Content uploaded by S. C. Bhan
Author content
All content in this area was uploaded by S. C. Bhan on Jan 31, 2015
Content may be subject to copyright.
Content uploaded by S. C. Bhan
Author content
All content in this area was uploaded by S. C. Bhan on Dec 29, 2014
Content may be subject to copyright.
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 90, NO. 10, 25 MAY 2006
1344
*For correspondence. (e-mail: rjenamani1@yahoo.co.in)
Observational/forecasting aspects of the
meteorological event that caused a record
highest rainfall in Mumbai
Rajendra Kumar Jenamani*, S. C. Bhan and S. R. Kalsi
Office of the Director General of Meteorology, Mausam Bhavan, India Meteorological Department, Lodhi Road, New Delhi 110 003, India
The Santacruz observatory at Mumbai airport experi-
enced unprecedented rainfall of 94.4 cm on 26 and 27
July 2005, an all-time highest record over the city. The
country’s commercial capital came to a complete standstill
due to severe flooding/deluge. It also caused severe
damage to life and property. In the present article, we
have brought out the characteristics of this unique
rainfall event, compared it with past intense rainfall
events in India and other places in the world, and also
analysed various data to find possible causes of occur-
rence of this event. Technology available for predic-
tion of such events is also briefly touched upon.
Keywords: Isohyet analysis, synoptic scale feature,
scale interaction, thunderstorm.
Mumbai (lat. 18.50°N, long. 72.52°E), formerly called
Bombay, situated on the west coast of India, is the com-
mercial and financial capital of the country. It generates
about 5% of India’s GDP and contributes over one-third
of the country’s tax revenues. It is among the few port
cities of the world having major industrial and economic
activities. The influx of labour and other work force over
the last two decades has resulted in the total population of
the metropolitan and the surrounding suburbs to increase
from 9.9 million in 1981 to 13.0 million in 1991 and 17.7
million in 2001. Mumbai is not only the most populated
city in India, but also the fourth most populated city in the
world after Tokyo (36.0 million), New York (22.0 million),
and Seoul (21.7 million). If only the population of the
metropolitan city, excluding its surrounding suburbs is
considered, Mumbai is the second most populated city in
the world, after Shanghai. Such densely populated cities
with highly developed infrastructure are always vulnerable
to natural disasters. The economic losses are also expected
to increase over time, as people move into more vulner-
able areas because of shortage of free space.
Parts of Mumbai experienced an unprecedented rainstorm
on 26 and 27 July 2005. The 24 h rainfall ending at 0830 h
IST of 27 July was 94.4 cm as per the meteorological ob-
servatory at Santacruz airport, Mumbai. Colaba, the other
meteorological observatory, which is roughly 27 km
away, recorded only 7.4 cm rain during the same period.
Very heavy rains also lashed other parts of Maharashtra
on subsequent days and completely disrupted normal life
across the state, with Mumbai being the most severely hit.
On the subsequent two days, though rainfall over Mumbai
reduced to 1.9 cm per day, it again increased to 20 cm on
1 August 2005; further worsening the situation over the
already deluged city. Mumbai was totally cut-off from the
remaining parts of the world almost for a week. Mumbai
airport was closed for two days on 26 and 27 July as the
runway was waterlogged, the terminal building flooded
and crucial navigation and landing aids damaged, thus
forcing all international and domestic flights to be diverted or
cancelled. Normal operations could resume only on 4
August 2005. Train services to and from Mumbai were
also cancelled for more than a week. Local trains which
are Mumbai’s life line – running every 3 min and ferrying
4.5 million passengers daily – were totally thrown off
gear as railway tracks were not only submerged, but also
damaged. Following were the initial damage assessments
due to this severe rainfall spell and flooding in the state,
as reported by Financial Times and Economic Times on 4
August 2005.
• The number of dead in the Maharashtra floods could
well be above 1000, with Mumbai alone accounting
for over 409.
• Around Rs 5000 crores estimated loss in the state,
with Mumbai accounting for half the amount.
• No electricity in certain areas of Mumbai city for up
to five days.
• One million people rendered homeless.
• About 1100 flights cancelled and airport closed for two
days.
• Five million mobile and land lines crashed.
In addition to the two observatories at Colaba and Santa-
cruz, Mumbai of India Meteorological Department (IMD),
the Maharashtra government and the Mumbai Municipal
Corporation have a separate network of rain gauges,
which also captured this intense rainfall event. The gauge
at Vihar Lake, located around 15 km northeast of Santacruz,
recorded 104.9 cm of rain on 27 July 2005. The measure-
ment of rainfall at Santacruz is of importance, particu-
larly in the field of national and international aviation.
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 90, NO. 10, 25 MAY 2006 1345
In the present article, we have brought out the charac-
teristics of this unique rainfall event that resulted in
94.4 cm rainfall over Santacruz on 27 July. An attempt is
made to examine spatial and temporal extent of this rain-
storm by pooling together all the rainfall reports on 27
July from other recording stations in the city and its suburbs,
and making an isohyetal analysis. We have also com-
pared this rainfall event with rainfall records available for
other places in India and elsewhere in the world. By ana-
lysing data from various sources, possible causes for such
exceptional rainfall have been investigated by examining
whether the event was purely a mesoscale system attributed to
thunderstorms having life period of a few hours or a syn-
optic scale event with intense atmospheric vortex/offshore
trough remaining with highly supporting, large-scale circula-
tion pattern already prevailing over the west coast of India
or an interaction between both the scales. We have also
referred to satellite and radar pictures to identify cloud-
development characteristics and discussed the possible
role of orography because of the presence of the Western
Ghats in the east in producing such heavy rainfall. Finally,
latest technological capabilities available in India and
abroad are discussed for predicting such important events.
Characteristics of the rainfall event
Normally, in July and August, many stations of Konkan
and Goa, including Mumbai and coastal Karnataka, which
are on the windward side of the Western Ghats of India,
receive high rainfall due to orographic effect. The strong
westerly moist wind from the Arabian Sea that strikes
perpendicularly to the hills extending roughly in a north-
south direction, is lifted up vertically during active monsoon
and causes very heavy rainfall on the windward side1,2.
Because of this orography, during many days in the mon-
soon season, strong westerly/southwesterly flow over the
Arabian Sea also results in the formation of offshore
trough/mesoscale vortices over the sea off the west coast,
causing very heavy rainfall activity along the west coast
of India, including Mumbai3. Strengthening of such westerly/
southwesterly flow over the Arabian Sea is observed
when the Arabian Sea branch of the monsoon is active or
when a depression/low pressure area forms over North
Bay of Bengal and moves to Central India1. Gujarat–Konkan
coast also experiences very heavy rainfall, up to 40 cm,
during active monsoon conditions due to the presence of
mid-tropospheric cyclone (MTC) over the region4.
Table 1 shows climatological frequency of heavy rainfall
distribution over the two main stations in Mumbai (Co-
laba and Santacruz) in different categories based upon
data of 1901–2004. It shows that there is a probability of
nearly 35–48% and 11–13% during each monsoon season
for Mumbai to have at least one day with rainfall of more
than 20 cm and more than 30 cm respectively. Hence
Mumbai is vulnerable to intense rainstorms during the
monsoon season.
The 24-h rainfall up to 0830 h IST on 27 July 2005
over Mumbai and its neighborhood is given in Figure 1.
On this day, not only did Santacruz record 94.4 cm, but
also Vihar Lake, which lies nearly 15 km northeast of
Santacruz, recorded higher amount of rainfall, i.e. 104.9 cm.
However, the amount decreased thereafter in the north-
east sector of Santacruz, with Bandhup and Tulsi reporting
rainfall of 81.5 and 60.1 cm respectively. Even stations
which are farthest northeast of Santacruz, e.g. Bhiwandi,
Thane and Kalyan, reported 75, 74 and 62 cm of rainfall
respectively. However, rainfall reduced significantly to a
few centimetres beyond 50 km to the northeast of San-
tacruz at lake stations, e.g. Tansa and Vaiterna reported 5
and 1 cm of rainfall respectively. South of Santacruz, Co-
laba recorded only 7.4 cm of rainfall. Isohytal contour
analysis in Figure 1 shows that the rainfall contour of
60 cm covered an area of nearly 35 × 20 km2 lying close
but to the northeast of Santacruz, while the highest con-
tour of 80 cm covered an area of nearly 10 × 10 km2, ly-
ing to the northeast, close to Santacruz.
From Figure 1, it is also interesting to note the occur-
rence of another very high rainfall area lying nearly 45–
50 km away to the southeast of Santacruz. In this area,
Matheran and Karjat recorded 84 and 69 cm of rainfall
respectively. Isohytal analysis also shows the location of
another rainfall maximum (> 60 cm) over the area covering
these stations having small spatial extent of 15 × 15 km2.
