Tornados in Bangladesh and East India

Jonathan D. Finch

National Weather Service, Dodge City KS

Ashraf M. Dewan

Department of Geography, University of Dhaka, Dhaka Bangladesh


A comprehensive spring (March-May) tornado climatology was developed for the eastern part of the Indian subcontinent (Bengal). This region includes most of Bangladesh as well as the Indian states of Assam, Tripura, West Bengal and Orissa, and generally encompasses the area between 86 and 94E and 21 and 27N. A plethora of sources was used in this study, including an exhaustive search of each daily March-May issue of the Bangladesh Observer (Pakistan Observer before 1972). Most tornados occurred from late-March to early-May, with a majority in early to mid-April. For the period of 1838 to 2001, 85 tornados are listed, 24 which killed at least 100 people. A majority (about 75%) of these occurred in a relatively small area of central, southern and southeastern Bangladesh (about 8,000 sq mi), which is about the size of the state of New Hampshire in the United States. Thirty-four of the tornados were previously undocumented except in newspaper reports, 15 which killed at least 30 people.

1. Introduction

Several climatological studies of tornados for the Indian subcontinent have been conducted. The most comprehensive works were by Petersen and Mehta (1981 and 1995), which documented 36 possible spring tornados across Bengal, 18 of which killed 10 people or more. Twelve of these occurred from 1838 to 1963, and 24 occurred after 1968. There was a large gap from 1889 to 1968 with no tornados for Bangladesh, and only two for India. Ono (1996) listed 28 spring events for 1990-1994 based on a search of each issue of the Bangladesh Observer, and found a tornado maximum in May. Six of the 28 events killed 10 people or more. Ono(1996) used a more liberal definition of a tornado. Goldar et al. (2001) documented 36 possible spring tornados for Bengal, 14 of which killed 10 people or more. While some events may not have been tornadic, this study partially fills the gap from the 1890's to early 1900's. Other studies such as Singh (1981) have listed a few tornados, but for India only. Each of these studies taken individually contains 40% or less of the tornados that are contained in the present study, and many of these events were not counted in the present study due to a lack of supporting evidence.In addition, many case studies have documented one or several tornados. All of these efforts and the present study have culminated in a comprehensive, spring tornado climatology for Bengal.

2. Methods

a. Tornado documentation

The authors examined each March-May issue of the Bangladesh Observer (Pakistan Observer before 1972) back to 1953. Cross-checking was performed using 'The Daily Ittefaq' and 'Sangbad', Bengali newspapers. The climatology was generally restricted to the area between 85 and 94E and 21 and 27N.Worldwide web disaster databases also provided valuable information for this study. The joint OFDA/CRED disaster database (, founded by 'The Office of U.S. Foreign Disaster Assistance' and 'The centre for Research on the Epidemiology of Disasters') provided documentation for several tornados. The British Association for Immediate Care ( also provided additional details.

b. Tornadic versus non-tornadic events

Since housing construction is so poor in Bangladesh, people are often killed by straight-line winds. Collapsing of roofs and capsizing of shps are responsible for many deaths. The collapse of a few poorly constructed huts, resulting in the deaths of several family members, is hardly evidence of a tornado. The same can be said of a 30 minute storm that collapses a church roof killing 30 people, or boat capsize killing 100 people. In addition, weaker tornados are indistinguishable from derechos and microbursts since documentation for such events is extremely limited. Newspapers tend to ignore events that are less than catastrophic. Since it is impossible to differentiate less than catastrophic tornados from straight-line winds, this research focuses on the more violent storms that are more likely to be tornadic.

