MAY 2016 KUMAR AND BHAT 1277

Vertical Profiles of Radar Reflectivity Factor in Intense Convective

Clouds in the Tropics

SHAILENDRA KUMAR AND G. S. BHAT

Centre for Atmospheric and Oceanic Sciences, Indian Institute of Science, Bangalore, India

(Manuscript received 25 April 2015, in final form 29 February 2016)

ABSTRACT

This study is based on the analysis of 10 years of data for radar reflectivity factor Ze as derived from the
TRMM Precipitation Radar (PR) measurements. The vertical structure of active convective clouds at the PR
pixel scale has been extracted by defining two types of convective cells. The first one is cumulonimbus tower
(CbT), which contains Ze $ 20 dBZ at 12-km altitude and is at least 9 km deep. The other is intense convective
cloud (ICC), which belongs to the top 5% of the population of the Ze distribution at a prescribed reference
height. Here two reference heights (3 and 8 km) have been chosen. Regional differences in the vertical
structure of convective cells have been explored by considering 16 locations distributed across the tropics and
two locations in the subtropics. The choice of oceanic locations is based on the sea surface temperature; that of
the land locations is based on propensity for intense convection. One of the main findings of the study is the close
similarity in the average vertical profiles of CbTs and ICCs in the mid- and lower troposphere across the ocean
basins whereas differences over land areas are larger and depend on the selected reference height. The foothills
of the western Himalaya, southeastern South America, and the Indo-Gangetic Plain contain the most intense
CbTs; equatorial Africa, the foothills of the western Himalaya, and equatorial South America contain the most
intense ICCs. Close similarity among the oceanic profiles suggests that the development of vigorous convective
cells over warm oceans is similar and that understanding gained in one region is extendable to other areas.

1. Introduction instrumented aircraft (e.g., Heymsfield et al. 2010,

hereinafter H10), with weather radars (e.g., Zipser and
The importance of cumulonimbus (Cb) clouds in the
Lutz 1994, hereinafter ZL94), and, in recent years,
energy balance of the tropical atmosphere has long been
with the Precipitation Radar (PR) on board the
recognized (e.g., Riehl and Malkus 1958). Even in a
Tropical Rainfall Measuring Mission (TRMM) satellite
mesoscale convective system (MCS), which is an orga-
(Kummerow et al. 1998) and a cloud-profiling radar on
nized form of Cb convection (e.g., Cotton and Anthes
board CloudSat (Stephens et al. 2002). In situ measure-
1989; Houze 1993), the vertical transport of water and
ments using instrumented aircraft are likely to be biased
energy between the atmospheric boundary layer and the
toward less intense or smaller clouds because of aircraft
upper troposphere is accomplished by Cb clouds. Our
safety considerations (H10). Hence, radar remote sensing
understanding of convective clouds is not commensu-
is the main observational approach available for investi-
rate with their role in weather and climate systems be-
gating intense convective clouds. Radar provides the
cause of a lack of observations on several key aspects
equivalent radar reflectivity factor (hereinafter Ze or
(e.g., vertical velocity, hydrometeor concentration, and
reflectivity), which is a proxy for hydrometeor concen-
the distribution of the latter between liquid and ice
tration (Houze 1993). The vertical distribution of Ze is
phases) in different parts of the globe. Studies on the
often used as a proxy for the intensity of convection
vertical structure of convective clouds are relatively few
because higher Ze in the mid- and upper troposphere
and are mainly based on measurements made with
corresponds to higher vertical velocities (Xu and Zipser
2012, hereinafter XZ12).
A majority of the studies that are based on the analysis
Corresponding author address: G. S. Bhat, Centre for Atmo-
spheric and Oceanic Sciences, Indian Institute of Science, Banga- of radar data are on cloud systems, for example, MCSs
lore 560012, India. (e.g., Houze 1993) and precipitation features (e.g.,
E-mail: bhat@caos.iisc.ernet.in Nesbitt et al. 2000), and the number of studies that

