Available online at www.sciencedirect.

com

ScienceDirect
Advances in Space Research 58 (2016) 1017–1032
www.elsevier.com/locate/asr

Three dimensional characteristics of precipitating cloud

systems observed during Indian summer monsoon
Shailendra Kumar ⇑
Room No. 116, CAOS, IISc, Bengaluru 560012, India

Received 19 November 2015; received in revised form 26 April 2016; accepted 29 May 2016
Available online 5 June 2016

Abstract

Ten year of TRMM PR data is used to study the properties of the precipitating cloud systems over Indian land and ocean areas.
TRMM PR reflectivity information is often displayed in three dimensions, making it excellent for study the three dimensional charac-
teristics of precipitating cloud systems. A cloud system (CS) is defined by connecting pixels of more than 17 dBZ, with at least one pixel
must be higher than 40 dBZ. Within the CSs two types of cloud cells are defined, namely most vigorous cell (MVC) and reflectivity profile
for cloud (RPC) for informing the vertical structure of precipitation. Most vigorous cell contains the maximum Ze at each altitude from
the CSs, whereas the RPC consists of the vertical extension of Ze P 40 dBZ at any altitude. Area covered by different reflectivity thresh-
olds (30, 35 and 40 dBZ), and analysis of three dimensional cloud vertical structure show the evolution characteristics of cloud systems.
Vertical structure of precipitation shows higher regional differences compared to cloud area and evolution of clouds. Head of Bay con-
sists higher number of precipitating cloud systems. Land areas show the intense cloud systems compared to oceanic areas and on an
average, precipitation intensity is least over Arabian Sea and highest over central India. Probability of warm rain is higher over Western
Ghats and Arabian Sea. Between 5 and 10 km altitude, reflectivity values show 5 dBZ difference in MVC and RPC. Western Ghats shows
higher reflectivity value above 12 km altitude and matches with land dominated areas. Land dominated area shows higher fraction of
deep clouds. Almost 50% MVCs cross 10 km altitude over land dominated areas whereas only 20% MVCs are crossing the 10 km altitude
over Arabian Sea. Maximum height of 30 and 40 dBZ consist the lower mode over Arabian Sea and Bay of Bengal compared to land
dominated areas. Small cloud systems consist higher area of intense precipitation (higher values of Ze) and as the area of cloud system
increases, the area of intense precipitation (high reflectivity) decreases. 30 and 35 dBZ cover more than 40% and 20% precipitation areas
in the cloud systems irrespective of subareas. Cloud areas show higher regional differences below the freezing level. Evolution of cloud
systems show similar characteristics, as land and oceanic areas consist of the similar patterns of evolution. Cloud systems with bright
band features consist of the higher reflectivity values and cloud liquid water near 5 km altitude and reflectivity values show a sharp
decrease just below the bright band.
Ó 2016 COSPAR. Published by Elsevier Ltd. All rights reserved.

Keywords: TRMM PR; Reflectivity; Vertical profile; Cloud systems; Cloud area; Three dimensional cloud vertical structure

1. Introduction cumulonimbus cloud is the primary block of a larger multi

thunderstorms (e.g., Houze, 1989). Understanding the
Deep clouds play an important role in transporting the structure of organized convective systems remain a frontier
net energy gain at the surface of equatorial regions to area of research in tropical meteorology (Randall et al.,
the upper atmosphere (Riehl, 1979). An individual 2007). The Tropical Rainfall Measuring Mission Precipita-
tion Radar (TRMM-PR henceforth) has been measuring
⇑ Tel.: +91 9535883265. the three dimensional structure of precipitation in terms
E-mail address: [email protected].

http://dx.doi.org/10.1016/j.asr.2016.05.052
0273-1177/Ó 2016 COSPAR. Published by Elsevier Ltd. All rights reserved.
1018 S. Kumar / Advances in Space Research 58 (2016) 1017–1032

of reflectivity (Ze) since December 1997 (Kummerow et al., observing the initial, mature and dissipating stages. Initial
1998), and provides large data for understanding of three stage consists more warm precipitation i.e. cloud tops
dimensional precipitating cloud systems. below the freezing height and Ze peaks in the upper tro-
TRMM-PR enabled the study of diurnal variation (e.g. posphere in the early mature phase of a cumulonimbus
Takayabu, 2002), spatial distribution (Hirose and cloud (Williams et al., 1989). Dutta et al. (2014) studied
Nakamura, 2005) and other characteristics (e.g., convective the bow echoes and observed their vertical evolution.
vs stratiform, Houze et al., 2007) of precipitating clouds Later stage of life cycle consists of high Ze values in
round the tropics. Petersen and Rutledge (2001) observed lower troposphere compared to early stage where higher
systematic variability in the vertical structure of precipita- Ze values occur at higher altitude.
tion above freezing level altitudes over selected global trop- Indian regions are highly inhomogeneous and provide a
ical locations. Alcala and Dessler (2002) observed deep suitable area to compare the precipitating cloud systems in
convection (P10 km) and overshooting convection detail. Most of precipitation occurs during JJAS months
(P14 km) using 4 months of TRMM data and observed and consists several highly precipitation areas (Fig. 1).
that, 5% clouds are involved in exporting the water from Our aim is not to separate the convective and stratiform
troposphere to stratosphere by deep convective clouds. precipitation, but to find the regional differences in the
TRMM based precipitation features (PFs) are important cloud vertical structure with Ze P 40 dBZ which indicate
to understand the properties of tropical cloud systems the intense precipitation (Dixon and Wiener, 1993;
(Nesbitt et al., 2000, 2006; Liu et al., 2008, 2012). The land Houze et al., 2007; Qie et al., 2014; Kumar, 2015). Frac-
areas (Africa, Argentina and India) contain deeper and tion area covered by different Ze thresholds (30, 35 and
horizontally more extensive PFs compared to oceanic areas 40 dBZ) within a cloud systems, provide the different pre-
(Zipser et al., 2006; Liu et al., 2007). Yuan and Qie (2008) cipitation intensity as well as different type of precipitation
observed the pre and post monsoon characteristics of light- within the cloud systems. Fraction area covered at differ-
ning activity and precipitation during China summer mon- ent altitudes (2–5 km) show the evolution of cloud systems.
soon, whereas Liu et al. (2012) observed the dependency of Derived cloud liquid water for a cloud system is used to
lightning flash rates with the radar reflectivity in the verti- separate the cloud systems within different phases of evo-
cal structure in thunderstorms. lution. Data and methodology are presented in Section 2,
Liu et al. (2008) used multiple TRMM sensors and where we introduce the concept of precipitating cloud sys-
observed that land area consists higher 20 dBZ area above tem (CS) based on TRMM sensitivity and different Ze
the freezing level, compared to oceanic areas. The pattern thresholds, ’most vigorous cell’ (MVC), ’reflectivity profile
and structure of deep convective and intense deep convec- of cloud ’(RPC) and three dimensional cloud vertical
tive systems over the Tibetan Plateau region based on structure (3DCVS). Results are presented in Section 3 fol-
TRMM data were investigated in detail by Qie et al. lowed by discussion in Section 4. Section 5 concludes the
(2014). Storms with Ze P 40 dBZ radar echoes with cloud paper.
top height above 6 km contribute significant amount of
rainfall (Xu and Zipser, 2012). However, along the western
Himalayas, observed precipitation is mainly contributed by
small scale but deep convective systems, whereas stratiform
clouds occur near eastern foothills (Houze et al., 2007;
Romatschke and Houze, 2011). TRMM PR based observa-
tion shows that land dominated areas consist more intense
convective clouds during Indian summer monsoon (Bhat
and Kumar, 2015; Kumar, 2015) whereas Western Hima-
layas foothills with south east south America consist most
intense cumulonimbus towers during their respective sum-
mer (Kumar and Bhat, 2016a).
The basic algorithm of convective and stratiform sepa-
ration consists the Ze peak near 0°C in most vigorous
convective regions (Steiner et al., 1995). TRMM PR
observation shows the highest stratiform rainfall fraction
area (35%) over the Arabian Sea (AS), Bay of Bengal
(BOB), eastern Brazil, followed by 20–30% stratiform rain
area over central Africa, parts of Maritime Continent and
the Caribbean (Schumacher and Houze, 2003). Ze
Fig. 1. Different areas are selected based on TRMM 3B42 spatial rainfall
changes during the life cycle of cloud systems and area
distribution. Western Ghats, head of Bay along the eastern coast of Bay
concentration changes drastically near the freezing height consist the precipitation maxima zone. Colorbar is in mm h 1. (For
(Szoke and Zipser, 1986). Byers and Braham (1948) interpretation of the references to colour in this figure legend, the reader is
studied the life cycle of convective cloud systems, by referred to the web version of this article.)
 S. Kumar / Advances in Space Research 58 (2016) 1017–1032 1019