There was a need to differentiate tornados from tropical cyclones. Newspapers often refer to tornadic events as gales, whirlwinds or cyclones. For example, after the tornado on 16 April 1967 in southern Bangladesh that killed 77 people, the 'Commoner', a Nepalese newspaper, referred to the tornado as a cyclonic storm. On the other hand, some events in the newspapers and literature are called tornados with no supporting evidence. The authors disregarded the events corresponding to the dates of land-falling tropical systems. In some cases, upper air charts were analyzed to make sure the event was not tropical in nature. High winds associated with a land-falling tropical system can simply not be distinguished from the accompanying tornados, especially since the term cyclone was used for tropical systems and tornados. While tornados are usually reported on days when tropical cyclones are absent and storminess is isolated, they are either ignored or irrelevant when large tropical systems are making landfall. In addition, summer and fall tornados were generally weak and caused less than catastrophic damage. These types of weak tornados are not possible to distinguish from non-tornadic events. Therefore, this climatology is strictly limited to March-May.

c. Single versus multiple tornados

The lack of intricate detail of tornados in the Bengal region precludes the division of tornado families into individual tornados, as is often done in the United States. For example, the well-documented tornado on 13 May 1996 that killed over 700 people was probably a family of several tornados since the damage path was intermittent (Ono 1996). So the actual number of violent tornados per area cannot be directly comparable to the United States or other countries. This research documents the dates and locations of violent tornadic storms rather than individual tornados.

d. Tornado thresholds

An event was classified a tornado if any of the following criteria were met. 

Thresholds are regrettable but necessary in the development of a tornado climatology. Unless there were at least 15 deaths, corroborating evidence was required to label an event a tornado. Even if at least 15 people were killed, an event was not labeled a tornado if the documentation suggested the presence of widespread straight-line winds. 

3. Bengal Tornado Climatology

a.Geographical distribution

A list of 85 tornados to strike Bengal, along with the date, time, location, latitude and longitude, number of people killed, path width, path length, direction of movement, and tornado justification criterion, are all shown in Table A. The number in the last column refers to the corresponding bibliographical reference. The tornados and larger towns of Dacca (D), Jessore (J), Faridpur (F), Madaripur (M) and Chandpur (C), are mapped in Fig. 1, clearly indicating that fewer tornados have been reported in northeast and northwest Bangladesh, and that part of southeast Bangladesh to the east of 91.5E. This figure also shows that southern and central Bangladesh between 22.5N and 24N has the highest incidence of tornados. The locations and number of deaths of the tornados that caused 30 or more deaths are plotted in Fig. 2. The Julian dates of the 43 tornados that killed 30 or more people are shown in Fig. 3.These figures clearly indicate that the highest frequency (about 75%) of the most deadly tornados is across central, south central and southeast Bangladesh, covering an area about 8,000 sq mi. This is about the size of the state of Massachusetts in the United States. 

The area around Dhaka to about 80 km south of Dhaka, or that area around and between Dacca, Faridpur, Madaripur and Chandpur, is especially tornado prone. Three violent tornados have struck Nadia and Bhederganj in southern Bangladesh. In fact, Kalikaprashad village under Naria police station was devastated by two tornados in back to back years on 16 April 1967 and 11 April 1968. Another devastating tornado swept through Naria on 10 April 1976. All three of these tornados were documented in the 'Pakistan Observer', and killed 77, 200 and 46 people respectively. This area is about 45 miles south of Dhaka, and is not the most populated area of Bangladesh. The city of Dacca has been hit by at least four violent tornados that this study was able to trace. These tornados occurred on 7 April 1888 (Crombie, 1888), 12 April 1902 (Dhaka district Gazetteer, 1976), 14 April 1969 (15th-22 April 1969 issues of the Pakistan Observer), and 5 April 1972 (2-5 April 1972 issues of Bangladesh Observer), and killed 118, 88, 660, and 75 people respectively. All of the 85 tornados documented here occurred at or north of 21N and all but three occurred north of 22N. Seven of the 85 tornados occurred north of 25N. Therefore, 75 out of 85 of the tornados occurred between 22N and 25N. All but five of the tornados occurred between 88E and 91.5E. The highest concentration of the most violent tornados (30 or more deaths) occurred in a 4000 sq mi area of central Bangladesh,or along and between 23N and 24N and between 89.5 E and 90.6 E. This is a little smaller than the state of New Hampshire in the United States.