Normally in the monsoon season from June to September,
Colaba receives 205 cm, while Santacruz receives 230 cm
of rainfall; 12% higher than the former though both are
within 27 km distance. To understand the microclimato-
logical characteristic of rainfall distribution over different
parts of Mumbai, monthly rainfall totals over Vihar Lake,
Santacruz and Colaba with their correlation coefficients
(CC) are shown in Figure 2, using long period dataset
from IMD. The analysis shows that the June rainfall is a
little higher over Colaba than over Santacruz and Vihar
Lake, and vice versa in September. This is because Co-
laba in south Mumbai receives more rainfall during the
Table 1. Occurrence of very heavy rainfall events over Mumbai
No. of occasions with No. of occasions with rainfall No. of occasions No. of occasions with rainfall
rainfall more more than 20 cm/day with rainfall more than 30 cm/day
Observatory than 20 cm/day consecutively for two days more than 30 cm/day consecutively for two days
Colaba 50 5 13 2
Santacruz 37 3 11 1
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 90, NO. 10, 25 MAY 2006
1346
Figure 1. Rainfall observations (in cm) reported from different parts of Mumbai and its sub-
urbs on 27 July and isohytal analysis over a 24 h period.
Figure 2. Monthly rainfall climatology over different stations in Mumbai and their correlation coefficients.
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 90, NO. 10, 25 MAY 2006 1347
onset phase in June. However, in July and August, stations
in the north, including Santacruz and Vihar Lake receive
higher rainfall than Colaba and rainfall difference during
both the months is higher than during the other months.
The CC between monthly rainfalls show a higher value
between rainfall of Santacruz and Colaba than that between
the rainfall of Santacruz and Vihar in June, while their
CC are more or less the same for other monsoon months.
Actual values of their CC are 0.75 and 0.8 in July, 0.9
and 0.92 in August, and 0.9 and 0.9 in September respec-
tively. Because of the existence of such high CC between
these stations, any event responsible for high rainfall over
any of these stations is always favourable for occurrence
of similar high rainfall over the other two stations. How-
ever, the present extreme rainstorm event which was re-
sponsible for the occurrence of the highest rainfall of
94.4 and 104.9 cm over Santacruz and Vihar Lake respec-
tively, has given rise to very less rainfall of 7.4 cm over
Colaba. Hence, the rainfall characteristics associated with
the present event have broken the climatic relationship
that normally occurs during monsoon.
Examination of past records of daily rainfall for more
than 100 years for Mumbai shows that the rainfall of
94.4 cm is the highest ever recorded in Mumbai. Earlier,
the highest was 57.8 cm on 5 July 1974, recorded over
Colaba. The other two very heavy rainfall events in de-
creasing order were also recorded over Colaba. These are
54.8 cm on 10 September 1930 and 47.8 cm on 10 June
1991. However, Santacruz had never recorded such a high
rainfall before. Its records show that the earlier highest
rainfall was 39.9 cm on 10 June 1991 followed by
37.5 cm on 5 July 1975, 34.6 cm on 23 August 1997 and
31.8 cm on 23 September 1981.
Figure 3 a and b shows 3-hourly rainfall and cumulative
rainfall distribution respectively over 24 h during 26–27
July in Santacruz and Colaba. For Santacruz, it shows
0.09 cm from 0830 to 1130 h IST and 1.8 cm from 1130
to 1430 h IST for 26 July. Hence in the first 6 h, the rainfall
was less at Santacruz. Then there was a sudden increase
in the intensity of rainfall in following 3 h in Santacruz,
when the 3-hourly highest realized rainfall of 38.2 cm
was recorded from 1430 to 1730 h IST on 26 July. This
was again followed by another intense spell of 26.8 cm
from 1730 to 2030 h IST, making a total of 65 cm (70%
of the 24 h rainfall) in 6 h in Santacruz between 1430 and
2030 h IST in the afternoon of 26 July. Rainfall intensity
decreased to almost one-third in Santacruz during the
next 6 h, with rainfall accumulation of 21.6 cm. It further
decreased drastically in Santacruz to about 6 cm in the
last 6 h ending at 0830 h IST of 27 July. In contrast, the
3-hourly rainfall data of Colaba for the same time period
plotted in Figure 3 a shows no rain or very less rain. The
highest was only of 2.4 cm observed between 0230 and
0530 h IST on 26 July. Similarly, it is interesting to note
the contrast characteristics of rainfall occurrences over
both stations from their 3-hourly cumulative observations
plotted in Figure 3 b. A continuous sharp raise in cumulative
rainfall value from 1.9 to 88.5 cm was observed over
Santacruz during the 3-hour period 1130–1430 till the
three hour period 2330–0230 h IST, in contrast to very
small increase in cumulative rainfall values from 0.7 to
3.3 cm over Colaba for the same period.
Hourly rainfall from IMD autographic rainfall recorder
could not be retrieved as the instrument was submerged
in rainwater after 1430 h IST. However, an IMD observer
working at Mumbai during the rainstorm period had taken
a lot of pains to preserve such important hourly observations
from 1430 h IST onwards, by continuing manual observa-
tions when the intense rainfall occurred. The observer
had to swim across the flood of water which accumulated
in the observatory compound as a result of ensuing severe
rainstorm, to rescue the rain gauge from the groundwater
and re-install it at the nearest roof top of the office building,
from where he finally succeeded to carry out the observa-
tions. Hence, the special effort and courage showed by
him need much appreciation. These hourly observations
taken manually by him are shown in Figure 3 c. It shows
that the highest hourly rainfall was 19 cm (190 mm) be-
tween 1430 to 1530 h IST. It was preceded by 10 cm of
rainfall between 1330 and 1430 h IST, followed by 9 and
10 cm in the subsequent 2 h.
Rainfall rates from rain gauges tend to have a serious
sampling problem, area wise, in association with this type
of convective rain events. The spacing between rain
gauges, even in relatively dense networks as in the Okla-
homa Mesonet, USA5, is not capable of resolving the details
of precipitation distribution. Though a digital radar could
have provided high-resolution precipitation observations,
in the absence of the same, an effort is made to use avail-
able data resources to analyse the present intense rainfall
event.
To compare the hourly rainfall variation or diurnal
variation of the present event with climatological ones,
hourly rainfall normal was prepared using hourly rainfall
data available daily for 10 years. It shows the occurrence
of highest rainfall in the early morning between 0100 to
1000 h IST with lowest in the afternoon/evening between
1400 and 1900 h IST, having intensity 1.0–1.5 and 0.3–
0.5 cm/h respectively, in contrast to occurrence of highest
rainfall between 1430 and 2030 h IST, as noted in Figure
3 c in the present extreme rainfall event.
The analysis of large-scale rainfall distribution data
over the west coast from 24 onwards till the occurrence
of this severe rainstorm, shows the occurrence and move-
ment of very high rainfall zone of the order of 15–40 cm
from south to north in the Maharashtra region. However,
rainfall on 27 July was higher compared to those on
earlier days over the Ghat section – Dhirpuri, 48 cm; Lona-
vala, 35 cm and Mahabaleshwar, 26 cm. Fairly wide-
spread rainfall with very heavy showers at a few places
was also reported from the rest of Maharashtra on 27
July – with Parbhani, 27 cm, and Aurangabad, 11 cm.
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 90, NO. 10, 25 MAY 2006
1348
Figure 3. a, Rainfall observations at 3-hourly intervals over Santacruz and Colaba on 26 and 27 July 2005. b, Same as (a) but cumulative totals.
c, Same as (a) but at 1-hourly intervals from 1430 to 0230 h IST of next day.
Figure 4. Daily rainfall over Santacruz during June to August 2005.
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 90, NO. 10, 25 MAY 2006 1349
Table 2. Highest 24 h rainfall records in India and the world
Rainfall
(cm) Station (State/UT) Date
India
116.8 Amini Devi (Lakshadweep) 5–6 May 2004
103.6 Cherrapunji (Meghalaya) 13–14 June 1876
99.8 Cherrapunji (Meghalaya) 11–12 July 1910
99.6 Kasauli (Himachal Pradesh) 17–18 June 1899
98.9 Mawsynram (Meghalaya) 9–10 July 1952
98.7 Dharampur (Gujarat) 1–2 July 1941
98.9 Cherrapunji (Meghalaya) 12–13 September 1974
94.4 Santacruz (Maharashtra) 26–27 July 2005
Rainfall
(cm) Station and country Date
World
184.1 Cilaos, Reunion Island 15–16 March 1952
179.6 Foc Foc, Reunion Island 7–8 January 1966
166.2 Belouvc, Reunion Island 27–28 February 1964
155.2 Aurere, Reunion Island 7–8 April 1958
137.8 Muuocaicang, Nei Mouggol China 1–2 August 1977
122.8 Paishih, Taiwan 10–11 September 1963
117.5 Halaho, Taiwan 9–10 September 1963
116.8 Amini Devi, India 5–6 May 2004
115.0 Bagerio, the Philippines 14–15 July 1911
112.3 Belledenker QLD, Australia 3–4 January 1979
Daily rainfall of Santacruz plotted in Figure 4 shows
reduction of rainfall from 11 cm on 21 July to 3.9 cm on
22 July. It further reduced to 0.4 and 1.2 cm on 25 and 26
July. Rainfall observations for other stations from India
show that mostly dry weather conditions prevailed over
North India and northern Maharashtra during 22–25 July,
though parts of southern and eastern India, including
south Madhya Maharashtra and South Konkan and Goa
had reported scattered to fairly widespread rainfall activity
during the same period. Occurrence of such contrasting
spatial rainfall pattern is normally associated with revival
of break/weak monsoon conditions over India6. Analysis
of circulation and rainfall pattern over India before and
during the occurrence of the present rainstorm also con-
firms that the break/weak monsoon prevailed over India
during 19–22 July, which was subsequently revived over
southern and eastern parts of India during 23–25 July. Af-
ter the occurrence of the devastating rainfall event of
94.4 cm during 26–27 July over Mumbai, Figure 4 shows
the occurrence of subdued rainfall thereafter till 31 July,
followed by very heavy rainfall of 20.8 and 14.9 cm res-
pectively, on 1 and 2 August. Then, rainfall was again
subdued over Mumbai till the end of the August (Figure 4).