2. Data and method of analysis

2.1. TRMM PR

TRMM PR V6 attenuation-corrected radar reflectivity

factor data is used in present study. The time span of data
is for JJAS months between the years 2001 and 2010.
TRMM PR works in Ku band with 2.2 cm frequency.
TRMM PR covers the tropical belt 15–16 times in a day
and provides the Ze data in dBZ unit in three dimensional
volumetric fields. Details on the TRMM satellite and PR
can be found in Kummerow et al. (1998). The horizontal
resolution of TRMM PR is 5 km  5 km with vertical res-
olution of 0.25 km. The maximum observable range of
TRMM PR is 19.25 km whereas the lowest altitude corre-
spond to Earth’s ellipsoid. The sensitivity of TRMM PR is
Fig. 2b. Projection of the maximum reflectivity in the longitudinal belt 60°
17 dBZ and could not measure the too less intense
E–100°E on the latitude–height plane for a single pass on 25 August 2010.
precipitation. The colour bar on right shows Ze in dBZ unit. (For interpretation of the
Fig. 2a shows the three dimensional radar reflectivity references to colour in this figure legend, the reader is referred to the web
patterns observed from TRMM PR. At the same time version of this article.)
TRMM PR can observe the Ze at different locations over
the tropics. The upper figure shows the swath of the
TRMM PR at 3 km altitude and vertical pattern of reflec- cells, which are crossing the 10 km, correspond to deep
tivity is shown at three different locations. We see the clouds and Ze is around 40 dBZ near the freezing level.
reflectivity values are crossing the 16 km altitude with Clouds with base height above the freezing level (5 km),
cloud base height <2 km and represent the cumulonimbus may be the part of stratiform or an anvil clouds. A partic-
towers. Fig. 2b is showing the projection of maximum ular meso-scale convective system (MCS) is observed in
reflectivity on 2D plane (latitude–height). Projection of Fig. 2c. Deep clouds with cloud top height (P10 km), fol-
maximum reflectivity shows the maximum vertical exten- lowed by bright band (horizontal pattern of Ze P 40 dBZ
sion of clouds present at time of observation. The cloud 5 km) and anvil clouds (Ze P 17 dBZ with echo base

Fig. 2a. Three dimensional field of reflectivity observed from TRMM PR. Upper figure shows the swath of TRMM PR. The vertical patterns of
reflectivity is observed at the three different locations. Here colorbar is not shown and for the colorbar information you can look Figs. 2b and c. (For
interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
1020 S. Kumar / Advances in Space Research 58 (2016) 1017–1032

area near the Bay of Bengal, which is primary area for gen-
erating the monsoon systems (Goswami et al., 2003), and
R-4 is chosen for this purpose. Besides of these high spa-
tially rainy areas, we selected the two land areas (R-2
and R-3) to compare the precipitation cloud systems with
highest precipitating areas as well as the oceanic ares. We
also chosen the boxes over the BOB (R-5 and R-6) and
AS (R-7) to compare the precipitating convective system.

2.4. Definition of precipitating cloud system (CS),

reflectivity profile for cloud (RPC), most vigorous cell
(MVC) and three dimensional cloud vertical structure
(3DCVS)