It must be noted that this research focused primarily on Bangladesh. The Bay of Bengal and the coastal mountain range just inland in East India restricts the area susceptible to tornados. Only a narrow strip along coastal East India is usually threatened by tornados, since spring moisture return over the Bay of Bengal is generally directed to the north and northeast across Bangladesh, instead of to the west into India. Three catastrophic tornados have occurred in the Indian state of Orissa. On 16 April 1978, a tornado moved south-southeast from west of Anandpur to southwest of Ramchandrapur (21N), killing 173 people. Another tornado on 17 April 1981 near Patna (21.7N) killed 120 people (Singh 1985). These two tornados are the southernmost on record for the Bengal region. On 24 March 1998, a tornado or tornado family killed 250 people along the Orissa-West Bengal border near Dantan (Associated Press). In fact, most tornados in east India have occurred very close to the Bangladesh border. One major tornado on 9 April 1993 killed 140 people near Kandi in West Bengal, India. This tornado was reportedly anticyclonic (Misra et al. 1996). On 8 April 1838 a tornado killed at least 200 people in the eastern suburbs of Calcutta (Crombie 1888). Since most of Bangladesh to the south of 22.5N is swampland and less populated, many tornados in this area have undoubtedly gone undocumented.

b. Seasonal and diurnal distribution

The total number of tornados for each third-month period (10 or 11 days) for March-May is shown in Fig. 4. The number of tornados that killed 30 or more and 100 or more people for the same third-month periods are also shown. This figure illustrates that 19 out of 25 (80%) of the tornados that killed 100 or more people, and 32 out of 43 (75%) of the tornados that killed 30 or more people, occurred during the first 20 days of April. Sixteen of the 25 (64%) tornados that killed 100 or more people occurred from April 8 to April 18. Out of the 20 central Bangladesh tornados that killed 30 or more, 12 (60%) occurred between April 5 and April 14. The 13 May 1996 tornado in Tangail that killed more than 700 people in northern and central Bangladesh was extremely unusual in that it occurred so late in the season. On the other hand, the tornado on 19 March 1961 in central Bangladesh that killed 210 people was unusually early in the season. In fact, this is the earliest spring tornado on record that killed 30 people or more. These findings strongly indicate that the violent tornado season for Bengal is very short, covering mainly the first 3 weeks of April. Almost all of the documented tornados occurred in the afternoon or evening, especially the most catastrophic tornados, with a peak around 1630 BST (Bangladesh local time). Very few violent tornados occurred between 2100 and 1200 BST.

c. Movement of tornados

The forward speed of Bengal tornados could only be ascertained in a few cases and the direction of movement could be ascertained in about half the cases. The preferred movement was to the southeast or south-southeast. A few of the tornados moved to the south, and a few to the east or northeast. The paths were not always along a straight line. The Manikganj tornado on 17 April 1973 moved in a zig-zag fashion to the southeast (Hasan 1979). The Saturia tornado on 26 April 1989 initially moved east, then hooked north before lifting (Bangladesh Met. Dept. Memo 1989). The Muhammadpur tornado on 11 April 1964 moved to the southwest, then curved to the southeast in a horseshoe fashion (Pakistan Observer). The tornado on 8 April 1838 that ravaged the eastern suburbs of Calcutta had a classic motion of south-37E (Floyd 1888). What makes this a very unusual tornado is the 2.5 h it took to traverse the 16 mile path. The anticyclonic Kandi tornado on 9 April 1993 moved to the east and northeast based on damage assessment (Misra et al. 1996). There is generally not enough information to determine the speed of the tornados. In fact the speed has been documented for only 2 tornados, and these were based on ground surveys. The Tangail tornado on 13 May 1996 moved about 40-45 kph to the south-southeast (Ono 1997), while the Cooch Behar tornado on 19 April 1963 moved about 75 kph to the east-southeast (Nandi and Mukherjee 1966).

