colostate.edu.]
Wang, Z., D. Vane, , G. Stephens, and D. Reinke, 2012: Level 2 combined radar and lidar cloud scenario classification product process description and interface control document, version 1.0. Colorado State University, 61 pp. [Available online at http://www.cloudsat.cira.colostate.edu.]
Winker, D. M., M. A. Vaughan, A. H. Omar, Y. Hu, K. A. Powell, Z. Liu, W. H. Hunt, and S. A. Young, 2009: Overview of the CALIPSO mission and CALIOP data processing algorithms. J. Atmos. Oceanic Technol., 26, 2310–2323, doi:10.1175/2009JTECHA1281.1.
Wu, G., and Coauthors, 2015: Tibetan Plateau climate dynamics: Recent research progress and outlook. Natl. Sci. Rev., 2, 100–116, doi:10.1093/nsr/nwu045.
Xu, W., and S. A. Rutledge, 2014: Convective characteristics of the Madden–Julian oscillation over the central Indian Ocean observed by shipborne radar during DYNAMO. J. Atmos. Sci., 71, 2859–2877, doi:10.1175/JAS-D-13-0372.1.
Zhang, Y., and S. A. Klein, 2013: Factors controlling the vertical extent of fair-weather shallow cumulus clouds over land: Investigation of diurnal-cycle observations collected at the ARM Southern Great Plains site. J. Atmos. Sci., 70, 1297–1315, doi:10.1175/JAS-D-12-0131.1.
Zhang, Y., D. Wang, P. Zhai, G. Gu, and J. He, 2013: Spatial distributions and seasonal variations of tropospheric water vapor content over the Tibetan Plateau. J. Climate, 26, 5637–5654, doi:10.1175/JCLI-D-12-00574.1.