Fig. 3 explains the different methods and definitions are

Fig. 2c. A vertical cross section of MCS is shown here. Bright band
used in the present study. TRMM PR data is re-gridded at
followed by the tall cumulonimbus clouds over central India is observed 1°  1° resolution for the present study. For each TRMM
here. The colour bar on right shows Ze in dBZ unit. (For interpretation of PR pass first we projected the maximum reflectivity on lon-
the references to colour in this figure legend, the reader is referred to the gitude–latitude plane (Fig. 3a). Cloud system (CS) consists
web version of this article.) of connected pixels, with Ze P 17 dBZ in x–y and along
the diagonal direction. Only the cases, where at-least one
height above 3 km, Li and Schumacher, 2011) are pixel of Ze is higher than 40 dBZ are considered as CSs
observed. to avoid the less intense precipitating CSs. Fig. 3a shows
an example of several CSs in a single TRMM pass over
BOB and shown by square boxes. Area covered by different
2.2. Reanalysis data reflectivity thresholds are calculated using (Ath/A17), where
Ath is the area covered by different reflectivity thresholds in
Surface and atmospheric conditions are obtained by the area of CSs (A17). The thresholds are chosen as 30, 35
using reanalysis data between 2001 and 2010 years for and 40 dBZ, as these reflectivity thresholds indicate the dif-
JJAS months. National Centers for Environmental Predic- ferent rain intensity and rain type (e.g., Houze (1993)) and
tion (NCEP, Kalnay et al., 1996) reanalysis data products their area fraction at different altitudes show the different
are used for surface temperature, humidity and water precipitation type evolves among the different subareas.
vapour column fields. For column water vapour horizontal Fig. 3a shows that a single CS may have many Ath area.
resolution of data is 0.25°  0.25°. Interim European Cen- The pixels with Ze P 40 dBZ represent the intense precip-
ter for Medium Range Weather Forecasting (ECMWF) itating area (Houze et al., 2007; Kumar, 2015) and their
Re-Analysis (ERA-Interim, Dee et al., 2011) data is used vertical structure are important to understand the intense
to characterize the vertical characteristics of the atmo- precipitation characteristics. RPC consists of the vertical
sphere. This reanalysis data has a horizontal resolution of profiles of radar reflectivity of pixels having Ze P 40 dBZ
approximately 1.5°  1.5° and 27 vertical levels between in each CSs (Fig. 3c). MVC consists of the maximum Ze
1000 and 100 hPa. at each altitude from different CSs in a single TRMM PR
pass (Fig. 3b). Three dimensional cloud vertical structure
2.3. Definition of sub areas (3DCVS) consists of the vertical profiles of radar reflectiv-
ity in the area of CS (Fig. 3d). To construct the 3DCVS in
Fig. 1 shows the sub-areas selected for comparison in each TRMM PR pass, we connected all the CSs within a
precipitating cloud systems based on TRMM 3B42 single TRMM pass and so one single TRMM PR pass pro-
(Huffman et al., 2007) spatial distribution of rainfall, dur- vides only one 3DCVS (Fig. 3d). Fig. 3d shows an example
ing the June–September period of 2001–2010. Box R-1 is of the 3DCVS for a CS, and X-axis is showing the number
chosen for understanding the clouds near the Western of connected pixels. We arranged the radar reflectivity pro-
Ghats (WGs), where orographic lifting promotes the for- files based on their echo base height (Ze P 17 dBZ). Radar
mation of precipitating clouds and copious rain fall is reflectivity profiles with base height 63 km, between 3–
observed (Hoskins and Karoly, 1981). TRMM PR obser- 5 km and P5 km are termed as type a, b and c respectively
vation shows that WGs and adjoining ocean consist the (Fig. 3f). Greene and Clark (1972) derived the cloud liquid
highest fraction of intense shallow precipitation and could water (CLW), using relation M = 3.44 * Z4/7 * 10 6 (where
be responsible for the intense precipitation near WGs M in unit of gm 3 and Z is in unit of mm6 m 3). We calcu-
(Kumar and Bhat, 2016b, under review). TRMM 3B42 lated the average CLW for 3DCVS at each altitude. Area
based spatial rainfall distribution also shows a heavy rain of different CSs varies more the 20 times, so absolute value
fall area in the box R-4 i.e. head of Bay and continental of CLW is rather meaningless, therefore we calculated the
 S. Kumar / Advances in Space Research 58 (2016) 1017–1032 1021

15
(b)MVC (c)RPC
(a)
40
16

Height (Km)
10
Latitude

15 30

15 20
5
13 10

12 0
0 20 30 40 50 20 30 40 50
73 74 75 76 77 77 Reflectivity(dBZ)
Longitude
20
(d)All radar beams
14 (e)CLW
12 15
Height (Km)

10
8 10
6
4 5
2
0 0
0.2 0.4 0.6 0.8 1
100 200 300 Normalized CLW
No. of connected pixels

Fig. 3. Concept of cloud systems, reflectivity profile for cloud, most vigorous cell and three dimensional cloud vertical structure are shown here. Fig. 3(a)
shows the cloud systems observed in a single TRMM pass and small boxes are showing Ze P 30 dBZ areas within a cloud systems. Fig. 3(b) shows an
example of MVC as it consists of maximum Ze at each altitude within a cloud systems. Fig. 3(c) shows the vertical profile of RPC of the areas, having
Ze P 40 dBZ. Fig. 3(d) shows the individual vertical profiles of for a CS and Fig. 3(e) shows the normalized cloud liquid water content at each altitude for
a CS.

TRMM PR : 2A25.030601.31604
60

17.5 type "a" type ’b’

type ’c’ 50
15
Height (Km)

50
Reflectivity (dBZ)

12.5
40 45
10
40
7.5
30 35
5.0
30
2.5 20
0 25
20 40 60 80 100 120 140
20
10
Connecred pixels 0
5
20 30 40 50 125
10 100
75
15 50
25
Height (Km) 20 0 Connected pixels

Fig. 3f. A three dimensional structure of the cloud system observed for TRMM PR for a particular CS. Each cloud systems are categorized into three
types based on their echo base height.

normalized CLW at each altitude by dividing CLW at each Our definition of CSs are similar to the definition used
altitude by maximum CLW (Fig. 3e). Normalized CLW by (Dixon and Wiener, 1993), except that we also used
shows the peak in CLW profile and could be important the connected pixels along the diagonal direction. RPC
for separating the CSs with and without bright band. does not discriminate between convective and stratiform
1022 S. Kumar / Advances in Space Research 58 (2016) 1017–1032

precipitation as Ze P 40 dBZ also belongs to stratiform the decreasing precipitation intensity. Below the 0°C the
precipitation, but they are the intense precipitation areas higher width informs the amount of condensed water,
within CSs (Kumar, 2015). MCS follows a life cycle, con- which is mainly produced by the precipitating hydromete-
sist of growing, mature and decaying stages (Leary and ors. Above the 0°C level the distribution of reflectivity
Houze, 1979) and Ze evolves continuously during the life becomes narrower. Below 5 km, the highest modes for
cycle and peaks during the mature stage (e.g., Williams MVCs are 40–45 dBZ, whereas for RPCs, mode is around
et al., 1989) and so the pixels of RPC come from the differ- 35–40 dBZ. The width in the CFAD for land dominated
ent mature cloud cells present at the time of observation. MVCs show higher width compared to that in oceanic
RPC provides the properties from same mature cell, as areas, whereas WGs (R-1) CFAD is mixture of land and
all Ze are connected, whereas MVC correspond to the ocean (below 5 km, like ocean whereas above 5 km conti-
strongest mature cell present at the time of observation. nental type). Oceanic areas (R-5, R-6, and R-7) show less
Maximum value of Ze represent the storm type (Zipser width compared to land areas and indicate the less intense
and Lutz, 1994) and so MVCs represent the intense precip- precipitation. The CFAD for RPC (Fig. 5b) shows similar
itation area in each CSs. Cloud cells with echo base height characteristics, and BOB (R-5 and R-6) and AS (R-7) show
P3 km may be the part of anvil or stratiform clouds the least width compared to land dominated areas.
(Dutta et al., 2014; Li and Schumacher, 2011).
3.2. Cloud top height
3. Results
Cloud top height is defined as the maximum height of
The study region (Fig. 1) encompasses, 10,800 satellite Ze P 20 dBZ (MH20) whereas maximum height of 30
passes during the entire study period. The number of CSs (MH30) and 40 dBZ (MH40) are the proxy for convective
varies by a factor of nearly 3, 2.5, 2.5 and 4 times in the intensity (Zipser et al., 2006; Xu and Zipser, 2012), as
boxes chosen with R-4 and R-7 have extreme occurrences height of 30(40) dBZ shows the 0.5(0.6) correlation with
(Fig. 4). R-4 (head of Bay) peaks with 679 CSs followed the vertical velocity (Heymsfield et al., 2010). Cloud top
by WGs (R-1, 550), land areas (300, R-2 and R-3), height is calculated at each 0.25 km altitude interval.
BOB (R-6, 200) and the minimum CSs are in AS (R-7, Fig. 6 shows cumulative (a and b) and frequency (c and
233) in each unit square grid box. d) distribution of MH20, MH30 and MH40 for MVC
and RPC. Oceanic areas (AS and BOB) show higher frac-
3.1. Contoured frequency by altitude diagram (CFAD) tion of low level clouds. Mode for MH20, MH30 and
MH40 is 8, 6 and 4–5 km respectively for MVC. Frequency
The CFAD of MVC and RPC are shown in Figs. 5(a distribution of RPC shows single mode at 4, 6 and 8 km for
and b) (Yuter and Houze, 1995). Broadly the CFAD can MH40, MH30 and MH20 respectively. WGs and AS show
be grouped into the two regions, below and above 0°C mode below 5 km, whereas other areas show the mode
(5 km). The CFAD shows the maximum width below above 5 km in MH40. AS has the least fraction of low level
0°C which indicate the higher precipitation intensity in clouds among all the areas. MH40 shows double mode
both RPC and MVC. Lower Ze at high altitudes shows compared to MH30 and MH20. MH20 for MVC does