DOI: 10.1175/JAMC-D-15-0110.1

Ó 2016 American Meteorological Society

1278 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 55

address Cb clouds at the scale of a radar beam is very 2000, 2009; Masunaga et al. 2002). The sensitivity of the
limited (e.g., ZL94; Yuan and Qie 2008; Liu et al. 2012; PR is ;17 dBZ, and its pixel size is 4.3 km 3 4.3 km
Saikranthi et al. 2014; Bhat and Kumar 2015, hereinafter (;5 km 3 5 km after the TRMM orbit boost in August
BK15). ZL94 used radar data collected at different lo- of 2001) in the horizontal plane, with samples collected
cations (covering coastal, land, and oceanic areas), and every 0.25 km along the beam (vertical) direction
constructed the vertical profile of radar reflectivity (Kummerow et al. 1998). There are 80 vertical levels,
(VPRR). ZL94 showed that Ze rapidly decreases with and the height corresponds to the distance measured
height above the freezing level, with the rate of decrease along the radar beam from the point of intersection
being larger in oceanic systems than in tropical conti- between the beam and Earth’s ellipsoid and not to the
nental and midlatitude systems. In the convective por- local vertical height. The difference in the maximum
tion of MCSs, Ze increases toward the surface below the heights at nadir and extreme angles is ;0.85 km (BK15).
freezing level over oceanic areas, whereas the opposite Hence, corrections to the pixel height have been applied
behavior is observed in continental systems (ZL94). By (see BK15), and the remapped data are used in the
analyzing data measured over the South China Sea area, analysis. Other data include TRMM Microwave Imager
Yuan and Qie (2008) show that high Ze in the mixed- (TMI) sea surface temperature (SST; Wentz et al. 2000)
phase region promotes lightning activity. XZ12 show (the data were downloaded from ftp://ftp.remss.com/
that the vertical structure of deep convective clouds over tmi/) and TRMM precipitation product 3B42 (Huffman
continents and oceans and in monsoonal systems de- et al. 2007), for a 10-yr period (2001–10).
pends on the updrafts in the mixed-phase region and on The PR data products have several advantages, in-
microphysics rather than on the cloud depth or ice cluding that the same instrument provides Ze around the
depth. BK15 used PR data to study the vertical structure globe, that there is a relatively long time series, and that
of Cb clouds embedded in South Asian summer mon- the quality of the data is high. The 2A25 data volume is
soon systems. BK15 adopted the VPRR method of very large, and carrying out the analysis at every hori-
ZL94 and defined convective cells with reference to 3-, zontal grid point is a difficult task. Hence we have se-
8-, and 12-km heights. BK15 show that some of the con- lected 18 locations on the basis of the following
clusions on the regional differences in the vertical struc- considerations. The choice of oceanic locations is based
ture of Ze depend on the reference height selected. on SST because of its strong influence on deep convec-
Measurements made in tropical convective systems tion (e.g., Gadgil et al. 1984; Graham and Barnett 1987;
with a nadir-looking Doppler radar fitted on a high- Waliser and Graham 1993). The chosen oceanic areas
flying aircraft show updraft widths of 5–6 and 8–10 km have climatologically the warmest SSTs in the respective
over oceanic and land areas, respectively (H10; Guimond seasons and basins (Fig. 1; Table 1). The ‘‘Maritime
et al. 2010). Here updraft width is defined as a stretch along Continent’’ (MC), one of the rainiest regions on the
the aircraft path that has vertical velocities of more than planet (Ramage 1968), is included. Land locations are
5 m s21 at 10-km altitude. In almost all of the cases, peak chosen on the basis of either seasonal rainfall amount or
updrafts were observed above the 10-km level and land- known propensity for intense convection. Six land areas
based and sea-breeze convection had higher Ze than did have been selected: equatorial Africa (AF), equatorial
oceanic and tropical-cyclone convection (H10). South America (LAM), the western Himalayan foothills
The spatial coverage of ground radars and aircraft (WHF), the Indo-Gangetic Plain (IGP), northern Aus-
campaigns is very limited. It is not clear from past tralia (AUS), and southeastern South America (SESA).
studies how the vertical profiles of hydrometeors in ac- Of the 18 areas, 16 are in the tropics (i.e., within 308 latitude
tive Cb clouds compare in different parts of the tropics. from the equator). The remaining two (WHF and SESA) lie
This study is aimed at filling this gap in knowledge, and at the edge of the tropics (Fig. 1; Table 1) and are included in
the main objective is to document the vertical structure the analysis since they are known to contain the most intense
of Ze in intense convective clouds in the tropics. We convection on Earth (Zipser et al. 2006, hereinafter Z06;
consider both land and oceanic areas. Section 2 de- Houze et al. 2007; Romatschke et al. 2010).
scribes the data and methods used in the study, sections The method for the identification of convective cells is
3 and 4 contain results and discussion, respectively, and explained in detail in BK15 and is discussed briefly here.
section 5 concludes the paper. Figure 2 shows a vertical cross section through a cloud
system captured by the PR. In Fig. 2, tops of three
convective clouds extend beyond 17 km, and their lat-
2. Data and methods
eral sizes are more than 10 km (i.e., much larger than
The primary data used in the study are TRMM PR that of a PR pixel), which could indicate areas of coherent
2A25 (version 6) attenuation-corrected Ze (Iguchi et al. updrafts or downdrafts (H10). These are examples of
MAY 2016 KUMAR AND BHAT 1279

FIG. 1. Seasonal-average TMI SST and TRMM 3B42 precipitation for the 10-yr period from 2001 to 2010: (a) SST for June–September,
(b) SST for January–March, (c) June–September daily average precipitation, and (d) January–March daily average precipitation. Color
bars on the right show SST (8C) or precipitation (mm day21). Areas selected for detailed analysis are shown by rectangular boxes. See
Table 1 for their description.