d. Variations over time

Prior to 1889 there were several documented cases of tornados across Bengal. After 1888 and prior to 1951, the authors could only document 5 tornados, based mostly on work by Goldar (1985). After 1960 the reported number of tornados dramatically increased, mainly due to the detailed reports in the Pakistan Observer (Bangladesh Observer after 1971). The period from 1967 to 1973 was catastrophic, with 12 killer tornados that caused 30 deaths or more. Since 1985, several violent tornados have occurred outside the peak season. Only three have occurred in early to mid-April, while four have occurred either in May or late-March. This is in contrast to the 1963 to 1977 period when nearly all of the most violent tornados occurred in early to mid-April.

e. Tornado risk

The population of Bangladesh has exploded in the past 40 years. Dhaka's population exceeds 10 million as of 2000, which represents a 10-fold increase since 1960. The population of Bangladesh has reached 150 million and its geographical area is about the same as the state of Oklahoma. Its high population density and 3rd world status makes Bangladesh extremely vulnerable to tornados. 

f. Fall and winter tornados

Although this research focused on spring tornados, there have been two noteworthy fall and winter tornados in Bangladesh . On 8-9 Jan 1993, tornados struck northern Bangladesh, resulting in over 50 deaths (Ono,1997). On 12 October 1997, 25 worshipers were killed by a tornado in the northern suburbs of Dhaka(Associated Press).Many other weak tornados have likely occurred in tropical weather regimes that are impossible to distinguish from tropical cyclones or derechos.

4. Illustrative cases of extremely violent tornados

A few examples of violent tornados which have struck Bangladesh and east India are presented here. These examples were selected because of the detailed information that was available.

1) 8 April 1838

The details of this tornado were furnished by engineer Mr. J. Floyd as communicated by J. H. Patterson, Esq., magistrate of the 24-Pergunnahs district of East India (Floyd, 1888). The path length of this devastating tornado that killed 215 people on the eastern suburbs of Calcutta, IN was about 26 km, the width of which varied from a quarter to a half-mile. This tornado moved very slowly, and some speculated that this was the reason for the massive destruction. According to Mr. Floyd, the tornado took 2.5 hours to move 26 km. In his own words, 

"In some places, not a house nor tree was left standing. In fact everything that opposed its progress was leveled to the ground. Coconut and date trees were twisted out of the ground and thrown 2-300 ft. "A slight bamboo was projected horizontally through a raised tile walk, which pierced through the whole breadth, breaking the tile on both sides. The grass in some places was completely scoured from the ground. Fisherman returned to their villages, with only a few left to account for the occurrence. There was little rain but a severe fall of hail." Hail was measured at 3.5 lbs at Dum-Dum. "A peepul tree, which had been standing time out of mind, and to the knowledge of the oldest inhabitants had never lost a bough, was the first to fall. The circle from whence the roots sprung was 35 ft. in diameter, and these being of extraordinary length, caused the earth to come away from the tree, leaving a chasm of about 38ft. in width and 14 ft. deep. Most of its stouter branches were wrenched off, and thrown to a sufficient distance as to attest to the violence of the winds. The bark of palm trees was peeled off as with a knife." 

2) 19 April 1963

On 19 April 1963 at 1900 IT, a tornado touched down near Cooch Behar, IN (about 20 miles from the base of the Himalaya mts.), and traveled east-southeast at 80 kph for about 32 km. This tornado clipped the extreme northern tip of Bangladesh, and is the northernmost devastating tornado on record for Bengal. 139 people were killed in India and more in Bangladesh. Pieces of scrap metal were found 48 km from the nearest point of the tornado. Persons were picked up and fatally thrown 300-600 m. The tornado was 100-200 m wide in most places but up to 250 m wide.