Fig. 4. No. of cases of CSs and for different reflectivity thresholds in each selected areas. From left to right number of cases correspond to 17, 30, 35 and
40 dBZ are shown.
 S. Kumar / Advances in Space Research 58 (2016) 1017–1032 1023

17.5 R−1 R−3

R−2 R−4
15.0
12.5
10.0
7.5
Height (Km)

5.0
2.5
0.0 20 30 40 50 20 30 40 50 20 30 40 50
20 30 40 50
1
17.5
R−5 R−6 R−7
15.0 0.8

12.5 0.6
10.0
0.4
7.5
5.0 0.2
2.5
0
0.0 20 30 40 50 20 30 40 50
20 30 40 50 Reflectivity (dBZ)

Fig. 5a. Contoured frequency by altitude diagram for MVC. Colorbar shows the relative frequency of occurrence of Ze at different altitude at different
reflectivity bins. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

17.5 R−2
R−1 R−3 R−4
15.0
12.5
10.0
7.5
5.0
Height (Km)

2.5
0.0
20 30 40 50 20 30 40 50 20 30 40 50 20 30 40 50
1
17.5
R−5 R−6 R−7
0.8
15.0
12.5 0.6
10.0
7.5 0.4

5.0
0.2
2.5
0.0 0
20 30 40 50 20 30 40 50 20 30 40 50
Reflectivity (dBZ)

Fig. 5b. Contoured frequency by altitude diagram for RPC. Colorbar shows the relative frequency of occurrence of Ze at different altitude at different
reflectivity bins. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

not show any regional differences and all the cloud cells are (MH20) shows that around 50% MVCs are above 10 km
crossing 3.5 km altitude. Frequency of cloud top height altitude in R-3, whereas corresponding height over AS is
decreases after 8 km altitude, and land area (R-3) shows 6 km. MH30 and MH40 also show the similar trends and
higher fraction after 15 km. Cumulative frequency distribu- land areas consist the highest fraction of deep clouds.
tion shows higher regional differences compared to fre- MH40 shows the highest regional differences and 50%
quency distribution. Cumulative frequency distribution MVCs are more than 6 km over land areas and least over
1024 S. Kumar / Advances in Space Research 58 (2016) 1017–1032

Fig. 6. Cumulative and frequency distribution of maximum height for 40, 30 and 20 dBZ in MVC (a and c) and RPC (b and d). Colours are same for
different areas as indicated only in one panel. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of
this article.)

AS with 4 km altitude. Cumulative frequency of RPC majority of the cases. Some profiles go beyond 18 km,
shows less regional differences compared to MVC, but with whereas some are well below 6 km. Average vertical pro-
similar trends. files for MVCs and RPCs are shown in Fig. 7. The land
areas (R-2 and R-3) show the higher Ze compared to ocea-
3.3. Average vertical profile nic areas (R-5, R-6 and R-7). Fig. 7(a;b) shows that on an
average Ze for the RPCs and MVCs are always higher for
Supplementary Fig. 1 shows the individual vertical pro- R-2 and R-3 from ground to top, followed by head of Bay
files for RPC over R-1, R-3 and R-4. Generally, maximum (R-4) and BOB (R-5). Below the 10 km, Ze difference
Ze is observed below 6 km and decreases rapidly above in a is  5 dBZ (0.08 mm/hr) among the extreme profiles.
 S. Kumar / Advances in Space Research 58 (2016) 1017–1032 1025

16 R−1 16 R−1
(a)MVC R−2 (b)RPC R−2
R−3 R−3
14 14
R−4 R−4
R−5 R−5
12 R−6 12 R−6

Height (Km)
R−7 R−7
10 10

8 8

6 6

4 4

2 2
20 25 30 35 40 45 20 25 30 35 40
Reflectivity (dBZ)

Fig. 7. The mean vertical profile for Ze in MVC (a) and RPC (b). Only those vertical levels are shown, which consist the 10% data points from the
population of MVCs and RPCs.

AS (R-7) has the least average Ze among all the areas. Head of Bay (R-4) shows the least ACRH between 2 and
Average Ze near WGs (R-1) is oceanic type below 0°C 4 km. For CSs higher than 1500 km2, ACRHs are always
and land type above the 0°C, where it has Ze value equal less than 10%. Higher differences in ACRH are observed
to land. The differences are higher above the 6 km. Average among ocean and land, and below 1500 km2 in (3 km and
vertical profiles belong to oceanic areas over Bay and AS 40 dBZ). ACRH for 30 dBZ is maximum near 2 km, and
(R-5, R-6 and R-7) show a sharp decrease in Ze compared gradually decreases at higher altitude, irrespective of areas.
to land areas. Regional differences are higher in MVCs in Decreasing rate of ACRH is highest in head of Bay (R-4),
comparison to RPCs below 5 km. compared to other subareas (Fig. 8b). Above 0°C (5 km),
ACRH decreases very rapidly (from 40% to 21%), and
reveals that how the precipitation intensity decreases above
3.4. Area covered of different reflectivity thresholds (ACR)
the freezing level in the CSs.
and at different altitude (ACRH)

