Atmospheric Research 95 (2010) 400–406

Contents lists available at ScienceDirect

Atmospheric Research
j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / a t m o s

Severe local convective storms in Bangladesh: Part I. Climatology

Yusuke Yamane a,⁎, Taiichi Hayashi b, Ashraf Mahmmood Dewan c, Fatima Akter d
a
Pioneering Research Unit for Next Generation, Kyoto University, Gokasho, Uji, Kyoto, Japan
b
Disaster Prevention Research Institute, Kyoto University, Gokasho, Uji, Kyoto, Japan
c
Department of Geography and Environment, University of Dhaka, Dhaka - 1000, Bangladesh
d
Habitat for Humanity International - Bangladesh, Dhaka - 1212, Bangladesh

a r t i c l e i n f o a b s t r a c t

Article history: This paper presents the climatology of severe local convective storms in Bangladesh. We
Received 13 May 2008 collected sufficient severe local convective storm events over a long period (2,324 events from
Received in revised form 30 October 2009 1990 to 2005) to compile reliable climatological statistics by surveying the literature. Severe
Accepted 9 November 2009
local convective storms frequently occur during the pre-monsoon season from March to May,
peaking in April. The highest frequency was found to be at 16–17 Local Standard Time (LST)
Keywords:
and 20–21 LST, with lower activity around midnight and early morning. It was also found that
Severe local convective storm
the occurrence of severe local convective storms is more concentrated in the central region of
Climatology
Bangladesh Bangladesh, and tends to spread northward during the pre-monsoon season.
© 2009 Elsevier B.V. All rights reserved.

1. Introduction ical department reports and newspapers. They reported that

“tornado” and nor'wester occurred most frequently in the
Bangladesh is located in the northeastern part of the Indian Dhaka district during the pre-monsoon season. In addition, it
subcontinent and faces the Bay of Bengal to the south and the was reported that the most frequent time of occurrence was in
Meghalaya plateau to the northeast. Almost the entire country is the evening (in their paper, however, the exact period of
less than 10 m above sea level and on a flat plane. Severe Local “evening” is not clearly specified).
Convective Storms (henceforth referred to simply as SLCS) Chowdry and Karmakar (1986) investigated the climatol-
frequently occur in Bangladesh during the pre-monsoon season ogy of nor'westers from 1964 to 1973 in Bangladesh based on
from March to May, causing deaths and damage to property the local literature. They reported that nor'westers occurred
every year. most frequently in the north central region of Bangladesh
In Bangladesh, SLCS are classified depending on the during the pre-monsoon season, peaking in April. Ono (1998)
magnitude of wind speed. The ones producing wind gusts collected 44 “tornado” events from 1990 to 1994 in
above 42 ms− 1 are defined as tornadoes, while those producing Bangladesh using newspapers published in Bangladesh and
wind gusts ranging from 11 ms− 1 to 42 ms− 1 are defined as found that “tornado” occurred most frequently in central
‘nor'westers’. The term ‘nor'wester’ means that SLCS come Bangladesh during the pre-monsoon season, peaking in May,
mostly from the northwestern direction. Despite being highly and with most events occurring in the evening (again, the
arbitrary, such criteria for classifying SLCS have been used in a exact period of evening is not clearly specified).
number of climatological studies addressing SLCS in Bangla- The above climatological studies were conducted based on a
desh. Afrose et al. (1981) collected data on “tornado” and survey of the local literature. In such type of approach, it is
nor'wester events from 1975 to 1979 (523 events) in desirable to sample for as long a period as possible and to collect
Bangladesh and examined their climatology using meteorolog- data from a variety of data sources in order to produce a reliable
climatological study. Although Afrose et al. (1981) investigated
⁎ Corresponding author. Present affiliation: Center for Southeast Asian
many “tornado” and nor'wester events (523 cases), the period of
Studies, Kyoto University, Kyoto, Japan. study was short (from 1975 to 1979), while Ono (1998) in-
E-mail address: yyamane@s06.mbox.media.kyoto-u.ac.jp (Y. Yamane). vestigated 44 “tornado” events referring to a short period of

0169-8095/$ – see front matter © 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.atmosres.2009.11.004
 Y. Yamane et al. / Atmospheric Research 95 (2010) 400–406 401

Fig. 1. Annual frequency of severe local convective storms in Bangladesh from 1990 to 2005. The maximum (year), minimum (year) and annual average are shown
on the upper side of the figure.