Comparison with other one-day rainstorms
Though the rainfall of 94.4 cm is an all-time record for
Mumbai, there have been records of heavier rainfall at
other places in the country before. Even during 26–27
July, as stated above, Vihar Lake near Santacruz, Mumbai
and a non-IMD observatory reported rainfall of 104.9 cm.
The other highest 24 h rainfall records in India (collected
from National Data Centre of IMD at Pune) and the
world7 are given in Table 2.
Table 2 shows that the most extraordinary rainstorms
ever recorded on earth have occurred on the small and
mountainous Indian Ocean Island of Reunion. This is located
21° south of the equator, some 500 miles east of the island
of Madagascar. It is located directly on the normal path
of the tropical cyclones in the southwest Indian Ocean.
The island is only about 50 miles in diameter at its widest
and has several tall volcanic mountain peaks, the height
of which reaches an elevation of 3069 m. When the tropi-
cal storm strikes, the mountains provide phenomenal oro-
graphic lift for the storm moisture, enhancing the rainfall
intensity.
It is interesting to note from Table 2 that the most extra-
ordinary rainstorm ever recorded in India has also occurred
in Amini Devi, a small Island in Lakshadweep. The rain-
storm occurred during 5–6 May 2004, due to the formation
of a depression near the island on 5 May, which intensi-
fied to a cyclonic storm on 6 May by remaining nearly
quasistationary there for almost 24 h. The other record-
holding stations, e.g. Cherrapunji, Kasauli and Mawsyn-
ram are in the hilly regions of Himalaya. In fact, Cherra-
punji and Mawsynram are two well-known stations in the
world climatology for the occurrence of the highest an-
nual rainfall. Hence, Table 2 shows that the heaviest 24 h
rainfall has been experienced in case of either a tropical
storm or in the hilly regions or due to a combination of
both. The other heaviest rainfall in India has also been re-
corded at Dharampur in the vicinity of Mumbai. Dharampur
in south Gujarat is about 250 km north of Mumbai. This
indicates that the region around Mumbai is prone to this
type of extreme weather events, which have occurred in
association with active monsoon conditions over the area,
the favourable conditions for which have been discussed
elsewhere in this article.
Possible causes
Mesoscale system embedded in favourable synoptic
scale flow pattern
Before the occurrence of the present severe rainstorm
over Mumbai, break/weak monsoon conditions were observed
over India during 19–22 July. The monsoon revived due
to the formation of a low-pressure area over north Bay of
Bengal off Gangetic West Bengal–Orissa coast on 23
July. The system persisted over the same area for two
subsequent days and became well marked on 25 July. It
moved slowly inland and lay over Orissa on 26 morning.
By then, the monsoon had revived over peninsular India
and parts of east coast of India, while most parts of
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 90, NO. 10, 25 MAY 2006
1350
northwest India, including Gujarat and northern Maharashtra
experienced dry conditions. It may be noted that the low-
pressure area moved fast on the day when Mumbai re-
ceived very heavy rainfall, and lay over central parts of
Madhya Pradesh on 27 morning. It became less marked
on 29 July over southeast Rajasthan. By then, the monsoon
had also revived over most parts of India.
Another low-pressure area formed over northwest Bay
of Bengal off Orissa coast on 28 July, which intensified
into a depression on 29 July and to a deep depression on
30 and lay close to Chandbali, Orissa. Then, it lay near
Jharsugura, Orissa on 31. It weakened to a well-marked
low and moved fast to Sagar in central Madhya Pradesh
on 1 August. It weakened into a low near Guna, northwest
Madhya Pradesh on 3 August and further weakened on 4
August over east Rajasthan.
Observations also show strengthening of the Arabian
Sea branch of the monsoon in association with formations
of both the systems in the Bay of Bengal. Both the systems
were responsible for overall strengthening of rainfall ac-
tivity over the west coast of India. As discussed earlier,
the occurrence of rainfall from 24 to 27 July over the
west coast, including the record highest rainfall of 26–27
July in Mumbai (Figure 4) was partly due to strengthening
of winds over the west coast, which in turn was because of
the formation of the first monsoon disturbance over north
Bay and its westward movement during 24–27 July, as
discussed before. The offshore trough was present in the
sea level chart from Konkan coast to Karnataka coast dur-
ing this period, but it was shallow. There was temporary
reduction in rainfall over the west coast, including Mumbai
for two days after 27 July, when this system moved far
westwards and weakened. With this movement, the rain-
fall zone also moved northwards to Gujarat after 27. There
was a second spell of very heavy rainfall with significant
increase over the state, including Mumbai from 31 July to
2 August, with the value reaching 20.8 cm on 1 August at
Santacruz (Figure 4). This is again because of the res-
ponse of Mumbai to the formation and movement of the
second monsoon disturbance from the Bay of Bengal to
Central India as described before and an offshore trough
off the west coast. It may be noted that both the monsoon
disturbances had moved fast to Central India during oc-
currences of both the heavy rainfall events over Mumbai.
An earlier study8 showed that a pressure gradient of
about 4–8 hPa along the west coast of India between 15
and 20°N is necessary for the occurrence of a heavy rainfall
event in the Mumbai belt. This combined with a trough
off Konkan coast or a low near Saurashtra or a depression
forming in the Bay or moving across west Madhya Pradesh,
is more effective than the pressure gradient alone8. In the
present case, it is also noted from 3-hourly synoptic sur-
face chart that pressure gradient remained nearly 4–6 hPa
from 0530 to 1730 h IST of 26 July, along the west coast
between 15 and 20°N, when this exceptional heavy rain-
fall occurred. Sarma9 studied the variation of heavy rain-
fall >5 cm at Mumbai between Santacruz and Colaba
during July–September. The study also shows that on days
of large variation in rainfall, westerly winds are of the
order of 30–40 knots and the depth 3 km or more, while
on days of little variation of rainfall, the speed is less
than 20 knots and depth 1.5 km. In the present rainfall
event of 26–27 July, when there was a record highest dif-
ference of rainfall received in both the stations, observed
westerly winds were also of the order of 30–55 knots,
with depth up to 5.8 km or more.
Hence synoptic-scale features were highly favourable
for the occurrence of heavy to very heavy rainfall over
Konkan and Goa and other parts of Maharashtra. This insight
seems to have helped forecasters to predict the same 48–
72 h in advance, in their daily weather forecasts. These
forecasts were issued mainly based on synoptic analysis
of large-scale features and certainly failed to capture day-
to-day variation of intense rainfall activity which occurred in
mesoscale, as is the case with Mumbai.
It is apparent that large-scale features alone cannot ex-
plain a phenomenal rainfall event of this kind. We have
tried to analyse both synoptic and mesoscale features further
to figure out scale interaction that might have led to the
occurrence of such a severe rainstorm over Mumbai. Fig-
ure 5 shows large-scale flow pattern over the Indian region
observed at 0000 UTC of 26 July at different standard
levels of the atmosphere. It shows the presence of strong
cyclonic circulation extending up to mid-tropospheric
levels tilting southwards with height near Orissa and east
Madhya Pradesh, associated with the well-marked low-
pressure area on the surface. But over the western part of
India, it does not show the presence of any MTC over
northeast Arabian Sea off Gujarat–Maharashtra coast.
One can clearly note the prevalence of strong westerly/
northwesterly winds with speed of 30–50 knots in the
lower level at 850 hPa over Mumbai (Figure 5 a), which
becomes northerly at 700 hPa (Figure 5 b), northeasterly
at 500 hPa (Figure 5 c) and easterly at 300 hPa (Figure 5 d).
Nearly similar conditions prevailed over Mumbai on 24
and 25 July, two days before.