For a given CSs, ACR (e.g., Ath/A17) is highest for 3.5. Three dimensional cloud vertical structure analysis using
30 dBZ, followed by 35 and 40 dBZ in all the areas cloud liquid water content
(Fig. 8), and varies 1–3 times for 30, 35 and 40 dBZ for dif-
ferent CSs area. For small CSs (200–500 km2), ACR is Fig. 9 shows the average of normalized CLW for
always higher. 30 dBZ ACR covers 50% cloud area for 3DCVS for type ‘a’ and type ‘a’ + b + ‘c’ cloud over head
600 km2, whereas corresponding coverage is 30% for of Bay. Line without marker is for type ‘a’ radar reflectivity
35 dBZ. The areal width for ACR correspond to 40– profiles whereas lines with the circle are for type ‘a + b + c’
50%, 30–40% and 10–20% for 30, 35 and 40 dBZ, irre- cloud. The effect of vertical profiles with echo base height
spective of different subareas and areas of CSs (Fig. 8a). P3 km is observed, when CSs are distinguish by a dimen-
ACR for 30 and 35 dBZ never comes down below 40% sionless ratio (r). Ratio (r) is defined as normalized CLW
and 20% irrespective of areas. Regional differences are (4 km)/CLW(2 km), and distinguish the CSs with
higher for 40 dBZ between 200 km2 and 2000 km2. Regio- (r P 0.75) and without (r < 0.75) bright band respectively.
nal differences are very less for 30 dBZ (<5%), followed Type b and c cloud enhances the CLW above 0°C (near 4–
by 35 dBZ (<10%) and 40 dBZ (<20%). Close observation 5 km), for all the ratios but with different magnitude. Ratio
reveals that eastern BOB has highest fraction of ACR, (r4) P 1, shows the sign of pronounced signature of bright
whereas WGs has the least. From the observation it is clear band. Insat figure shows that how the type ‘b’ and ‘c’
that small CSs consist more intense rain (higher values of clouds affect the CLW in lower and upper troposphere.
Ze) and as the cloud area increases lower reflectivity values Fig. 10a shows CFAD for 3DCVS for different ratio’s
contribute. Almost equal ACR reveals that precipitation and shows the probability of Ze at different altitudes in
mechanism may be same over different areas. the CSs for different ratios. Each CFAD shows the different
ACRH is calculated by evaluating ACR (i.e. Ath/A17) at characteristics, as higher probability of Ze occurs within
2 km, 3 km 4 km and 5 km respectively (Fig. 8b). Observa- different altitude range. For r1 the maximum probability
tions show that ACRH decreases as CSs area increases. of Ze is just below the 5 km altitude for all the areas and
ACRH for 30, 35 and 40 dBZ is always less than 40%, indicate the intense precipitation comes from below the
20% and 10% respectively, irrespective of height, whereas freezing level. For r2, maximum probability of Ze occurs
at 5 km, ACRH for 35 dBZ is less than 10% (Fig. 8b). below as well as above the freezing level and shows that
1026 S. Kumar / Advances in Space Research 58 (2016) 1017–1032

1
(a) 30 dBZ (b)
35 dBZ
0.8

0.6

0.4

0.2
17
A /A

0
1 2−5 6−10 11−15 16−20 >21 1 2−5 6−10 11−15 16−20 >21
th

1
(c)
0.8 40 dBZ R−1
R−2
0.6 R−3
R−4
0.4 R−5
R−6
0.2 R−7

0
1 2−5 6−10 11−15 16−20 >21
2
Cloud system area (100*Km )

Fig. 8a. Fraction area covered by reflectivity thresholds (a) 30 dBZ, (b) 35 dBZ and (c) 40 dBZ, in CSs for different areas.

0.8 2 km & 30 dBZ 3 km & 30 dBZ 4 km & 30 dBZ 5 km & 30 dBZ

0.6

0.4

0.2

0
Area(th)/Area (17dBZ)

0.8 2 km & 35 dBZ 3 km & 35 dBZ 4 km & 35 dBZ 5 km & 35 dBZ

0.6

0.4

0.2

0
1

0.8 2 km & 40 dBZ 4 km & 40 dBZ 5 km & 40 dBZ

3 km & 40 dBZ
0.6

0.4

0.2

0
1 2−5 6−10 11−15 16−20 >21 1 2−5 6−10 11−15 16−20 >21 1 2−5 6−10 11−15 16−20 >21 1 2−5 6−10 11−15 16−20 >21
2
Area of CSs (100*km )

Fig. 8b. Fraction area covered by reflectivity thresholds at different altitudes in CSs. Colours are same for different areas and similar to used in Fig. 8a.
(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

ice microphysics involve in precipitation processes. For r3 (near 0°C) over all areas. This indicate the CSs in their
and r4 maximum probability of Ze occurs above the initial stage (Szoke and Zipser, 1986), and collision-
freezing altitude, but r4 shows a clear picture of bright coalescence process dominates and has the similar nature
band as Ze probability shows a peak near freezing altitude. over all the areas. Land and oceanic areas show different
Fig. 10b shows the CFAD over the other areas and they characteristics for r2–r4. The CFAD for r2 shows the
also show the similar patterns. r1 shows the similar charac- two maxima mode over land (R-2 and R-3), and coastal
teristics as higher width occurs below the 5 km altitude area (R-1 and R-4), and single maxima mode over oceanic
 S. Kumar / Advances in Space Research 58 (2016) 1017–1032 1027

Fig. 9. Average of normalized CLW of the cloud systems based on different ratios’ in region-1. Bright band signature appears when ratio (see the text) is
more than 0.75.

15.0 15.0
12.5 r<=0.5 12.5 0.5<r<0.75

10.0 10.0
7.5 7.5

5.0 5.0

2.5 2.5
Height (km)

17 27 37 47 17 27 37 47
15 15
0.75=<r<1.0 r>=1.0
12.5 12.5

10 10

7.5 7.5

5.0 5.0

2.5 2.5

17 27 37 47 17 27 37 47
Reflectivity (dBZ)

Fig. 10a. Contoured frequency by altitude diagram for different ratios over Western Ghats. CFAD shows the signature of the bright band when r > 1.

areas (R-5, R-6 and R-7). The CFAD for r2 shows that indicate the evolution of bright band and stratiform
intense precipitation sustains for longer time over the land precipitation (Houze et al., 2007) in the CSs. The CFAD
and coastal areas, whereas intense precipitation dies early for r4 is similar to that r3 but peak in Ze is more pro-
over the oceanic areas. This also shows the intense precip- nounced near 0°C (5 km) and shows the highest signature
itation over land and coastal dominated areas have higher of bright band and CSs may be in decaying stage. We do
fraction of precipitation, which comes from above the not track the life cycle but a broad features of life cycle
freezing level compared to oceanic areas, and ice micro- of CSs have been observed through the CLW and CFAD
physics dominate. The CFAD for r3 shows the maxima profiles. The characteristics of CSs look similar over differ-
mode only above the 0°C, as well as the shifting of Ze peak ent areas compared to the vertical and areal structure of
towards the higher altitude over all areas near 5 km. This precipitation.
1028 S. Kumar / Advances in Space Research 58 (2016) 1017–1032

Fig. 10b. Contoured frequency by altitude diagram for different ratios over other areas. CFAD shows almost similar patterns over the different areas.