5 years (from 1990 to 1994). Chowdry and Karmakar (1986) did convections including not only the wording “tornado” and
not state the number of events examined in their study. But, the “nor'wester” but also wind gusts, hail and lightning. These
period of their analysis covered 10 years, as mentioned later, reports can be regarded as the evidence of occurrence of SLCS,
which is shorter than that of our study (16 years). and were used to investigate their climatology in Bangladesh.
Again, it is important to stress that the definition of tornadoes Wind gusts also can be induced by local topography and
and nor'westers used in Bangladesh, based on arbitrary thresh- produced by tropical cyclones. Since Bangladesh is almost
olds of wind speeds, may be considered flawed. It is also doubtful entirely flat, no damaging gust produced by topographically-
whether the wind speed was accurately estimated. Hence, induced windstorms has been reported. We also excluded
Chowdry and Karmakar (1986), that showed the climatology of wind damage reports with tropical cyclones by utilizing best
nor'westers, may have included tornadic events in their study. track data and meteorological data. Based on these points,
Ono (1998) may have included non-tornadic events in his we can state that the events analyzed in this study rea-
climatology of “tornado”. sonably can be considered as related to local deep convec-
In this study, in order to exclude such confusion, we did tion. Our objective is to produce a reliable database upon
not discriminate between tornadoes and nor'westers. We which we can study the climatology of convective storms in
collected reports of severe weather accompanied with deep Bangladesh.

Fig. 2. Monthly frequency of severe local convective storms in Bangladesh from 1990 and 2005.
402 Y. Yamane et al. / Atmospheric Research 95 (2010) 400–406

Fig. 3. Hourly frequency of severe local convective storms in Bangladesh from 1990 to 2005. BST means Bangladesh Standard Time (GMT + 6 h).

Fig. 4. Geographical distribution of severe local convective storms in Bangladesh from 1990 and 2005. Dots are located at the headquarters of districts and the
shade and size of a dot indicates the number of events for each district. D: Dhaka (90.3°E 23.7°N), T: Tangail (89.9°E 24.2°N), N: Netrakona (90.7°E 24.8°N), S:
Sirajganj (89.6°E 24.0°N), C: Comilla (91.1°E 23.4°N) and P: Pabna (89.2°E 23.9°N), respectively.
 Y. Yamane et al. / Atmospheric Research 95 (2010) 400–406 403

Fig. 5. Same as Fig. 4, except for in March. B: Barisal (90.3°E 22.8°N), L: Lakshmipur (90.8°E 22.9°N), respectively.

2. Data and method of mean sea level pressure (contoured at 1 hPa intervals) and
wind vector maps were prepared by utilizing ECMWF 40 years
In the present study, we produced a database of SLCS in re-analysis data (ERA-40) and NCEP FNL data. ERA-40 and NCEP
Bangladesh based on the survey of storm reports published in the FNL data are available on the website of ECMWF and NCEP,
local literature (mostly media-based literature). This same respectively. The area of influence of the tropical cyclone was
approach has been employed by many researchers. Niino et al. delineated by the outmost closed isobar associated with a tropical
(1997) developed the climatology of tornadoes in Japan based on system. Any report placed within this area of influence was
the survey of newspapers and reports compiled by the Japan excluded from the database in the present study.
Meteorological Agency. The newspapers surveyed here included The database of SLCS constructed in the present study
“The Bangladesh Observer” in English and “Daily Ittefaq” in contains date, location (district and upazila1), duration, type
Bengali. In addition, the “Bangladesh Disaster Report” complied of phenomena (e.g., hail and lightning) and the damage to
by the Disaster Forum (NGO in Bangladesh) and the annual property. The district of occurrence is identified for most
reports of Bangladesh Space Research and Remote Sensing reports, and the upazila is also identified in some events. The
Organization (SPARRSO) were also used in this study. We occurrence time is identified for some events. Details of
searched and collected articles and reports of severe weather damage (e.g., the number of deaths and injuries, the damage
utilizing these data sources. to buildings and crops) are also included in the database.
In our database the exclusion of wind damage reports asso- In this study, we collected 2,324 SLCS events from 1990 to
ciated with tropical cyclones was accomplished as follows. First, 2005 in Bangladesh. There were more events and the period
we searched for tropical cyclone activity affecting Bangladesh by of analysis is longer than those of previous studies.
using the best track data made available by the Joint Typhoon
Warning Center (JTWC). This best track data is available on the
JTWC website (https://metocph.nmci.navy.mil/jtwc.php). Then, 1
Bangladesh is divided into 64 districts. Moreover, each district is divided
we defined the size of tropical cyclones by producing an analysis into a number of Upazilas (sub-district).
404 Y. Yamane et al. / Atmospheric Research 95 (2010) 400–406