To support this, we have also plotted wind speed and
wind direction from radio sounding data of Santacruz in
Figure 6 a and b at different levels, for 24–27 July up to
300 hPa. It may be noted that the upper air radio sounding
data of Mumbai is available for all days for all levels up to
150 hPa or above, except at 0000 and 1200 UTC of 26,
for which these data are only available up to 250 and
550 hPa respectively. Figure 6 a and b shows that wind
speed between 925 and 850 hPa reached up to 48–52
knots on 25 and 26 July in a westerly direction. However,
wind speeds near the surface during these days were of
the order of 5–10 knots. Figure 6 b shows angle of wind
direction towards the station. It shows that wind direction
in the lower levels remained between 270 and 310° from
1000 hPa up to 600 hPa during 24–26 July. Hence angle
of wind direction at the lower levels over the station was
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 90, NO. 10, 25 MAY 2006 1351
Figure 5. Circulation pattern at 0000 UTC on 26 July 2005 at (a) 850 (b) 700 (c) 500 and (d) 300 hPa (Wind barb indicates wind direc-
tion and wind speed. It points in the direction ‘from’ which the wind is blowing, with each short barb representing 5 knots and each long
barb 10 knots). Contours are geopotential heights in metres.
mainly westerly, with least variation before and during
the occurrence of the present severe rainstorm. Thus, wind
conditions were highly favourable at the lower tropo-
sphere for continuous moisture incursion from the adjoining
Arabian Sea on both a few days before and during the oc-
currence of the present severe rainstorm. Analysis of the
data for further higher levels in Figure 6 a and b shows de-
crease in the wind speed to below 30 knots and change in
wind direction, with pressure height from 315–355° (i.e.
northwesterly or northerly) at 700 hPa to 5–60° (i.e.
northerly or northeasterly) at 300 hPa, indicating that
winds were veering with height before and during the oc-
currence of the present rainstorm in Mumbai. One may
further note from these figures about the least changes of
both wind speed and direction with height from near the
surface till 500 hPa at 1200 UTC of 27 July over Mumbai.
But during 25–26 July (1200 UTC), i.e. before and during
the occurrence of the present severe rainstorm, there was
abrupt strengthening of wind speed between 925 and
850 hPa followed by abrupt weakening at subsequent lev-
els, with direction changing from westerly to northeasterly
with height. Veering is defined as a clockwise turning of
wind direction as we move up through the atmosphere. It is
always accompanied by warm air advection10, which also
happened over Mumbai region before and during the oc-
currence of the present severe rainstorm.
Figure 5 also shows the interaction of two kinds of air
masses near north of Mumbai. The air mass prevailing at
the lower level up to 700 hPa and having strong westerly/
northwesterly direction, was moist in characteristic with
completely maritime origin, as it originated from the sea
area and was part of the low-level Somali jet which became
westerly over the Arabian Sea then westnorthwesterly
near Mumbai. Vertical relative humidity profile plotted
for Mumbai in Figure 6 c also confirms this, as their values
are higher during 25–26 at the lower levels compared to
24 and 27 July. The flow prevailing at the mid-level was
mostly dry northnorthwesterly winds and colder in char-
acteristic as it had continental origin. As discussed before,
very heavy rainfall over the west coast of India was con-
fined to south of Mumbai from 24 to 26 July, which was
associated with the revival of break/weak monsoon that
prevailed before over India during 19–22 July, while
north of Mumbai, Gujarat and West Rajasthan experienced
mostly dry weather from 23 to 26 July. Thus, it supports
the fact that the winds which came through the latter re-
a
b
d
c
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 90, NO. 10, 25 MAY 2006
1352
Figure 6. (a) Wind speed, (b) direction, (c) relative humidity, (d) vertical wind shear and (e) θe (all parameters are for 24–27 July).
gion from the north at upper levels towards Mumbai until
the morning of 26 were relatively dry northerly.
Daily NCEP temperature and moisture data (www.cdc.
noaa.gov) are also critically analysed for the period 24–
27 July to find whether these two air masses have such
distinct characteristics or not. Daily spatial composites of
air temperature, relative humidity and precipitable water
content at different standard levels from 1000 to 200 hPa
for 24–27 July, showed that the winds in the lower levels
up to 700 hPa coming from the west were more moist and
warmer, while those at mid-tropospheric level from the
north were relatively drier and colder.
It may be noted that this interaction of air masses had
happened during 24–26 July over Mumbai just when
break monsoon which prevailed during 19–22 July, had
started reviving due to the formation of a low pressure
over the Bay of Bengal on 23 July. Hence, except the warm
isotherm west of Mumbai in the Arabian Sea at lower
a
b
d
c
e
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 90, NO. 10, 25 MAY 2006 1353
levels with colder isotherms at its north, isolated warmer
isotherms at lower levels are also observed north of
Mumbai. However, advection was from west at these levels
due to prevailing of strong westerlies up to 50 knots from
the Arabian Sea, thus bringing warm and moist air from
the west. Rao1 has also shown that in break/weak monsoon
conditions, in the west of Mumbai it is warmer and in the
north up to the Himalayas, temperature decreases at mid
levels. It is perhaps the warmer air spreading across
northern India from the west and the cooling associated
with rainfall along the Himalayas that may cause the
temperature to increase towards south from the Himala-
yas in break monsoon situation.
Sawyer11 studied the boundary between the monsoon
air mass and the continental air mass over northwest India
during advance phase of the monsoon, which is similar to
the revival of break monsoon as it is in the present case.
He demonstrated that continental air warmer than the mon-
soon air up to 700 hPa, lies above the latter at these levels.
Above 700 hPa, the monsoon air being warmer will over-
lie the continental air. It has been observed in association
with severe thunderstorms that a lid of warm, dry air at
the top of the moisture boundary layers also appears to be
favourable for the development of severe thunderstorms.
Such a lid inhibits early release of convective instability
which finally helped in establishing intense convective
instability in the atmosphere with time. Interaction of
these two peculiar types of air masses and their veering
are very much favourable for severe thunderstorms12,13.
We have also shown from observations that this rainfall
event was partly associated with the occurrence of severe
thunderstorms.
Study of relative humidity at different levels for 24–27
July in Figure 6 c, shows that the relative humidity
reached its highest values at 1200 UTC of 26 July com-
pared to all other days. It was nearly 90–95% at various
levels for which observations are available. An updraft in
the low levels (0–3 km) along with relatively strong ambient
vertical wind shear might have triggered the development
of additional lifting forces needed for the occurrence of
severe localized rainstorm like the present one. Hence,
we have computed vertical wind shear of zonal component
of wind over Mumbai between different standard levels
from 1000 to 300 hPa and plotted in Figure 6 d, rather
than computing only one value between 850 and 200 hPa.
It shows the presence of strong vertical wind shear over
Mumbai at lower level between 925 and 1000 hPa with
the order of magnitude as 23–40 knots from 1200 UTC of
25 till 1200 UTC of 26 July, i.e. just before and during
the event. In fact, the highest ever positive wind shear for
the data period irrespective of different pressure levels
was observed at 1200 UTC of 25 with the order of 40
knots between 925 and 1000 hPa, which was just one day
before the occurrence of the event. It was followed by an-
other high value of 30 knots at 1200 UTC of 26 observed at
further higher level between 700 and 850 hPa, which was
just during the time of occurrence of the most severe part
of the present rainstorm when the rainfall intensity
reached its peak. However, it was of low value on 24 and
27 July between respective levels.
In Figure 6 d, vertical wind shear between the subsequent
higher levels shows the presence of negative wind shear,
i.e. subsidence, most of the time with large variation in
their values. In Figure 6 d, the highest negative wind
shear values were observed at 1200 UTC of 25 and at
0000 UTC 26 between 850 and 925 hPa as also 500 and
700 hPa respectively. In other words, at 1200 UTC of 25,
warm and moist westerly air which lay at a very lower
level was strongly lifted up to 925 hPa from the lower
level with simultaneously strong subsidence of dry and
cool winds over the region in the aloft. On the next day,
at 0000 UTC of 26, i.e. just before the rainfall event,
though vertical uplifting continued up to 925 hPa, the
strong subsidence of dry and cool air which was there be-
tween 850 and 925 hPa in the previous day’s observation
at 1200 UTC, weakened to nearly 0 (see Figure 6 d) at
0000 UTC of 26. Finally on 1200 UTC of 26, such verti-
cal wind shear was observed to be positive and very high
between 700 and 850 hPa (no observation is available at
500 hPa for 1200 UTC of 26). Hence, vertically uplifting
of moist air mass was highest at 1200 UTC of 26 among
all observations between 24 and 27 July, when rainfall in-
tensity reached its peak.