4. Discussion show the effect of mixed-phase dynamics (Heymsfield

et al., 2010; Xu and Zipser, 2012).
The properties of precipitating CSs, based on 10 years of
TRMM PR data is presented during Indian summer mon- 4.1. Land vs ocean
soon. In particular, radar defined properties of CSs, MVC,
RPC, ACR, ACRH and 3DCVS over Indian region are Strongest (higher Ze at higher altitude) RPC and MVC
looked into in this work. MVC and RPC are the most correspond to land areas, and consistent with the earlier
intense cloud cells, and exhibit both convective and strati- studies reported by (e.g., Zipser and Lutz (1994), Zipser
form precipitation. Several factors influence the vertical et al. (2006), Bhat and Kumar (2015), Kumar (2015),
and horizontal distribution of hydrometeors. ACR, ACRH Kumar and Bhat, 2016a) whereas oceanic areas show the
and 3DCVS provide the information of evolution and areal weakest (low Ze values) vertical profiles and Ze decreasing
coverage of CSs. Surface properties are important in draw- rapidly above 4–5 km altitude. The highest differences in
ing cloud properties. Sherwood et al. (2004) explained the the vertical distribution of Ze are observed in the mid-
requirement of moist air for deep convection. Fig. 11a troposphere (5–12 km range, (Fig. 7a;b), as average Ze in
shows the spatial pattern of surface moist static energy land areas (R-2 and R-3) are 6–10 dBZ higher compared
(MSE) derived from NCEP reanalysis data. Head of Bay to that over the AS (R-7) and BOB (R-5 and R-6). Cloud
shows higher number of CSs and coincides with higher sur- top height (MH20), MH30 and MH40 also show more
face MSE, but not consistent with vertical profile. AS and deep clouds over land (R-2 and R-3) compared to oceanic
southern BOB show lower surface MSE and coincide with areas (R-5, R-6 and R-7). Cloud condensation nuclei
less CSs and the weakest (lowest Ze) vertical profile. But it (CCN) concentration, their size distribution (Rosenfeld
is not true for all the cases. Fig. 11b shows vertical profile et al., 2008; Konwar et al., 2012) and updraft velocity
of MSE over selected areas derived from ERA-Interim (Lucas et al., 1994a) are important in defining the cloud
data. Central India (R-2 and R-3) shows the strongest top height as well as higher reflectivity values at higher alti-
(highest Ze) vertical profile between mid and upper tropo- tude. Land areas have stronger updrafts (e.g., Lucas et al.,
sphere consistent with the higher MSE. This may partially 1994b) and could explain the higher hydrometeors at
explain the stronger vertical profile in central India. Fig. 12 higher altitudes. Smaller concentration of CCN with salty
(a;b) shows the column water vapour and column liquid CCN over ocean (Pruppacher and Klett, 1997) favours
water derived from NCEP reanalysis data. Head of Bay the warm rain processes (Konwar et al., 2012). Higher
consists higher number of CSs and matches with higher CCN concentration over land delays the precipitation pro-
amount of column water vapour and liquid water. The cess (Rosenfeld et al., 2008; Konwar et al., 2012) and pro-
highest regional differences in average vertical profiles vides higher water at higher vertical levels (e.g., Wang
and cloud top height occur in 5–12 km altitude range and (2005)).
 S. Kumar / Advances in Space Research 58 (2016) 1017–1032 1029

Fig. 11a. Spatial distribution of surface moist static energy derived from NCEP reanalysis data for JJAS months during 2001–2010. Unit is in kJ kg 1.

summer, salty aerosols are observed over the AS, and the
100 aerosol optical depth increases exponentially with wind
R−1
150 R−2 speed (e.g., Vinoj and Satheesh, 2003). Further, water in
R−3 the AS is more saline (36 psu, Rao and Sivakumar,
200 R−4
R−5 2003) compared to that over the BOB (29 psu, Bhat
250 R−6 et al., 2001). AS consists high concentration of land origin
R−7
CCN along with some giant CCN, whereas, it is more in
Pressure (mbar)

300
original conditions over the BOB (Konwar et al., 2012).
400 The resulting wider size distribution of the droplet spectra
over the AS in the presence of giant CCN suits the
500 collision-coalescence mechanism (Konwar et al., 2012),
and condensed water precipitate out early compared to
600 Bay clouds.

700
775 4.3. Two and three dimensional characteristics of cloud
850 systems
925
1000
330 340 350 360 370
Due to the high sensitivity PR misses  50% of the rain
MSE(kJ kg−1) area higher than 0 dBZ (Schumacher and Houze, 2000),
and this could affect the analysis of area coverage. ACR
Fig. 11b. Vertical distribution of surface moist static energy derived from
decreases exponentially as the area of CSs increase and
ERA reanalysis data for JJAS months during 2001–2010. Unit is in
kJ kg 1. reveals that larger CSs area consist of less intense precipi-
tation (low reflectivity value). ACRH shows two ways of
observations (Fig. 8b), one for constant Ze (left to right),
4.2. Comparison between Arabian Sea and Bay of Bengal and other for constant height (top to bottom). Ze thresh-
olds variation are more sensitive compared to changing
Figs. 5–8 reveals the BOB (R-5, R-6) shows higher altitude, as we move from 2 to 5 km, ACRH changes
intense precipitation compared to AS (R-7), as AS has less 5% for all the thresholds, whereas if Ze threshold varies
Ze in average vertical profile and higher fraction of low at constant altitude, ACRH changes by 20–30%
level clouds (Fig. 6). High wind speed (8–16 ms 1) per- (Fig. 8b). This critical information reveals that precipita-
sists over the AS throughout the monsoon season following tion formation mechanism, which controls the size and dis-
the monsoon onset (Weller et al., 1998), whereas, it fluctu- tribution of raindrops are different among the selected
ates between high and low values on intra-seasonal time areas, for Ze thresholds but almost similar at constant alti-
scales over the BOB (Bhat et al., 2001). During the tudes. 3DCVS does not show the regional differences and
1030 S. Kumar / Advances in Space Research 58 (2016) 1017–1032

Fig. 12a. Column water vapour derived from NCEP data for JJAS months during 2001–2010. Unit is in kg m 2.

Fig. 12b. Column liquid water vapour derived from NCEP data for JJAS month during 2001–2010. Unit is in kg m 2.

bright band feature is observed in CFAD for r > 1, both difference in CFAD for r2. We see the CFAD over the land
over land and oceanic areas (Figs. 10a and b), and shows dominated aress (R-2 and R-3) and coastal areas (R-1 and
the similar evolution of CSs over all the selected areas. A R-4) show the higher probability of Ze below as well as
close similarity in the CFAD for different ratio’s reveals above the freezing level. This may represents that ice
that intense precipitation areas follow the similar evolu- microphysics such as riming and warm rain processes such
tion, either they have bright band or not. Whenever as collision–coalescence contribute the higher intense pre-
CLW shows a peak near the freezing level, 3DCVS exhibits cipitation over the land and coastal areas compared to
the bright band. Land and oceanic areas show the oceanic area.
 S. Kumar / Advances in Space Research 58 (2016) 1017–1032 1031