3. Results wind shear between the surface and the midlevel of the
troposphere in Bangladesh using ERA-40. They showed that
3.1. Annual frequency both CAPE and vertical wind shear are high during the pre-
monsoon season with a peak in April. Atmospheric conditions
Fig. 1 shows the annual frequency of SLCS in Bangladesh from displaying high CAPE and strong vertical wind shear are favorable
1990 to 2005. The maximum is 288 in 1990, and the minimum is for convective storms (e.g., Brooks et al. (2003)). Thus, it appears
38 in 1999. The annual average is 145. The number of events that the environmental conditions observed during the pre-
fluctuates largely from year to year, and no obvious trend was monsoon season in Bangladesh are not only conductive to local
found. convection in general but also favor the severe mode of local
convection. This is in agreement with the fact that many of the
3.2. Monthly frequency convective events sampled in the present study were based on
reports of phenomena associated with severe thunderstorms.
Fig. 2 shows the monthly frequency of SLCS in Bangladesh
from 1990 to 2005. SLCS occurs frequently during the pre- 3.3. Hourly frequency
monsoon season from March to May. Of the total number of
events, 84% occur during the pre-monsoon season. The peak Fig. 3 shows the hourly frequency of the occurrence of
was found in April, with 829 events, which is 36% of the total SLCS. The time of occurrence was identified for 224 events.
number of events. The frequency decreases sharply in the The frequency has two peaks. One is between 16 Bangladesh
monsoon season. From September in the late monsoon season Standard Time (BST) and 17 BST with 21 events, and the other
to October, the frequency slightly increases. SLCS is rare during is between 20 BST and 21 BST with 22 events. The frequency
the winter season (December–February). is large in the evening and small at midnight and early
Yamane and Hayashi (2006) showed the seasonal variation morning. The minimum of the frequency is between 4 BST
of Convective Available Potential Energy (CAPE) and the vertical and 5 BST with no event.

Fig. 6. Same as Fig. 4, except for in April.

 Y. Yamane et al. / Atmospheric Research 95 (2010) 400–406 405

Although this hourly frequency analysis was based on a lower population density in these two areas may account for
very limited number of reports containing time of occurrence this finding since the number of reports can be influenced by the
(less than 10% of the total amount of reports), we have the population distribution (King, 1997).
strong perception –– based on years of experience observing Afrose et al. (1981) showed the frequency of “tornado”
convective activity in Bangladesh –– that the results found and nor'wester was the least in the Pabna (89.2°E and 23.9°N
here are representative of the typical hourly distribution of (indicated as “P” in Fig. 4)), while our results show a relatively
SLCS in Bangladesh. high frequency in Pabna (Fig. 4).
We investigated the geographical distribution of the
3.4. Geographical distribution monthly frequency during the pre-monsoon season, when
SLCS is dominant. Figs. 5–7 show the geographical distribu-
Fig. 4 shows the geographical distribution of SLCS across tion of SLCS in March, April and May. In March, the events
Bangladesh. A dot locates at the headquarters of a district, and occur more frequently in the central part of Bangladesh, for
the shade and size of the dot indicate the magnitude of the example, Dhaka, Comilla, Sirajganj and Pabna. In addition, the
frequency for each district. SLCS occurs most frequently in events are more frequent in Barisal (90.3°E and 22.8°N
Dhaka (90.3°E and 23.7°N (indicated as “D” in Fig. 4)) with 111 (indicated as “B” in Fig. 5)) and Lakshmipur (90.8°E and
events. In addition, SLCS occurs more frequently in Netrakona 22.9°N (indicated as “L” in Fig. 5)) near the Bay of Bengal. In
(90.7°E and 24.8°N (indicated as “N” in Fig. 4)) with 109 events, April and May, the events appear to spread toward the
Tangail (89.9°E and 24.2°N (indicated as “T” in Fig. 4)) with 90 northern area of Bangladesh. SLCS is the most active in April,
events, Sirajganj (89.6°E and 24.0°N (indicated as “S” in Fig. 4)) especially over the central region of Bangladesh. In May, SLCS
with 83 events, Comilla (91.1°E and 23.4°N (indicated as “C” in occur more frequent in Nilphamari (88.9°E and 25.9°N
Fig. 4)) with 83 events. Thus, SLCS occurs most frequently in the (indicated as “N” in Fig. 7)) and Rangpur (89.3°E and
central part of Bangladesh. On the contrary, SLCS is rare in the 25.7°N (indicated as “R” in Fig. 7)) in the northwestern area
southwestern and eastern hill regions. To some extent, the of Bangladesh.