convective cells referred to in this study. For identifying its base is located below the 3-km altitude (BK15). ICCs
convective cells embedded in a 3D reflectivity field, a Ze and CbTs have been constructed following the VPRR
threshold needs to be specified, often at some pre- method (ZL94), and the procedure is the same as that
determined height. For example, Dixon and Wiener employed in BK15. For each TRMM pass over a loca-
(1993) use thresholds in the 40–50-dBZ range for iden- tion, the highest Ze (i.e., Zemax) at the reference height
tifying individual convective cells in the Thunderstorm lying within the boundaries of the location is searched in
Identification, Tracking, Analysis and Nowcasting (TITAN) the remapped 2A25 data. If Zemax equals or exceeds the
algorithm. ZL94 specify 40 and 35 dBZ at 4.4- and 3.9-km threshold, a cloud cell is constructed by picking up the
heights, respectively, for defining convective clouds over local maximum Ze at each level from the radar volume
the midlatitudes and tropics, respectively. H10 specified a containing the beam having Zemax and its immediate eight
Ze threshold of 20 dBZ at 12-km height for defining neighboring beams. After excluding the beams associated
hot towers. with the cell(s) already constructed, the procedure is re-
In this study, the threshold for defining an intense peated until the pixels at the reference level that satisfy the
convective cell (ICC) is different at each location (and threshold criteria are exhausted. The lateral dimensions of
reference height) and is chosen such that no more than CbT and ICC are the same as that of a PR pixel.
5% of the pixels at the reference height have their Ze The physical interpretation of CbT and ICC, assuming
above this value. We consider two heights for ICCs (3 that they belong to the Cb cloud type, is as follows. In a
and 8 km), and the ICCs that correspond to these two Cb cloud, hydrometeors are lifted to the upper tropo-
heights are denoted by ICC3 and ICC8, respectively. sphere during its growth phase, and, once the updrafts
Table 1 shows Ze thresholds in different areas. It is ob- start weakening, larger hydrometeors begin their
served that ICC8 and ICC3 thresholds are in the 28–35- downward journey and grow along their trajectory. The
and 40–43-dBZ ranges, respectively. Similar to H10, a Ze reaches a peak value in the upper troposphere in the
cumulonimbus tower (CbT) is defined by referring to a early mature phase of a Cb cloud (Williams et al. 1989).
fixed Ze threshold value of 20 dBZ at 12-km altitude, and For example, cell B in Fig. 2 has high Ze values in the
1280 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 55

TABLE 1. Areas selected for comparison, reflectivity thresholds for ICC3 and ICC8 cells, and the number of convective cells. For the area/
region names, lowercase letters s and w refer to boreal summer and boreal winter, respectively.

ICC3 ICC8
threshold threshold No. of No. of SST
No. Area/region Boundaries and season (dBZ) (dBZ) No. of CbTs ICC8s ICC3s (8C)
1 Bay of Bengal (BOB) 158–198N, 868–938E; JJAS 42 32 7650 1881 18 490 29.0
2 Indian Ocean west (IOWs) 58S–38N, 558–798E; JJAS 41 32 9476 1410 37 853 29.4
3 Indian Ocean east (IOEs) 58S–38N, 818–1008E; JJAS 41 28.5 13 859 3400 55 408 29.0
4 Maritime Continent (MCs) 98S–108N, 1058–1308E; JJAS 43 32.5 42 571 6421 63 429 29.2
5 North Pacific (NPs) 58–208N, 1358–1558E; JJAS 41 30 22 622 5700 16 222 29.7
6 Central Pacific (CPs) 158S–58N, 1568E–1708W; 40 30 13 890 2547 15 067 29.3
JJAS
7 Atlantic Ocean (ATs) 08–18N, 208–508W; JJAS 41 28 11 937 3527 48 069 28.0
8 Indian Ocean west (IOWw) 108S–28N, 558–798E; JFM 40 29 2181 667 28 081 29.3
9 Indian Ocean east (IOEw) 118S–28N, 878–958E; JFM 41 31 4487 2853 37 155 29.4
10 Maritime Continent 128S–08, 1108–1408E; JFM 41 33.5 25 880 6793 74 981 29.6
(MCw)
11 Central Pacific (CPw) 118S–28N, 1458E–1708W; JFM 40.5 30.5 19 152 1426 71 904 29.8
12 Atlantic Ocean (ATw) 38S–78N, 108–308W; JFM 41 30.5 5309 1556 24 306 28.3
13 Equatorial Africa (AF) 08–108N, 48–288E; JJAS 43 35 54 211 11 995 22 604
14 Equatorial South 108S–28N, 508–758W; JJAS 42.5 34 26 541 8058 37 586
America (LAM)
15 Western Himalaya 288–338N, 658–728E; JJAS 43 32 13 944 2287 9289
foothills (WHF)
16 Indo-Gangetic Plain (IGP) 218–268N, 768–868E; JJAS 42 31 9878 5856 22 298
17 Northern Australia (AUS) 188–108S, 1328–1458E; JFM 42 32 17 054 1435 9679
18 Southeastern South 258–328S, 558–708W; JFM 41 31 14 859 2491 13 953
America (SESA)