5. Conclusions Dee, D.P., Uppala, S.M., Simmons, A.J., Berrisford, P., Poli, P.,
Kobayashi, S., Andrae, U., Balmaseda, M.A., Balsamo, G., Bauer,
P., Bechtold, P., 2011. The ERA-Interim reanalysis: configuration and
Using the TRMM-PR 2A25 data, the properties of three performance of the data assimilation system. Q. J. R. Meteorol. Soc.
dimensional precipitating cloud systems are studied over 137, 553–597.
seven climatologically important areas in the Indian region. Dixon, M., Wiener, G., 1993. TITAN: thunderstorm identification,
Properties of these precipitating cloud systems include the tracking, analysis, and nowcasting – a radar base methodology. J.
vertical structure, areal distribution and evolution of cloud Atmos. Ocean Technol. 10 (6), 785–797.
Dutta, D., Sarma, D.K., Sharma, S., 2014. A multisensor analysis of the
systems. Following are the major conclusions of the study: life cycle of bow echo over Indian region. Int. J. Atmos. Sci., 207064
Greene, D.R., Clark, R.A., 1972. Vertically integrated liquid water – a
(1) Individual vertical profiles (1-dimensional character- new analysis tool. Mon. Weather Rev. 100 (7), 548–552.
istics) show higher regional differences, followed by Goswami, B.N., Ajayamohan, R., Xavier, P.K., Sengupta, D., 2003.
area covered by different reflectivity thresholds (2- Clustering of synoptic activity by Indian summer monsoon intrasea-
sonal oscillations. Geophys. Res. Lett. 30 (8), 141–144. http://dx.doi.
dimensional characteristics), whereas three dimen- org/10.1029/2002GL016734.
sional cloud vertical structures show the least regio- Heymsfield, G.M., Tian, L., Heymsfield, A.J., Li, L., Guimond, S., 2010.
nal differences. Characteristics of deep tropical and subtropical convection from nadir-
(2) Probability of intense precipitation is higher over viewing high-altitude airborne Doppler radar. J. Atmos. Sci. 67, 285–
land in comparison to ocean, as average vertical 308.
Hirose, K., Nakamura, K., 2005. Spatial and diurnal variation of
reflectivity profile has higher reflectivity value at precipitation systems over Asia observed by the TRMM precipitation
higher altitude. On an average intense precipitation radar. J. Geophys. Res. 110 (D05106), 2004J. http://dx.doi.org/
is least over Arabian Sea and highest over central 10.1029/D004815.
India. Probability of warm rain is highest over Wes- Hoskins, B.J., Karoly, D.J., 1981. The steady linear response of a spherical
tern Ghats and Arabian Sea. atmosphere to thermal and orographic forcing. J. Atmos. Sci. 38 (6),
1179–1196.
(3) Small cloud systems consist higher area of intense Houze Jr., R.A., 1989. Observed structure of mesoscale convective systems
precipitation (higher Ze) compared to large cloud sys- and implications for large-scale heating. Q. J. R. Meteorolog. Soc. 115,
tems, and follow the similar trends over land and 425–461.
oceanic areas. Different reflectivity threshold has Houze Jr., R.A., 1993. Cloud Dynamics. Academic Press, San Diego, p.
nearly equal precipitation area at the same altitude 573.
Houze Jr., R.A., Wilton, D.C., Smull, F.B., 2007. Monsoon convection in
in different areas. the Himalayan region as seen by the TRMM precipitation radar. Q. J.
(4) Three dimensional cloud vertical structure shows the R. Meteorolog. Soc. 133, 1389–1411.
similar climatology over land and oceanic areas and Huffman, G.J., Bolvin, D.T., Nelkin, E.J., Wolff, D.B., Adler, R.F., Gu,
all the areas show the signature of bright band near G., Hong, Y., Bowman, K.P., Stocker, E.F., 2007. The TRMM
5 km altitude. Land and coastal areas show the multisatellite precipitation analysis (TMPA): quasi-global, multiyear,
combined-sensor precipitation estimates at fine scales. J. Hydromete-
higher probability of intense precipitation above the orol. 8 (1), 38–55.
freezing level. Kalnay, E., Kanamitsu, M., Kistler, R., Collins, W., Deaven, D., Gandin,
L., Iredell, M., Saha, S., White, G., Woollen, J., Zhu, Y., 1996. The
NCEP/NCAR 40-year reanalysis project. Bull. Am. Meteorol. Soc. 77
Appendix A. Supplementary data (3), 437–471.
Konwar, M., Maheskumar, R.S., Kulkarni, J.R., Freud, E., Goswami, B.
N., Rosenfeld, D., 2012. Aerosol control on depth of warm rain in
Supplementary data associated with this article can be convective clouds. J. Geophys. Res. 117, D13204. http://dx.doi.org/
found, in the online version, at http://dx.doi.org/10.1016/ 10.1029/2012JD017585.
j.asr.2016.05.052. Kumar, S., 2015. A 10-year climatology of vertical properties of most
active convective clouds over the Indian regions using TRMM PR.
Theor. Appl. Climatol. http://dx.doi.org/10.1007/s00704-015-1641-5.
References kumar, S., Bhat, G.S., 2016a. Vertical profiles of radar reflectivity factor in
intense convective clouds in the tropics. J. Appl. Meteorol. Climatol.
Alcala, C., Dessler, A., 2002. Observations of deep convection in the http://dx.doi.org/10.1175/JAMC-D- 15-0110.1.
tropics using the tropical rainfall measuring mission (TRMM) Kumar, S., Bhat, G.S., 2016b. Active shallow and mid level clouds
precipitation radar. J. Geophys. Res. 107 (D24), 4792. observed over South Asia during the summer monsoon season. J.
Bhat, G.S., Kumar, S., 2015. Vertical structure of cumulonimbus towers Earth Syst. Sci., JESS-D-16-00190
and intense convective clouds over the South Asian region during the Kummerow, C., Barnes, W., Kozu, T., Shiue, J., Simpson, J., 1998. The
summer monsoon season. J. Geophys. Res. Atmos. 120. http://dx.doi. tropical rainfall measuring mission (TRMM) sensor package. J.
org/10.1002/2014JD022552. Atmos. Ocean Technol. 15, 809–817.
Bhat, G.S., Gadgil, S., Hareesh Kumar, P.V., Kalsi, S.R., Madhu- Li, W., Schumacher, C., 2011. Thick anvils as viewed by the TRMM
soodanan, P., Murty, V.S.N., Prasada Rao, C.V.K., Ramesh Babu, V., precipitation radar. J. Clim. 24, 1718–1735.
Rao, L.V.G., Rao, R.R., Ravichandran, M., 2001. BOBMEX: the Bay Liu, C., Zipser, E.J., Nesbitt, S.W., 2007. Global distribution of tropical
of Bengal monsoon experiment. Bull. Am. Meteorol. Soc. 82 (10), deep convection: different perspectives from TRMM infrared and
2217–2244. radar data. J. Clim. 20, 489–503.
Byers, H.R., Braham Jr., R.R., 1948. Thunderstorm structure and Liu, C., Zipser, E.J., Cecil, D.J., Nesbitt, S.W., Sherwood, S., 2008. A
circulation. J. Meteorolog. 5 (3), 71–86. cloud and precipitation feature database from nine years of TRMM
1032 S. Kumar / Advances in Space Research 58 (2016) 1017–1032