Fig. 7. Same as Fig. 4, except for in May. N: Nilphamari (88.9°E 25.9°N) and R: Rangpur (89.3°E 25.7°N), respectively.
406 Y. Yamane et al. / Atmospheric Research 95 (2010) 400–406

4. Conclusions the effort to collect events and renew our database needs to
be continued for a longer time in the future.
This paper presents the climatology of Severe Local
Convective Storm (SLCS) in Bangladesh. We constructed a Acknowledgments
vast database of SLCS from 1990 to 2005 in Bangladesh by
surveying the literature. The database includes more events This study was supported by Program for Improvement of
(2,324 events) for a longer period than those of previous Research Environment for Young Researchers from Special
climatological studies. We attempted to provide reliable Coordination Funds for Promoting Science and Technology
climatological characteristics of SLCS in Bangladesh based on (SCF) commissioned by the Ministry of Education, Culture,
our database. The results of this study are as follows: Sports, Science and Technology (MEXT) of Japan. This work
was also supported in part by Global COE Program: “In search
• SLCS occurs most frequently during the pre-monsoon of Sustainable Humanosphere in Asia and Africa”, MEXT,
season, peaking in April. The occurrences sharply decrease Japan. The best track data used in the present study are from
in the monsoon season and slightly increase in September Joint Typhoon Warning Center (JTWC). The re-analysis data
and October. The frequency of occurrences is the least in used in this study are NCEP FNL data and ECMWF 40 years re-
November. analysis data.
• There are two peaks of the hourly frequency between 16 BST
and 17 BST, 20 BST and 21 BST. The frequency is high in the References
evening and low in the early morning and midnight.
• SLCS frequently occurs in the north central part of Bangla- Afrose, S., Munim, S.S., Begum, Q.N., Choudhury, A.M., 1981. Investigation of
nor'westers and tornadoes in Bangladesh by weather satellite pictures.
desh, peaking in Dhaka district and is rare in the southwest-
Nucl. Sci. Appl., Ser. B 12&13, 100–104.
ern part and the east hill region. Brooks, H.E., Lee, J.W., Craven, J.P., 2003. The spatial distribution of severe
• The occurrence tends to spread toward the northern area of thunderstorm and tornado environments from global reanalysis data.
Bangladesh during the pre-monsoon season. Atmos. Res. 67–68, 73–94.
Chowdry, M.H.K., Karmakar, S., 1986. Premonsoon nor'westers in Bangladesh
with case studies. Proc. Seminar on Local Severe Storms, Dhaka,
It was reported that the geographical distribution of SLCS Bangladesh Meteorol. Dep, pp. 147–166.
may depend on the distribution of the population to some King, P., 1997. On the absence of population bias in the tornado climatology
of southwestern Ontario. Weather. Forecast 12, 939–946.
extent. The number of reports may be small in a less populous
Niino, H., Fujitani, T., Watanabe, N., 1997. A statistical study of tornadoes and
region. The small number of SLCS in the southwestern and waterspouts in Japan from 1961 to 1993. J. Climate 10, 1730–1752.
eastern hill regions may be due to these regions being less Ono, Y., 1998. Climatology of tornadoes in Bangladesh, 1990–1994. J. Met.
populous to some extent. In addition, we should mention that 22 (222), 325–340.
Yamane, Y., Hayashi, T., 2006. Evaluation of environmental conditions for the
the hourly frequency presented in this study is based on a formation of severe local storms across the Indian subcontinent.
very limited database. In order to obtain more reliable data, Geophys. Res. Lett. 33, L17806. doi:10.1029/2006GL026823.