upper troposphere, whereas cell C has relatively weak among the pure oceanic CbTs are small (,2 dBZ), and
echo in the upper troposphere but high Ze in the lower the weakest CbTs (Fig. 3a) are present over the Indian
troposphere and is perhaps in a later stage of the life Ocean western box during the boreal winter [January–
cycle relative to that of cell B. ICC8 cells are represen- March (JFM)]. Among the land areas, the decreasing
tative of clouds such as cell B. As hydrometeors descend
below the freezing level, Ze increases rapidly because of
cloud microphysical processes and the change from ice
to liquid phase (Houze 1993; Fabry and Zawadzki 1995).
We expect peak Ze to occur at 3-km altitude in the
middle and later stages of the mature phase, and ICC3
should correspond to such clouds. With a Ze threshold of
20 dBZ, CbTs may contain both growing and mature Cb
clouds. The ICC8 reference level lies in the upper
mixed-phase region in the tropical atmosphere (e.g.,
Stith et al. 2002) where convective and stratiform clouds
constitute the cloudy pixels (Li and Schumacher 2011).
Precipitating convective and stratiform clouds contrib-
ute to the population of the cloudy pixels at 3 km. Thus,
CbT, ICC8, and ICC3 capture Cb clouds in different
stages of their development, and the microphysical
processes associated with them are also different.
FIG. 2. Section through a cloud system observed by the PR over
the Indian region on 24 Jul 2008. The inset shows a horizontal
3. Results section taken at 3.5-km altitude (the abscissa and ordinate are
longitude and latitude, respectively). The vertical section is taken
Figure 3 shows the average vertical profiles of CbT, along the lines that cut through active convective clouds and is
ICC8, and ICC3 over different locations. The differences plotted against latitude. The color bar on the right shows Ze (dBZ).
MAY 2016 KUMAR AND BHAT 1281

FIG. 3. Average profiles of convective cells: (a) CbT, (b) CbT25, (c) ICC8, and (d) ICC3. To reduce confusion, mainly extreme cases
have been labeled and ‘‘TOC’’ refers to the remaining cases. Land and oceanic profiles are shown in thick solid and dashed lines,
respectively. A level is shown for an area if the number of data points in averaging is not less than 10% of the maximum number of profiles
for the area or 50, whichever is higher.
1282 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 55

order in the intensity of cells (in the layer between 5

and 10 km) is WHF, SESA, IGP, AUS, AF, and LAM.
CbTs over the Maritime Continent and northern Bay of
Bengal (BOB) are almost identical above 6 km, and
their profiles are placed between continental and other
oceanic profiles. The BOB box is adjacent to the
landmass, and continental influence is felt here (i.e., it
is not purely oceanic). In a similar way, oceanic influ-
ence cannot be ignored over the Maritime Continent.
To highlight the most active CbTs, the averages of
those that contain the top 25% Ze values at 12-km
height among the CbTs (denoted by CbT25) are shown
in Fig. 3b. The largest differences observed in the ver-
tical profiles of Ze are seen in the CbT25 profiles. The
most intense CbT25s form over WHF and SESA, and
the 40-dBZ echo extends above 12 km. This observa-
tion is consistent with the findings reported in Z06,
Houze et al. (2007), and Romatschke and Houze (2010).
The differences among the oceanic CbT25 profiles are
relatively small (maximum difference is ;3 dBZ), and
profiles over BOB and the Maritime Continent area are
closer to the oceanic ones.
The differences among the land locations are fewer
for ICC8 (Fig. 3c) relative to those seen in CbT25. One
important observation from Fig. 3c is that WHF and AF
contain the most intense ICC8 clouds at subfreezing
temperatures. SESA has the weakest ICC8 clouds
among the land areas above 3 km, and the weakest ICC8
cells among all locations are seen over the Atlantic
Ocean [June–September (JJAS)] followed by the Indian
Ocean [east and summer (IOEs); west and winter
(IOWw)]. ICC8 clouds over IGP and AUS are almost
identical. Oceanic profiles of ICC8 cluster together. In
the case of ICC3 (Fig. 3d), land cells are stronger than
their oceanic counterparts; the interarea differences are
relatively small below 5 km, however, and land and
ocean profiles separate above 7 km. Location AF con-
tains the strongest ICC3, followed by WHF and LAM;
the weakest ICC3 is observed over the central Pacific
Ocean during the boreal summer.
The 40-dBZ Ze threshold is often used as a signature
of convective precipitation (Awaka et al. 1997; Steiner
et al. 1995), and the maximum height of 40-dBZ Ze is
also used as a measure of the intensity of convection
(Z06; XZ12). Figure 4 shows the distribution of the
FIG. 4. Frequency distribution of the top height of 40-dBZ reflectivity
maximum heights of 40-dBZ Ze for CbTs and ICCs. For
for (a) CbT, (b) ICC8, and (c) ICC3. Oceanic areas are shown by dashed
CbTs (Fig. 4a), the mode in the distribution of 40-dBZ lines, and solid lines show land areas. To reduce confusion, mainly ex-
top heights occurs between 4 and 6 km over oceans and treme cases have been labeled and ‘‘rest’’ refers to the remaining cases.
the Maritime Continent, and the fraction of clouds with
Ze $ 40 dBZ above 10-km altitude is extremely low over (;7 km) in clouds over WHF, AF, and LAM relative to
these locations. The height of penetration of 40-dBZ Ze other areas (Fig. 4b). For all ICC3s, the modal height of
is the maximum in CbTs over WHF and SESA. The 40-dBZ tops lies between 4 and 5 km (Fig. 4c), and two
modal height of 40-dBZ tops in ICC8s is elevated locations—namely, central Pacific in summer (CPs) and
MAY 2016 KUMAR AND BHAT 1283