As discussed before in the present study, the large-scale/
synoptic atmospheric flow over Mumbai city was highly
favourable from 24 July onwards. It was followed by a
local environmental setting, which provided a conducive
environment for the occurrence of intense thunderstorm
on 26 July. We have supported with observations in the
next section, that a mesoscale severe thunderstorm developed
in the highly favourable large-scale/synoptic flow pattern
that existed before, which subsequently resulted in excep-
tional heavy rainfall in the city. Hence the rainfall event
that resulted in exceptional heavy rainfall of 94.4 cm in a
24 h time period can be defined as the combination of an
intense mesoscale event (called as intense thunderstorm)
with time period of 6 h and scale of the system as
35 × 20 km2 that occurred in the afternoon due to interaction
of two air masses embedded on already occurring very
heavy rainfall due to highly favourable synoptic-scale/
large-scale monsoon features prevailing over the region
before the event, having timescale of 3–4 days and scale
of the system in terms of 1000 km. Thus, we conclude
that in the tropics, the interaction of two distinct air masses
giving intense thunderstorm towards the afternoon/
evening, as it happened in Mumbai with abundant mois-
ture available from the Arabian Sea, when 70% of the to-
tal rainfall was realized in 6 h, might not be unusual.
What is unusual is that the occurrence of the 24 h event
that resulted in 94.4 cm of rainfall was a scale interaction
with simultaneous occurrence of an intense thunderstorm
embedded on already prevailing highly favourable synop-
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 90, NO. 10, 25 MAY 2006
1354
tic-scale/large-scale monsoon feature over the Maharashtra
region before and after.
The event was an intense thunderstorm
Severe thunderstorms are normally of high frequency and
most violent over India during summer from April to
June, when the atmosphere is highly unstable because of
high temperatures prevailing at lower levels. However, in
the main monsoon months of July and August, their fre-
quencies over India are less. For the occurrence of thun-
derstorms, three factors are essential, i.e. instability, an
uplifting mechanism and moisture in the lower and mid
levels of the atmosphere. An unstable air mass is warm and
moist near the ground and relatively cold and dry in the
upper atmosphere. If an air mass is unstable, air that is
pushed upward will continue to rise. Lift is the mechanism
that pushes the air upward. Sources of lift can be synop-
tic-scale low-level convergence, orography, or differen-
tial heating. These ‘warm pockets’ are less dense than the
surrounding air and bound to rise once pushed up. In case
sufficient moisture is available, moisture condenses into
small water drops as air rises in a thunderstorm updraft
which forms clouds. When the moisture condenses, heat is
released into the air, making it warmer and less dense than its
surroundings. This helps the updraft to continue.
Precipitation during the monsoon over India could be
in the form of thundershowers, showers or rain1. Each
type occurs according to the prevailing synoptic conditions,
e.g. onset, active/break monsoon, withdrawal, etc. The
most striking features of monsoon rain is the occurrence
of lesser number of thunderstorms along the west coast of
India to the south of 20°N, once the monsoon is establi-
shed. This is because of the prevalence of a highly moist
air mass in great depth in the lower levels with instanta-
neous availability of weak low convergence, which does
not leave any chance for intense convection to develop
that may lead to the occurrence of a thunderstorm. It has
also been found that the growth of cumulus clouds above
6 km is inhibited by the wind shear due to the easterlies
aloft14. However, during the onset/withdrawal phase or
prevalence/revival of break monsoon conditions15, fairly
widespread thunderstorm activities have been observed
over many stations, especially from those located in
northwestern and peninsular India, including Mumbai.
This is mainly because of the factors discussed by Sawyer11.
As the rainfall over Mumbai after arrival of the mon-
soon is generally not associated with severe thunderstorm
activity, the monsoon cloud tops are of relatively lower
height along the west coast of India3. Climatological
study of thunderstorm days over Mumbai16,17 shows occur-
rence of highest in June with frequency of 5.5 days per
month, during which the monsoon normally advanced
over the region, followed by October with 4.0 when mon-
soon started withdrawing from the region, few thunder-
storm activities are observed in July and August. Their
mean frequency is only three per month for July as
against 24 rainy days in the month. Mostly, these thun-
derstorms might have occurred during weak monsoon/
break monsoon conditions when there is incursion of dry
air at mid-troposphere1. Climatological study of squall
reports over Mumbai too shows highest occurrence in
June and October, one per year for both months followed
by few in July (0.3 per year) and August having no
squall. However, autographic weather observations in
Santacruz on 26–27 July 2005 showed the occurrence of
94.4 cm rainfall in 24 h, which reached the highest inten-
sity in the afternoon/evening (Figure 3), was associated
with severe thunderstorm activity. Autographic weather
observations and synoptic observations at Santacruz re-
ported thunderstorms and lightning with rain at 1400,
1700, 2202 h IST of 26 and at 0530 h IST of 27 July.
Squall with wind speed reaching 42 knots was experi-
enced at 1630 h IST of 26. Because of intense thunder-
storm activity with the presence of deep clouds in the
afternoon, the minimum temperature over Santacruz be-
tween 0830 h IST of 26 and 0830 h IST of 27 was re-
corded as 24.1°C at 1700 h IST, i.e. in the afternoon of 26
July. The period of lowest temperature occurred between
1430 and 1730 h IST, whereas this is the epoch of the
maximum temperature over the station under normal cir-
cumstances.
We have compared the timing of occurrence of this
thunderstorm and squalls with the climatological timing
of occurrences. Climatological study of daily occurrence
of thunderstorm and squall timings in July16,17 showed
commencement timing of thunderstorms distributed through-
out the day, least in the early morning, while in case of
squalls, the most favourable timing is from 1430 to
2330 h IST, when maximum squalls occurred. Hence
there is nothing unusual with the occurrence timing of both
squalls and thunderstorms, as had happened in the present
case.
McNulty18 argued that it is important for forecasters to
be able to distinguish between severe and non-severe
thunderstorms. It has been found that if the environment
(uplifting, instability or moisture) of a thunderstorm changes,
then the type and characteristics of the thunderstorm that
formed before may also change. The amount of vertical
wind shear in a storm’s environment at lower levels, is
critical in determining the type of storm that will form. If
the vertical wind shear is weak, multicellular storms with
short-lived updrafts will be favoured. Studies by McNulty18,
and Johns and Doswell19 have indicated that the three de-
termining factors in such a decision are, extreme instability,
strong vertical wind shear, and mid-level dry air or an in-
trusion of dry air at mid-level. If the vertical wind shear
at very lower level is stronger as was the case for Mumbai,
then storms with longer-lived updrafts will develop.
Closely related to the concept of vertical wind shear is
veering, which was also favourable over Mumbai on 26
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 90, NO. 10, 25 MAY 2006 1355
July. Veering of low-level wind is instrumental in the pro-
duction of thunderstorm rotation. Once this vertical rotation
is established, a mesocyclone can develop which may
produce a supercell or significant severe weather. The
amount of moisture in the air has an effect on thunderstorm
too. If the amount of moisture is low, then the storms
tend to have high bases. If the amount of moisture is
high, the storms tend to have low bases. Lower the cloud
base, better is the chance for flash flood-producing rains.
It may be noted that the existed environment over Mumbai on
26 July was characterized by high instability, strong ver-
tical wind shears at the lower levels and veering of the
wind with height with sufficient moisture supply from the
Arabian Sea by the prevailing large-scale monsoon circu-
lation. As a result, development of the severe thunderstorm in
the form of supercell or mesocyclone towards after-
noon/evening over Mumbai region, when the most in-
tense spell of rainfall occurred, cannot be fully ruled out.
To find out whether atmospheric thermodynamic condi-
tions prevailing over Santacruz have supported the occur-
rence of such localized severe thunderstorm embedded in
large-scale favourable monsoon circulation due to interaction
of two distinct air masses as discussed earlier, we have
analysed daily atmospheric vertical sounding data from
22 July to 4 August when such an event occurred on 26
July and rainfall again increased on 31 July and 1 August.
An attempt is also made to examine different instability
indices.
Variation of convection in the atmosphere depends
upon dynamics as well as thermodynamics instability indices.
θe or equivalent potential temperature is the prime para-
meter for measuring moist convective instability. Steeper
the fall in the values of equivalent potential temperature,
more is the presence of convective instability. A critical
parameter for atmospheric convection is the convective
available potential energy (CAPE), a measure of the vertical
instability of the atmosphere under moist convection.
CAPE is the work done by the buoyancy force on a parcel
lifted through the atmosphere moist adiabetically and is
given by20
CAPE = –LNB
vpveD
LFC
()
TTR
−
∫d(ln p),
where RD is the gas constant of dry air; Tvp and Tve are re-
spectively, the virtual temperatures of the parcel and the
environment at pressure p, LFC and LNB are levels of
free convection and neutral bouancy. Deep clouds can
develop due to the ascent of air from a given level, only if
its CAPE is greater than zero.
When disturbances occur, precipitation, strong winds
and downdrafts decrease the energy of the air from a given
level, only if its CAPE is greater than zero.