observations. J. Appl. Meteorol. Climatol. 47 (10), 2712–2728. http:// Schumacher, C., Houze Jr., R.A., 2003. Stratiform rain in the tropics as
dx.doi.org/10.1175/2008JAMC1890.1. seen by the TRMM precipitation radar. J. Clim. 16, 1739–1756.
Liu, C., Cecil, D.J., Zipser, E.J., kronfeld, K., Robertson, R., 2012. Schumacher, C., Houze Jr., R.A., 2000. Comparison of radar data from
Relationships between lightning flash rates and radar reflectivity the TRMM satellite and Kwajalein oceanic validation site. J. Appl.
vertical structures in thunderstorms over the tropics and subtropics. J. Meteorol. 39, 2151–2164.
Geophys. Res. 117 (D06212), 2011J. http://dx.doi.org/10.1029/ Sherwood, S.C., Minnis, P., McGill, M., 2004. Deep convective cloud-top
D017123. heights and their thermodynamic control during crystal-face. J.
Leary, C.A., Houze Jr, R.A., 1979. The structure and evolution of Geophys. Res. 109. http://dx.doi.org/10.1029/2004JD004811.
convection in a tropical cloud cluster. J. Atmos. Sci. 36, 437–457. Szoke, E.J., Zipser, E.J., 1986. A radar study of convective cells in
Lucas, C., Zipser, E.J., Lemone, M.A., 1994a. Vertical velocity in oceanic mesoscale systems in GATE. Part II: Life cycles of convective cells. J.
convection off tropical Australia. J. Atmos. Sci. 51 (21), 3183–3193. Atmos. Sci. 43, 199–218.
Lucas, C., Zipser, E.J., LeMone, M.A., 1994b. Convective available Takayabu, Y.N., 2002. Spectral representation of rain profiles and diurnal
potential energy in the environment of oceanic and continental clouds; variations observed with TRMM PR over the equatorial area.
correction and comments. J. Atmos. Sci. 51 (24), 3829–3830. Geophys. Res. Lett. 29 (12), 1584. http://dx.doi.org/10.1029/
Nesbitt, S.W., Zipser, E.J., Cecil, D.J., 2000. A census of precipitation 2001GL014113.
features in the tropics using TRMM: radar, ice scattering, and Vinoj, V., Satheesh, S.K., 2003. Measurements of aerosol optical depths
lightning observations. J. Clim. 13, 4087–4106, http: //dx.doi.org/ over Arabian Sea during summer monsoon season. Geophys. Res.
10.1175/1520- 0442(2000) 13<4087:ACOPFI>2.0, Co;2. Lett. 30 (5), 1263. http://dx.doi.org/10.1029/2002GL016664.
Nesbitt, S.W., Cifelli, R., Rutledge, A., 2006. Storm Morphology and Wang, C., 2005. A modeling study of the response of tropical deep
Rainfall Characteristics of TRMM Precipitation Features. Mon. convection to the increase of cloud condensation nuclei concentration:
Weather Rev. 134, 2702–2721. 1. Dynamics and microphysics. J. Geophys. Res. 110 (D21). http://dx.
Petersen, W.A., Rutledge, S.A., 2001. Regional variability in tropical doi.org/10.1029/2004JD005720, D21, 211.
convection: observation from TRMM. J. Clim. 13, 3566-2586. Weller, R.A., Baumgartner, M.F., Josey, S.A., Fischer, A.S., Kindle, J.C.,
Pruppacher, H.R., Klett, J.D., 1997. Microphysics of Clouds and Precip- 1998. Atmospheric forcing in the Arabian Sea during 1994–1995.
itation, second ed. Kluwer Academic Publishers, London, p. 954. observations and comparisons with climatology and models. Deep Sea
Qie, X.S., Wu, X.K., Yuan, T., Bian, J.C., Lu, D.R., 2014. Comprehensive Res. Part II 45, 1961–1999.
pattern of deep convective systems over the Tibetan Plateau-South Williams, E.R., Weber, M.E., Orville, R.E., 1989. The relationship
Asian monsoon region based on TRMM data. J. Clim. 27, 6612–6626. between lighting type and convective state of thunderstorms. J.
http://dx.doi.org/10.1175/JCLI-D-14-00076.1. Geophys. Res. 94, 13213–13220.
Randall, D.A. et al., 2007. Climate models and their evaluation. In: Xu, W., Zipser, E.J., 2012. Properties of deep convection in tropical
Solomon, S. (Ed.), Climate Change 2007: The Physical Science Basis. continental, monsoon, and oceanic rainfall regimes. Geophys. Res.
Contribution of Working Group I to the Fourth Assessment Report of Lett. 39, L07802. http://dx.doi.org/10.1029/2012GL051242.
the Intergovernmental Panel on Climate Change. Cambridge Univer- Yuan, T., Qie, X., 2008. Study on lightning activity and precipitation
sity Press, Cambridge, United Kingdom and New York, USA. characteristics before and after the onset of the South China Sea
Rao, R.R., Sivakumar, R., 2003. Seasonal variability of sea surface summer monsoon. J. Geophys. Res. 113, D14101. http://dx.doi.org/
salinity and salt budget of the mixed layer of the north Indian Ocean. 10.1029/2007JD009382.
J. Geophys. Res. 108 (C1), 3009. http://dx.doi.org/10.1029/ Yuter, S.E., Houze Jr., R.A., 1995. Three-dimensional kinematic and
2001JC000907. microphysical evolution of Florida cumulonimbus: Part II. Frequency
Riehl, H., 1979. Climate and Weather in the Tropics. Academic Press, distribution of vertical velocity, reflectivity, and the differential
London, xii + 611 pp. reflectivity. Mon. Weather Rev. 123, 1941–1963.
Romatschke, U., Houze Jr., R.A., 2011. Characteristics of precipitating Zipser, E.J., Lutz, K.R., 1994. The vertical profile of radar reflectivity of
convective systems in the South Asian monsoon. J. Hydrometeorol. 12, convective cells: a strong indicator of storm intensity and lightning
3–26. http://dx.doi.org/10.1175/2010JHM1289.1. probability? Mon. Weather Rev. 122, 1751–1759.
Rosenfeld, D., Lohmann, U., Raga, G.B., O’Dowd, C.D., Kulmala, M., Zipser, E.J., Cecil, D.J., Liu, C., Nesbitt, S.W., Yorty, D.P., 2006. Where
Fuzzi, S., Reissell, A., Andreae, M.O., 2008. Flood or drought: how do are the most intense thunderstorms on Earth? Bull. Am. Meteorol.
aerosols affect precipitation? Science 321, 1309–1313. Soc. 87, 1057–1071.
Steiner, M., Houze, R.A., Yuter, S.E., 1995. Climatological characteriza-
tion of three-dimensional storm structure from operational radar and
rain guage data. J. Appl. Meteorol. 34, 1978–2007.