IOWw—show a secondary mode at 3 km. In general, a easily released over WHF because of a stable layer that
narrow distribution implies close similarities among caps the moist boundary layer. The potentially unstable
convective clouds whereas a wider distribution suggests flow is orographically lifted to saturation over small hills,
larger cloud-to-cloud differences. and intense convection develops (Medina et al. 2010).
Table 1 shows that Ze thresholds of ICC3 are greater This suggests that, to understand convection over land
than or equal to 40 dBZ. Such high values of Ze at 3-km areas, advection of moisture, the local surface charac-
altitude are normally associated with deep convective teristics, and environmental conditions aloft should be
clouds. The ICC8 Ze threshold is minimum for the At- taken into account.
lantic Ocean in summer (ATs; 28 dBZ), and maximum The maximum in the average profile of CbTs (Figs. 3a,b)
for AF (35 dBZ). The ratio of the number of CbTs to the is observed between 3 and 5 km; Ze decreases toward
number of ICC3s differs among the regions; the maxi- the surface or remains nearly constant below the peak
mum value of the ratio is 2.4 and occurs over AF, and the (depending on whether the location is continental or
minimum value (0.08) occurs over the IOWw area. (A oceanic) while it decreases with height above. The
value of unity for the ratio means that the fraction of change in slope above 12 km could be due to Ze values
cloudy pixels at 3 km that qualify as CbTs is 5%.) Note approaching the PR detection limit. In the case of ICC8
that the average SST over IOWw is 29.38C, which is and ICC3, Ze decreases toward the surface in the lowest
comparable to SSTs at other locations (Table 1). The 3 km of the atmosphere at all locations. The decrease of
ratio of the number of CbTs to the number of ICC8s is Ze in the continental clouds below 3 km is related to the
more than unity, with its maximum and minimum values humidity structure of the lower troposphere and cloud
of 13.4 and 1.6 found over CPw and IOEw, respectively. microphysics (Liu and Zipser 2013). In the case of ICC3,
Thus, more than 65% of the cloudy pixels at 8 km qualify Ze decreases rapidly between 5 and 7 km (i.e., in the
as CbTs over the CPw area, whereas the corresponding mixed-phase region) and absolute slopes of the profiles
value is just 8% over IOEw. North Pacific in summer are small above 9 km. The average absolute slopes of the
(NPs) and CPw have comparable SSTs (Table 1), but the Ze profiles between 5- and 7-km altitudes have been
ratio of the number of CbTs to the number of ICC8s calculated and are shown in Fig. 5a. The slopes of ICC3
over these areas is very different. The above observa- (4–6 dBZ km21) profiles are nearly 2 times that of CbTs
tions suggest that, in addition to SST, local factors (such and ICC8s, and land cells have smaller slopes relative to
as vertical wind shear and dryness of the midtropo- oceanic ones. CbT25 cells have the smallest slopes (;1–
sphere) could matter when finer details of intense con- 3 dBZ km21), and those of ICC8s are marginally larger.
vective clouds are considered, which is not obvious from Oceanic CbT25s have comparable slopes that are larger
the average profiles (Fig. 3). than those of land clouds. The slopes shown in Fig. 5a
are related to the precipitation process and to the phase
change of water. Above the freezing level (which is
4. Discussion
typically around 5 km in the tropical atmosphere during
This study quantified the vertical structure of intense summer), conversion from liquid to ice phase decreases
convective clouds, primarily in the tropics. The average Ze by 6.5 dBZ (Fabry and Zawadzki 1995) because Ze is
vertical profiles of intense convective clouds (Fig. 3), as calculated assuming the complex refractive index of
depicted by CbT, ICC8, and ICC3, show similarities as liquid water, which is nearly 4.5 times that of the ice. Ice
well as differences. Close similarity among the oceanic phase also accelerates the growth of hydrometeors as
cells, especially in the mid- and lower troposphere, the cloud air mass ascends in the mixed-phase region;
suggests that the development of vigorous convective larger hydrometeors descend and grow further along
cells over warm oceans is similar and that understanding their downward trajectory and thus Ze increases, since it
gained in one region is extendable to other areas. This is proportional to the sixth power of hydrometeor size
result is encouraging from a cloud-modeling perspec- (Houze 1993). The net result is that Ze decreases rapidly
tive. For land areas, the regional differences are large. with height in the mixed-phase region. This mechanism
The strongest CbTs form over WHF and SESA, which is favored if the updraft velocities in the mixed-phase
are located in the subtropics. It is observed from Fig. 1 region are low, as in, for example, oceanic cloud systems
that these two regions are not the areas of highest (Lucas et al. 1994). In the case of intense convective
rainfall over land. So, convection is not frequent, but, clouds, a different mechanism may be operating, as
when it occurs, it tends to be very intense. Medina et al. explained below.
(2010) argue that very intense convection develops over The convective cells studied in this work may be
WHF as a result of surface flux feedbacks and region- treated as very intense since either they extend deep
specific features. Instability builds up but does not get into the upper troposphere (CbT) or are in the top-5%
1284 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 55

FIG. 5. (a) Average slope of Ze (absolute value) in the atmospheric layer between 5- and 7-km
altitudes for the profiles shown in Fig. 3. The numbers along the x axis refer to the number in the first
column in Table 1. (b) Change in cloud ice water content between 7- and 5-km altitudes (g m23).