Normally, air is not saturated to start with and a finite
vertical displacement (a few hundred metres to a few
kilometres) is needed for the rising air parcel to become
saturated and reach the level of free convection. Some
energy is required for this process, and is called convection
inhibition energy (CINE)20
CINE = –LFC
vpveD
SURFACE
()
TTR
−
∫d(ln p).
It is expected that a larger value of CINE means an increased
barrier to convection. If CINE is large, deep clouds will
not develop even if CAPE is positive, while low values of
CINE imply a favourable condition for convection.
Figure 6 e shows values of θe with height over Mumbai
for the period 24–27 July, which covers a few days before
the present rainstorm till 27 July. It shows that fall of θe
with height was high at 1200 UTC of 25 and at 0000
UTC of 26 July between 1000 and 925 hPa as well as 700
and 500 hPa respectively. Hence intense convective in-
stability was present in the lower and mid-level a day be-
fore the start and at the time of occurrence of the present
event which led to severe thunderstorm in the after-
noon/evening of 26 July. We have also computed the net
difference of θe between 1000 and 500 hPa during these
dates and the values are 39.2 K for 1200 UTC of 24,
38.3 K for 1200 UTC of 25, 39.1 K for 0000 UTC of 26,
9.5 K for 1200 UTC of 26, and 18.0 K for 1200 UTC of
27 July. This comparison also confirms that instability
was very high before the occurrence of the event, which
reduced significantly just following the event.
Figure 7 a and b shows CAPE, CINE and precipitable
water content (PWC) respectively from 22 to 28 July over
Santacruz. Figure 7 a shows continuous and steep increase
in CAPE values from 702 J kg–1 on 22 July to its highest
values of 4341 J kg–1 on 25 July, just one day before the
occurrence of the rainfall event. The value decreased to
3267 J kg–1 on the day of the occurrence of the rainfall
and further reduced to 252 J kg–1 on 27 July. It increased
again after the event. Hence, CAPE increased by 3000–
4000 J kg–1 before convection and decreased by nearly
similar magnitude following convection. It again recovered
after 1–2 days causing another spell of heavy rainfall.
Figure 7 b also shows the presence of high CINE values
of the order of –22.6 and –06.6 J kg–1 on 23 and 24 July
respectively, i.e. before the rainfall event and of –5.51
and –29.86 J kg–1 on 27 and 28 July respectively, i.e. after
the event with the lowest values of 0 on 25 and 26 July
when the event occurred. Bhat et al.21 also observed that
CAPE decreased by 3000–4000 J kg–1 following convec-
tion and recovered in a time period of 1–2 days, while
CINE reached its lowest value on the day of rainfall.
When the disturbances attenuate, the air–sea fluxes in-
crease the energy of the surface air, while the temperature
of the air aloft decreases because of radiative cooling.
These factors destabilize the atmosphere and build up
CAPE. For the Pacific warm pool, when disturbances occur,
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 90, NO. 10, 25 MAY 2006
1356
Figure 7. (a) CAPE values and (b) CINE and perceptible water content (PWC) during 23–28 July (all values are for 0000 UTC).
precipitation, strong winds and downdrafts decrease the
energy of the air near the surface, while deep cloud activity
makes the upper troposphere warmer. As a result, the atmos-
phere becomes less unstable and CAPE is substantially
reduced during disturbances21,22. The period between suc-
cessive disturbances is expected to depend upon the time
it takes for CAPE to build up. Bhat et al.21 have analysed
the BOBMEX data and have shown for the Indian region
that the magnitudes of CAPE and convective inhibition
energy were comparable to those for the atmosphere over
the west Pacific warm pool. PWC values in Figure 7 c
also show gradual increase in value from 42.1 mm on 22
July to 59 mm on 25 July one day before the event, followed
by a decrease in value to 53.8 mm on the day of the event.
It again shows a gradual increase in value to 67 mm on 28
July. Such a small decrease in PWC values on the day of
the event followed by increase in PWC values the next
day, may be attributed to prevailing strong westerly wind
from the Arabian Sea in the large-scale flow, which per-
sisted at the lower levels a few days before as well as a
few days after the event.
Various satellite pictures of 26 July available for the
Indian region received either through Indian satellite
(Kalpana-1) or other international satellites are analysed
at different resolutions to identify whether any area over
Mumbai had distinct intense convective clouds which
may correspond to the most severe mesoscale rainstorm
that occurred on 26–27 July. Unfortunately, no such distinct
cloud mass could be identified to have direct correspon-
dence with the present severe rainstorm over Mumbai.
However, we analysed these pictures to identify charac-
teristics of development of clouds on a larger scale that
might have worked as a feeder band for the occurrence of
the present rainstorm.
From satellite pictures given in Figure 8, one can hardly
find the development of any intense convective cloud
over the area north of 20°N of India in the 0300 UTC pic-
tures, in contrast to maximum area covered by clouds at
its southern parts. Development of similar cloud pattern
is normally observed over India during the revival of
break monsoon, which has been studied earlier by Jena-
mani6. Hence the present analysis of satellite pictures also
confirms that the severe rainstorm occurred when mon-
soon was in its revival phase. Together with this, one may
also note the development of new intense convective clouds
in the area demarcated by white circle in the 0900 and
1200 UTC pictures (Figure 8), mostly confined to the
northeast of Mumbai. At 0300 UTC, the same area was
mostly free from clouds, while at 0600 UTC, only one
cloud cell had been clearly developed. Later, the same
cloud cell further developed in a north-south orientation
as marked in a line in the satellite pictures at 0900 UTC
(Figure 8 c) and 1200 UTC (Figure 8 d). These newly deve-
loped clouds in the afternoon are associated with intense
convection which caused severe thunderstorm/rainstorm
northeast of Mumbai. Hence, these pictures confirm that
afternoon convection was more intense and extended in
the northeast of Mumbai than southeast, where rainfall
was high. One may also note the presence of intense con-
vective clouds from these pictures over Central India as-
sociated with the low-pressure area.
Outgoing Longwave Radiation (OLR) as observed through
satellites, is an index of convective activity/cloudiness.
Cloudy skies generally disrupt the OLR. Hence, higher
the fractional cloudiness, lower are the OLR values. In
other words OLR would have an inverse relation with
rainfall but the relationship is not one-to-one23. Spatial
analysis of OLR data from Indian satellite (Kalpana-1)
on 26 July for different synoptic hours does not show the
presence of any area near Mumbai with lowest OLR con-
tour compared to that of Central India, where large-scale
intense convective clouds were present due to the fast
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 90, NO. 10, 25 MAY 2006 1357
Figure 8. Kalpana-1 pictures at 3-hourly intervals from 0300 UTC of 26 July.
movement of the low towards central Madhya Pradesh
from Orissa on this day. Analysis of grid point OLR data
around Mumbai showed that the values at the grid nearest
to Mumbai decreased to the lowest of 115 W/m2 at 0900
UTC from 157 W/m2 observed at 0300 UTC on the same
day. It may be noted that the rainfall intensity had
reached its highest ever value between 0900 and 1200
UTC, as shown in Figure 3 a.
We have analysed all pictures of S-band radar taken by
IMD, Mumbai. The two main types of scans carried out
by S-band radar for detecting characteristics of different
clouds around the given location are the Plan Position In-
dicator (PPI) and the Range Height Indicator (RHI) scans.
When scanning in PPI mode, the radar holds its elevation
angle constant, but varies its azimuth angle. The returns
can then be mapped on a horizontal plane. When scanning
in RHI mode, the radar holds its azimuth angle constant
but varies its elevation angle. The returns can then be mapped
on a vertical plane. In other words, PPI and RHI images
represent respectively, a horizontal and vertical cross-
section of reflecting clouds.
Figure 9 a and b shows radar pictures for 1130 h IST
and 1430 h IST respectively, at PPI mode. Figure 9 c shows
the radar picture for 1430 h IST at RHI mode. Figure 9 a
shows that at 1130 h IST, formation of clouds is mainly
confined to the northeast sector with northwest to southeast
elongation and anvil shape, while in Figure 9 b, at the
1430 h IST, it took an oval shape with diameter of nearly
80 km due to the development of new intense clouding at
its southsoutheastern sector. Further, clouds were mostly
developed at the eastern half of the radar circle in both
the pictures, where very high rainfall had occurred (Figure
1).
Figure 9 c shows a cloud top of about 17 km height,
about 10 km away from the centre. A scan of successive
radar observations shows that the cloud of highest vertical
growth was constantly observed at a distance of 15–20 km
north of Colaba. This indicates that some clouds were of
a
b
c
d
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 90, NO. 10, 25 MAY 2006
1358
Figure 9. PPI and RHI cloud photographs from the storm detecting radar at Mumbai on 26 July 2005. a, PPI at 1130 h IST having radar range at
100 km with each range ring of 20 km scale. b, Same as in (a), but at 1430 h IST. c, RHI at 1430 h IST having radar range at 50 km with scale
height as 5 km each in vertical, while horizontal scale is 10 km each.
intense convective type, which resulted in severe rain-
storm over Mumbai. It may be noted that the present radar
is not capable of measuring any characteristics of the pre-
sent rainfall event, including rainfall rate or its spatial
characteristics. Hence, pictures available from other
agencies, e.g. Tropical Rainfall Measuring Mission (TRMM)
are also referred.