Ze bracket at their respective levels. In intense convec- volume of cloud air) needs to be considered. These two
tive clouds, updraft velocities tend to be high, leaving quantities are empirically related by an expression of the
less time for hydrometeors to grow in size during the form M 5 cZed (here Ze has for units: mm6 m23; Smith
growth phase, and more hydrometeors are carried to the 1984). Values of c and d depend on the type of pre-
upper troposphere (H10). Once the updrafts weaken in cipitation (Black 1990). For glaciated convective clouds
the upper troposphere, larger hydrometeors start their inside a hurricane (mostly containing graupel), Black’s
downward journey and grow along their path through (1990) study gives c 5 915 and d 5 1.51 (for which M has
ice-phase cloud microphysical processes up to the for units: g m23). Assuming these values for the present
melting level. In this scenario, the slope of the profile in convective cells (the main purpose is to get at least an
the mixed-phase region depends more on the growth of approximate picture of changes in the hydrometeor
hydrometeors than that due to the phase-change effect concentration in clouds over different regions), the in-
(by the time hydrometeors arrive here from the upper crease in M between the 7- and 5-km levels (DM) has
troposphere, the cloud would have glaciated). Then, a been calculated and is shown in Fig. 5b. It is observed
convective cloud with a larger (absolute) slope suggests that the growth rate of hydrometeors is higher in ICC8
a higher growth rate of hydrometeors. From this point of than in ICC3. The lowest growth rate is observed in the
view, hydrometeor growth rate along the downward oceanic CbTs, whereas over land it is in ICC3. The
trajectory is highest in ICC3s, slowest in CbT25s, and highest growth of hydrometeors (;0.5 g m23) occurs in
more efficient in oceanic clouds than in continental CbT25s over IGP.
clouds (Fig. 5a). A careful analysis reveals that this need The reasons for smaller slopes of the average profiles
not always be true, however, because the slope shown in above 12 km are not clear. The effect of the sensitivity of
Fig. 5a is calculated taking Ze expressed in reflectivity the PR being ;17 dBZ will be felt at these levels, and
decibel units (i.e., the original Ze has been subjected to a sampling could be an issue. For example, Fig. 6 shows
logarithmic transformation). A consequence of the log the number of ICC3 profiles as a function of height. It is
transformation is that small values get magnified and observed that almost all ICC3 clouds extend up to 5 km
larger values get compressed. Therefore, to understand and that their number decreases rapidly with height
which cloud cells have a greater precipitation growth above 6 km. Figure 6 should be interpreted with care.
rate, the relationship between Ze and cloud ice water Over the tropical oceans where fallout of precipitation
content M (the amount of glaciated water per unit could be rapid above the freezing level, hydrometeor
MAY 2016 KUMAR AND BHAT 1285

Acknowledgments. This work is supported by grants

from the Department of Science and Technology,
New Delhi, Indian, and the Ministry of Earth Sci-
ences, New Delhi. TRMM 2A25 (http://mirador.gsfc.
nasa.gov/cgi-bin/mirador/presentNavigation.pl?tree=
project&dataset=2A25%20%28Version%20007%29:%
20Radar%20Rainfall%20Rate%20and%20Profile%
20%28PR%29&project=TRMM&dataGroup=Orbital&
version=007) and 3B42 (http://mirador.gsfc.nasa.gov/cgi-bin/
mirador/presentNavigation.pl?tree=project&dataset=
3B42:%203-Hour%200.25%20x%200.25%20degree%
20merged%20TRMM%20and%20other%20satellite%
20estimates&project=TRMM&dataGroup=Gridded&
version=007) data are taken from NASA’s Earth–Sun
System Division website. The authors thank the anony-
mous reviewers for their constructive suggestions. TMI
data were produced by Remote Sensing Systems, Inc., and
were sponsored by the NASA Earth Sciences Division.

REFERENCES
FIG. 6. Number of ICC3 profiles at different altitudes normalized
Awaka, J., T. Iguchi, H. Kumagai, and K. Okamoto, 1997: Rain
by their number at 3 km.
type classification algorithm for TRMM Precipitation Radar.
Proc. 1997 Int. Geoscience and Remote Sensing Symp., Sin-
concentration may reduce to a level at which the echo gapore, Institute of Electrical and Electronics Engineers,
strength is less than the PR detection limit. The actual 1633–1635, doi:10.1109/IGARSS.1997.608993.
cloud-top height as well as the number of convective Bhat, G. S., and S. Kumar, 2015: Vertical structure of cumulo-
nimbus towers and intense convective clouds over the South
clouds could be much higher at upper levels than that
Asian region during the summer monsoon season. J. Geophys.
indicated by Fig. 6 (Sindhu and Bhat 2013). If the PR Res. Atmos., 120, 1710–1722, doi:10.1002/2014JD022552.
sensitivity is not a factor, then smaller slopes above Black, R. A., 1990: Radar reflectivity–ice water content relation-
12 km mean low growth of hydrometeor concentration/ ships for use above the melting level in hurricanes. J. Appl.
size at these heights. Meteor., 29, 955–961, doi:10.1175/1520-0450(1990)029,0955:
RRIWCR.2.0.CO;2.
Cotton, W. R., and R. A. Anthes, 1989: Storm and Cloud Dy-
5. Conclusions namics. Academic Press, 882 pp.
Dixon, M., and G. Wiener, 1993: TITAN: Thunderstorm Identifi-
There are four main conclusions of this study: cation, Tracking, Analysis, and Nowcasting—A radar-based
methodology. J. Atmos. Oceanic Technol., 10, 785–797, doi:10.1175/
1) Intense convective clouds over the ocean have similar 1520-0426(1993)010,0785:TTITAA.2.0.CO;2.
average vertical structure across basins. Convective Fabry, F., and I. Zawadzki, 1995: Long-term radar observations of
clouds over land exhibit larger regional variability, the melting layer of precipitation and their interpretation.
J. Atmos. Sci., 52, 838–851, doi:10.1175/1520-0469(1995)052,0838:
and intense convective clouds over the northern Bay of
LTROOT.2.0.CO;2.
Bengal and the Maritime Continent are closer to Gadgil, S., P. V. Joseph, and N. V. Joshi, 1984: Ocean–atmosphere
oceanic cells. coupling over monsoon regions. Nature, 312, 141–143, doi:10.1038/
2) Reflectivity decreases toward the surface below 4 km 312141a0.
for ICC8 and ICC3 over both land and ocean, whereas Graham, N. E., and T. T. Barnett, 1987: Sea surface temperature,
this happens for land areas only in the case of CbTs. surface wind divergence, and convection over tropical oceans.
Science, 238, 657–659, doi:10.1126/science.238.4827.657.
3) The slope of the radar reflectivity factor is larger for Guimond, S. R., G. M. Heymsfield, and F. J. Turk, 2010: Multiscale
the oceanic profiles between 7 and 5 km; the increase observations of Hurricane Dennis (2005): The effect of hot
in the amount of hydrometeors in this layer is not less in towers on rapid intensification. J. Atmos. Sci., 67, 633–654,
continental clouds relative to oceanic ones, however. doi:10.1175/2009JAS3119.1.
4) The most intense CbTs form over the foothills of the Heymsfield, G. M., L. Tian, A. J. Heymsfield, L. Li, and
S. Guimond, 2010: Characteristics of deep tropical and sub-
western Himalaya and southeastern South America, tropical convection from nadir-viewing high-altitude airborne
and the most intense ICC3s develop over equatorial Doppler radar. J. Atmos. Sci., 67, 285–308, doi:10.1175/
Africa. 2009JAS3132.1.
1286 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 55