TRMM is a joint mission between NASA and the Japan
Aerospace Exploration Agency (JAXA), to monitor tropical
and subtropical precipitation and to estimate its associated
latent heating. TRMM (http://trmm.gsfc.nasa.gov/home-
page.html) provides systematic visible, infrared and mi-
crowave measurements of rainfall in the tropics as key
inputs to weather and climate research. Satellite observa-
tions are complemented by ground radar and rain-gauge
measurements to validate the satellite rain estimation
techniques. TRMM satellite has been collecting rainfall
data since its launch back in November 1997.
TRMM captured an image of the rain over Bombay at
15:39 h IST on the 26 July 2005 (Figure 10 a). The image
displays the horizontal distribution of rain intensity as ob-
tained from the TRMM sensors. Rain rates in the central
part of the swath are from the TRMM Precipitation Radar
(PR), the only radar capable of measuring precipitation
from space. Rain rates in the outer swath are from the
TRMM Microwave Imager (TMI). The rain rates are over-
a
b
c
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 90, NO. 10, 25 MAY 2006 1359
Figure 10. a, TRMM captured image of the rain rate over Mumbai at 15.39 h IST on 26 July 2005. b, Rainfall accumulation from the TRMM-
based near-real time multi-satellite precipitation analysis at the NASA Goddard Space Flight Center for the period 25–27 July 2005.
laid on infrared (IR) data from the TRMM Visible Infra-
red Scanner (VIRS). The image shows a localized area of
heavy precipitation (red area) directly over Mumbai, of
the order of 4–5 cm per hour.
The second image (Figure 10 b) shows rainfall accumula-
tion from the TRMM-based, near-real time Multi-satellite
Precipitation Analysis (MPA) at the NASA Goddard
Space Flight Center, USA for the period 25–27 July 2005.
The dark red area of heavy rainfall (indicating a foot of
rain) is concentrated right over Mumbai and shows local-
ized extreme rainfall values of the order of 22–30 cm.
Discussion
It is well known that orography at the Western Ghats plays
an important role in enhancing rainfall over the west
coast, including that of Mumbai1. Except this, climato-
logical data over different parts of Mumbai region show a
large difference of seasonal mean monsoon rainfall, e.g.
Colaba, Santacruz and Matheran. Colaba and Santacruz
differ in seasonal mean monsoon rainfall amounts by
12%, though both stations are located within a distance of
nearly 27 km. A hillock of 300 m within 5 km of Santa-
cruz apparently accounts for this increase1. Matheran,
which is close to the coastal stations of Mumbai and having
height of 756 m, gets high rainfall because of its location
on the plateau of a minor range, almost at the beginning
of its lee side.
While analysing the micro characteristics of the present
rainfall event in Figure 1, observations also confirm the
occurrence of two high rainfall areas, one near Santacruz
and the other near Matheran. Hence the effect of orography
cannot be completely ruled out. We have not shown any
result as evidence to prove how much orography has con-
tributed. However, to elucidate the effect of orography,
we would like to show variation of climatological rainfall
at coastal stations and rainfall at hill stations located over
the crest/top of the Western Ghats available from an earlier
study1. Rainfall enhancement on account of orography
may depend on the slope of the ground, height, configura-
tion of land around the synoptic system and strength of
westerly winds to stations. Height alone is not a decisive
factor on the role of orography. Agumbe (with height
from sea level 659 m) in the Western Ghats gets 718 cm,
while Cherrapunji (1313 m) gets 837 cm. In the Western
Ghats, Mahabaleshwar at nearly twice the height as that
of Agumbe, records 16% less rainfall compared to the lat-
ter. Hence the effect of orography, if any, in causing the
present rainstorm over Mumbai needs to be studied criti-
cally. An important question arises: ‘Is the deluge in
Mumbai an impact of climate change?’
Scientifically speaking, global warming is expected to
accelerate the hydrological cycle and lead to precipitation
extremes in some areas at the cost of other places, where
droughts may develop. The Intergovernmental Panel on
Climate Change24 outlined in its Third Assessment Report,
the likelihood of occurrence of such extremes with higher
frequencies in the 21st century. Simple extremes, such as
higher maximum land temperatures and more intense
precipitation, are projected to be likely with increased
chances of their occurring repeatedly25,26. These amplified
simple extremes could lead to extreme weather events
like drought and flooding. Current trends in one-day and
multi-day precipitation events are revealing. It seems
there is now a tendency to have more days with heavier
a
b
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 90, NO. 10, 25 MAY 2006
1360
24-h precipitation, even if some of these areas are experi-
encing a decrease in total annual rainfall. This is yet to be
confirmed from larger datasets.
The case of India is interesting while looking at extreme
weather events. For a country that has more than 60% of
its population relying on agriculture directly or indirectly,
the impact of extreme weather events is critical. In the
last decade, India has been repeatedly battered by monsoon
droughts in 2002 and 2004. Also, one must take into ac-
count whether an extreme event is truly extreme. The
natural variability of extreme weather events is still to
some extent unknown, since most historical records on
events can only stretch back to less than a hundred years.
The cycles that extreme weather events undergo may be
on a timescale that is longer than our records, thus mak-
ing them difficult to foresee. For example, statistically
speaking, the probability of an extreme annual precipita-
tion event of 1 in 1000 years still falls within the range of a
‘normal’ climate. An event that may seem extreme to us
may be a purely random variation or part of the natural
variability of the earth’s climate in the last four hundred
years or more. How can the public be better informed
about extreme weather events? Historical records do pro-
vide some reliable insight into the frequency and inten-
sity of extreme weather events. Yet, as mentioned before,
these records are insufficient to give an accurate picture of
what to expect. Predictive climate models have been util-
ized to analyse extreme weather events, but unpredictabil-
ity of extreme weather events is impeding the process.
However two earlier studies27,28 have improved our con-
fidence in estimating the risk of flooding and extreme
precipitation. The first study27 found that the frequency of
severe floods in large river basins has increased during
the 20th century. It was established that the likelihood of this
increase arising from natural climate variability is small.
The other study28 analysed the output of 19 climate models,
predicting that wet winters will be five times more likely
in northern and central Europe over the next century. It
was also estimated that the Asian monsoon region will
experience an increase in wet summers of a similar scale,
escalating the risk of flooding in already flood-prone areas.
So, the debate continues29. No clear consensus exists link-
ing the frequency and intensity of extreme weather events
to changes in climate patterns, but it would be reckless
not to expect that climate change will have some impact
on extreme weather events. Today, there are over five bil-
lion people living on the earth, often in areas that are
known to be vulnerable to extreme weather events. The
potential for catastrophic damage and loss of lives is
enormous. There is an urgent need for better understand-
ing of the changing climate patterns in a regional scale over
the Indian subcontinent and how they affect extreme
weather events over India. Further study, especially on
the natural variability and cycles of extreme weather
events as well as data on a smaller (i.e. local) scale is re-
quired before making any remark on the role of climate
change in the occurrence of the present extreme rainfall
event as observed over Mumbai.
With our present limited understanding about the role
of different factors that caused such unusual rainstorm
over Mumbai, another important challenge before us is how
better India and other global forecasting centres are
equipped both technically and scientifically to forecast
important events, either in real time or for academic interest.
To briefly explain this, such forecasting accuracy of dif-
ferent operational models available from different Nu-
merical Weather Prediction (NWP) centres in India and
abroad are analysed to find their levels of skills for pre-
dicting severe rainstorms. We have not considered forecast
products beyond 24 h in advance, because in the present
case rainfall is due to mesoscale weather system embed-
ded on highly favourable synoptic/large-scale pattern.
NWP models generally lose their skill in predicting such
events with increase in forecasting time period.
Mesoscale/regional models normally perform better in
simulating the localized heavy rainfall event compared to
global models. Though both adequate and accurate sur-
face and upper air data are a prerequisite, they need to be
appropriately assimilated on very high resolution to cap-
ture the phenomenal rainstorm discussed above. IMD is
going out in a big way for modernization and upgradation
of its observing and forecasting system at the earliest.
However, it is still working under constraints, where it is
not possible to run high resolution models due to lack of
high power computing resources and also because the
kind of data required to be fed into these models are at pre-
sent lacking. At present, IMD is running regional and
mesoscale models (www.imd.gov.in) with initial and
boundary conditions either from National Center for Me-
dium Range Weather Forecasting (NCMRWF) or NCEP,
USA at coarse resolution, which is not sufficient enough
to resolve features responsible for predicting such devas-
tating mesoscale rainfall events. We reproduce results of
integration of our regional and mesoscale models. Figure
11 a and b shows rainfall forecast of Limited Area Model
(LAM) and mesoscale model (MM5) used operationally
by IMD valid for 24 h of 0000 UTC of 27 July. Though
both the models brought out in general the spatial distri-
bution of rainfall over the west coast as was observed, the
highest rainfall of 4–7 cm predicted by both the model
was small.