Houze, R. A., 1993: Cloud Dynamics. Academic Press, 496 pp. Romatschke, U., and R. A. Houze Jr., 2010: Extreme summer
——, D. C. Wilton, and F. B. Smull, 2007: Monsoon convection in the convection in South America. J. Climate, 23, 3761–3791,
Himalayan region as seen by the TRMM Precipitation Radar. doi:10.1175/2010JCLI3465.1.
Quart. J. Roy. Meteor. Soc., 133, 1389–1411, doi:10.1002/qj.106. ——, S. Medina, and R. A. Houze Jr., 2010: Regional, seasonal, and
Huffman, G. J., and Coauthors, 2007: The TRMM Multisatellite diurnal variations of extreme convection in the South Asian
Precipitation Analysis (TMPA): Quasi-global, multilayear, region. J. Climate, 23, 419–439, doi:10.1175/2009JCLI3140.1.
combined-sensor precipitation estimates at fine scales. Saikranthi, K., T. N. Rao, B. Radhakrishna, and S. V. B. Rao, 2014:
J. Hydrometeor., 8, 38–55, doi:10.1175/JHM560.1. Morphology of the vertical structure of precipitation over
Iguchi, T., T. Kozu, R. Meneghini, J. Awaka, and K. Okamoto, India and adjoining oceans based on long-term measurements
2000: Rain-profiling algorithm for the TRMM Precipita- of TRMM PR. J. Geophys. Res. Atmos., 119, 8433–8449,
tion Radar. J. Appl. Meteor., 39, 2038–2052, doi:10.1175/ doi:10.1002/2014JD021774.
1520-0450(2001)040,2038:RPAFTT.2.0.CO;2. Sindhu, K. D., and G. S. Bhat, 2013: Comparison of CloudSat and
——, ——, J. Kwiatkowski, R. Meneghini, J. Awaka, and K. Okamoto, TRMM radar reflectivities. J. Earth Syst. Sci., 122, 947–956,
2009: Uncertainties in the rain profiling algorithm for the TRMM doi:10.1007/s12040-013-0316-9.
Precipitation Radar. J. Meteor. Soc. Japan, 87A, 1–30, doi:10.2151/ Smith, P. L., 1984: Equivalent radar reflectivity factors for snow
jmsj.87A.1. and ice particles. J. Appl. Meteor. Climatol., 23, 1258–1260,
Kummerow, C., W. Barnes, T. Kozu, J. Shiue, and J. Simpson, 1998: doi:10.1175/1520-0450(1984)023,1258:ERRFFS.2.0.CO;2.
The Tropical Rainfall Measuring Mission (TRMM) sensor Steiner, M., R. A. Houze, and S. E. Yuter, 1995: Climatological
package. J. Atmos. Oceanic Technol., 15, 809–817, doi:10.1175/ characterization of three-dimensional storm structure from op-
1520-0426(1998)015,0809:TTRMMT.2.0.CO;2. erational radar and rain gauge data. J. Appl. Meteor., 34, 1978–
Li, W., and C. Schumacher, 2011: Thick anvils as viewed by the 2007, doi:10.1175/1520-0450(1995)034,1978:CCOTDS.2.0.CO;2.
TRMM Precipitation Radar. J. Climate, 24, 1718–1735, doi:10.1175/ Stephens, G. L., and Coauthors, 2002: The CloudSat mission and
2010JCLI3793.1. the A-TRAIN: A new dimension to space-based observations
Liu, C., and E. J. Zipser, 2013: Why does radar reflectivity tend to of clouds and precipitation. Bull. Amer. Meteor. Soc., 83,
increase downward toward the ocean surface, but decrease 1771–1790, doi:10.1175/BAMS-83-12-1771.
downward toward the land surface? J. Geophys. Res., 118, Stith, J. L., J. E. Dye, A. Bansemer, A. J. Heymsfield, C. A.
135–148, doi:10.1029/2012JD018134. Grainger, W. A. Petersen, and R. Cifelli, 2002: Microphysical
——, D. J. Cecil, E. J. Zipser, K. Kronfeld, and R. Robertson, 2012: observations of tropical clouds. J. Appl. Meteor., 41, 97–117,
Relationships between lightning flash rates and radar re- doi:10.1175/1520-0450(2002)041,0097:MOOTC.2.0.CO;2.
flectivity vertical structures in thunderstorms over the tropics Waliser, D. U., and N. E. Graham, 1993: Convective cloud systems
and subtropics. J. Geophys. Res., 117, D06212, doi:10.1029/ and warm-pool sea surface temperatures’ coupled interactions
2011JD017123. and self-regulation. J. Geophys. Res., 98, 12 881–12 893, doi:10.1029/
Lucas, C., E. D. Zipser, and M. A. LeMone, 1994: Vertical velocity 93JD00872.
in oceanic convection off tropical Australia. J. Atmos. Wentz, F. J., C. Gentemann, D. Smith, and D. Chelton, 2000:
Sci., 51, 3183–3193, doi:10.1175/1520-0469(1994)051,3183: Satellite measurements of sea surface temperature through
VVIOCO.2.0.CO;2. clouds. Science, 288, 847–850, doi:10.1126/science.288.5467.847.
Masunaga, H., T. Iguchi, R. Oki, and M. Kachi, 2002: Comparison Williams, E. R., M. E. Weber, and R. E. Orville, 1989: The re-
of rainfall products derived from TRMM Microwave Imager lationship between lighting type and convective state of
and Precipitation Radar. J. Appl. Meteor., 41, 849–862, doi:10.1175/ thunderstorms. J. Geophys. Res., 94, 13 213–13 220, doi:10.1029/
1520-0450(2002)041,0849:CORPDF.2.0.CO;2. JD094iD11p13213.
Medina, S., R. A. Houze Jr., A. Kumar, and D. Niyogi, 2010: Xu, W., and E. J. Zipser, 2012: Properties of deep convection in
Summer monsoon convection in the Himalayan region: Ter- tropical continental, monsoon, and oceanic rainfall regimes.
rain and land cover effects. Quart. J. Roy. Meteor. Soc., 136, Geophys. Res. Lett., 39, L07802, doi:10.1029/2012GL051242.
593–616, doi:10.1002/qj.601. Yuan, T., and X. Qie, 2008: Study on lightning activity and pre-
Nesbitt, S. W., E. J. Zipser, and D. J. Cecil, 2000: A census of cipitation characteristics before and after the onset of the
precipitation features in the tropics using TRMM radar, South China Sea summer monsoon. J. Geophys. Res., 113,
ice scattering, and lightning observations. J. Climate, D14101, doi:10.1029/2007JD009382.
13, 4087–4106, doi:10.1175/1520-0442(2000)013,4087: Zipser, E. J., and K. R. Lutz, 1994: The vertical profile of radar
ACOPFI.2.0.CO;2. reflectivity of convective cells: A strong indicator of storm intensity
Ramage, C. S., 1968: Role of a tropical ‘‘Maritime Continent’’ in and lightning probability? Mon. Wea. Rev., 122, 1751–1759,
the atmospheric circulation. Mon. Wea. Rev., 96, 365–370, doi:10.1175/1520-0493(1994)122,1751:TVPORR.2.0.CO;2.
doi:10.1175/1520-0493(1968)096,0365:ROATMC.2.0.CO;2. ——, D. J. Cecil, C. Liu, S. W. Nesbitt, and D. P. Yorty, 2006:
Riehl, H., and J. S. Malkus, 1958: On the heat balance in the Where are the most intense thunderstorms on Earth? Bull. Amer.
equatorial trough zone. Geophysica, 6, 503–538. Meteor. Soc., 87, 1057–1071, doi:10.1175/BAMS-87-8-1057.