Figure 11 c shows rainfall forecast from global model
(T80) used operationally by NCMRWF (www.ncmrwf.
gov.in) valid for 24 h of 0000 UTC of 27 July. We have
also considered 24 h rainfall forecast of Eta model from
the same centre. Though both the models were able to pre-
dict the observed spatial pattern of rainfall over the west
coast, the highest rainfall over the west coast predicted
was less: 1–2 cm in the global model and 8–16 cm in the
Eta model, values below the realized rainfall of 94.4 cm.
Among the global model products available from other
centres of the world, IMD at present receives its products
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 90, NO. 10, 25 MAY 2006 1361
Figure 11. The 24-h forecast of rainfall as simulated by different NWP models valid for 27 July 2005: (a) LAM, (b) MM5, and (c) T-80. d, An
UK Met. Office Model could predict up to 30 cm of rain on 26 July, 24 h in advance. e, Re-run (not on real time) of UK Met. Office high resolution
model predicted up to 80 cm of rain in Mumbai on 26 July, 24 h in advance.
c
d
e
a
b
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 90, NO. 10, 25 MAY 2006
1362
regularly through global telecommunication system
(GTS)/Internet. The COLA (Center for Ocean–Land–Atmo-
sphere) model of USA also gives station-specific forecast
(www.monsoondata.org/wx/meteogram2.html) for Asian
monsoon region. Its 24 h meteogram for Mumbai shows
only 4 cm of rainfall for 24 h. It is not possible to validate
European Center for Medium Range Weather Forecasting
(ECMWF) model forecast, as we did not get any rainfall
forecasts from ECMWF operationally. It only provides
forecasts of large-scale flow pattern for 850, 500 and
200 hPa. These forecast charts more or less matched with
the observations.
In Figure 11 d and e, we have given 24 h rainfall fore-
cast of UK Meteorological Office (UKMO) model, which
was on real-time basis and the re-run of this model from
UKMO after the event respectively. The UKMO Model
could predict up to 30 cm of rain on 26 July, 24 h in ad-
vance over the Konkan coast, with highest reaching up to
51 cm (Figure 11 d). A re-run of this model by incorporating
more data, predicted up to 80 cm of rain in a small area
covering Mumbai on 26 July (Figure 11 e), 24 h in ad-
vance, which is somewhat close to the observed amount
of 94.4 cm over Santacruz.
Latest observing systems based on modern technology
as available, e.g. Doppler Radar, TRMM, etc. could be help-
ful in monitoring successfully different development as-
pects of the present rainstorm. In fact, TRMM (Figure 10)
was the only observing system which could measure the
prevailing intensity of rainfall rate and spatial distribution
of the present rainstorm that occurred over Mumbai. This
shows the capability of satellite-based observations to
monitor such intense events and a more frequent coverage
of the area could enable now casting of same. The proposed
Doppler weather radars would supplement such space-
based observations and become an effective aid in moni-
toring and now casting of such events. For highly vulnerable
and economically important centres like Mumbai, a net-
work of Doppler weather radars would certainly go a long
way in meeting this requirement. Also, occurrence of
such high rainfall over the west coast of India has again
re-established the fact that both the coasts of India are
highly vulnerable to disastrous weather and hence some
observatories along both the coasts urgently require to be
equipped with modern instruments like wind profilers,
well-calibrated automatic surface weather stations, etc.
1. Rao, Y. P., Southwest monsoon. Synoptic Meteorology, Meteorol.
Monograph, India Meteorological Department, 1976, No. 1.
2. Sarker, R. P., Some modification in a dynamic model or orogra-
phic rainfall. Mon. Weather Rev., 1967, 95, 673–684.
3. George, P. A., Effect of off-shore vortices on rainfall along the
west coast of India. Indian J. Meteorol. Geophys., 1956, 7, 225–240.
4. Miller, F. R. and Keshavamurthy, R. N., Structure of an Arabian
Sea summer monsoon systems-IIOE. 1968, Meteorol. Monograph
No. 1, University of Hawaii, Honolulu.
5. Brock, F. V., Crawford, K. C., Elliot, R. L., Cuperus, G. W.,
Stadler, S. J., Johnson, H. L. and Eilts, M. D., The Oklahoma Mesonet:
A technical overview. J. Atmos. Ocean. Technol., 1995, 12, 5–19.
6. Jenamani, R. K., Distinct synoptic patterns associated with pre-
break onset phase and revival of normal monsoon phase. Mausam,
2004, 55, 591–598.
7. Burt, C. C., Extreme Weather – A Guide and Record Book, W. W.
Norton and Co, New York, 2004.
8. Jayaram, M., A preliminary study of an objective method of fore-
casting heavy rainfall over Mumbai and neighborhood during the
month of July. Indian J. Meteorol. Geophys., 1965, 16, 557–564.
9. Sarma, V. V., A study of heavy rainfall occurrence over Santacruz
(Mumbai) with special reference to supersonic flight operations.
Indian J. Meterol. Geophys., 1969, 20, 263–266.
10. Holton, J. R., An Introduction to Dynamic Meteorology, Academic
Press, 1979.
11. Sawyer, J. S., The structure of the inter-tropical front over the
north-west India during southwest monsoon. Q. J. R. Meteorol.
Soc., 1947, 73, 346–369.
12. Rudari, R., Entekhabi, D. and Roth, G., Terrain and multiple-scale
interactions as factors in generating extreme precipitation events.
J. Hydrometeorol., 1999, 5, 390–404.
13. Lin, Y. L., Chiao, S., Wang, T. A., Kaplan, M. L. and Weglarz, R.
P., Some common ingredients for heavy orographic rainfall.
Weather Forecast, 2001, 16, 633–660.
14. Bhaskara Rao, N. S. and Dekate, M. V., The effect of vertical
wind structure on some aspects of convective activity at Mumbai.
Indian J. Meteorol. Geophys., 1971, 22, 59–66.
15. Rao, Y. P. and Hariharan, P. S., Unusual thunderstorm activity in
Pensinsular India possibly due to upper divergence. Indian J. Me-
teorol. Geophys., 1957, 8, 452–455.
16. Vishvanathan, T. R. and Faria, J. F., A climatological study of
thunderstorms at Mumbai airport. Indian J. Meterol. Geophys.,
1962, 13, 377–383.
17. Biswas, B. K., An analytical study of thunderstorm over Mumbai.
Mausam, 2005, 56, 690–692.
18. McNulty, R. P., Severe and convective weather: A central region
forecasting challenge. Weather Forecast, 1995, 10, 187–202.
19. Johns, R. H. and Doswell III, C. A., Severe local storms forecast-
ing. Weather Forecast, 1992, 7, 588–612.
20. Williams, E. R. and Renno, N. O., An analysis of the conditional
instability of the tropical atmosphere. Mon. Weather Rev., 1993,
121, 21–36.
21. Bhat, et al., BOBMEX: The Bay of Bengal monsoon experiment.
Bull. Am. Meterol. Soc., 2001, 82, 2217–2243.
22. Emanuel, K. A., Atmospheric Convection, Oxford University
Press, 1994, p. 580.
23. Morrissey, M. L., A statistical analysis of the relationship among
rainfall, outgoing long wave radiation and the moisture budget during
January–March 1979. Mon. Weather Rev., 1986, 114, 931–942.
24. Intergovernmental Panel on Climate Change, Climate Change
2001: The Science of Climate Change, Cambridge University
Press, Cambridge, UK, 2001.
25. Tompkins, H., Climate change and extreme weather events: Is
there a connection? Center for International Climate and Environ-
mental Research, Oslo, 2002, Cicerone(3).
26. Easterling, D. R. et al., Observed variability and trends in extreme
climate events: A brief review. Bull. Am. Meteorol. Soc., 2000, 81,
417–425.
27. Milly, P. C. D., Wetherald, R. T., Dunne, K. A. and Delworth, T.
L., Increasing risk of great floods in a changing climate. Nature,
2002, 415, 514–517.
28. Palmer, T. N. and Raisanen, J., Quantifying the risk of extreme
seasonal precipitation events in a changing climate. Nature, 2002,
415, 512–514.
29. Srinivasan, J., Hottest decade: Early warning or false alarm? Curr.
Sci., 2006, 90, 273–274.
ACKNOWLEDGEMENTS. We thank Mr B. Lal, Director General of
Meteorology, India Meteorological Department, New Delhi for encour-
agement and permission to publish this paper. We also thank the anony-
mous referee for suggestions and comments to improve the manuscript.
Received 3 November 2005; revised accepted 24 April 2006