3) 11 April 1964

According to the Pakistan Observer (April 12-19th issues, 1964), a violent tornado developed northeast of Muhammadpur and moved to the southwest through seven villages of Muhammadpur union to the Nabaganga river. Upon reaching the river, the tornado turned to the southeast and widened as it entered Narail zila, demolishing 27 villages in Naogram, Naldi, Chandibarpur, Kashmipur and Lakhipassa unions. The tornado ended just west of Lohagara. The tornado track was in the shape of a horseshoe. Many dead bodies were found hanging in treetops. There was no sign of any of the 400 inhabitants of village Bhabanipur seven days after the tornado. According to the Pakistan Observer, "It's is only a matter of formality to proclaim this (Bhabanipur) population as dead". Three unions comprising seven villages of Mohammadpur under Magura subdivision were wiped out without any trace of human habitation. The Pakistan Observer showed pictures of cooking utensils lodged in trees. The unofficial death toll was over 500 as of April 14. Although the official body count on April 18 was 280, hundreds of people were still missing. The joint OFDA/CRED disaster database (founded by 'The Office of U.S. Foreign Disaster Assistance' and 'The centre for Research on the Epidemiology of Disasters') shows a death toll ranging from 300 to 1400. The final death toll will never be known. The total path length of the tornado was about 32 km. This tornado ranks as one of the deadliest tornados on record for the world.

4) 14 April 1969

On 14 April 1969, two and possibly three devastating tornados struck central and eastern Bangladesh as documented by the Pakistan Observer (April 15-21st 1969 issues). Many victims were horribly mutilated. One 18 year-old boy was flung 400 m into a pond. According to a government official upon his visit to the scene, "I saw what no amount of words could aptly describe. Damage was colossal. Tragedy was harrowing." One tornado struck the northeast suburbs of Dhaka, killing 660 people. Another separate tornadic storm struck 48 km to the east in Comilla, killing 263 people (Mowla, 1969).

5) 1 April 1972

On 1 April 1972 a vicious tornado struck northern Bangladesh around 1830 BT, killing at least 200 people. The details of this tornado were provided by theBangladesh Observer (2-5th April 1972 issues). This death toll was only two days after the tornado and the final death toll was probably much higher. A 24 km by 1.6 km area was wiped clean. Trees were carried for a mile. Innumerable houses were blown away leaving no trace. It was as if the place was "leveled by a 1000 giant dozers". The Bangladesh Observer proclaimed Fulbaria, a town just southwest of Mymensingh, a dead village. Crumbled, corrugated iron sheets were found perched in tree-tops miles away.

6) 17 April 1973

In one of the worst tornado tragedies to hit Bangladesh, 681 people were killed in the Manikganj subdivision of Dhaka district on 17 April 1973. The details of this tornado were provided by both the Bangladesh Observer (18-23 April 1973) and Hasan (1985). In Balurchar not a single dwelling was traceable according to the Prime Minister. This village was completely leveled. Almost all houses were leveled in the eight villages along the Kaliganga river. "Uprooted trees were cris-crossed and bodies were strewn all around. An unofficial death toll was over 1000. There were two funnels that merged together to form one large tornado that moved in a zig-zag path. A boat with 3 people was blown 1000 m from the bank. A pacca concrete structure that was 4.6 m high and 2.7 m in diameter was thrown a few meters from its original position.

7) 1 April 1977

A tornado struck Madaripur and Shibchar around 1600 BT and killed 500 people as described in Ahmed (1977). Forty-three bodies were found floating in the river. According to a spokesperson, "During my visit to the villages battered by Friday's tornado, I found that destruction was complete. Not a single dwelling nor a tree I found standing erect." Corrugated iron sheets were "blown like kites" and hit many villagers. Solenama village was completely devastated.

8) 26 April 1989

This was possibly the deadliest tornados on record in the world as documented by (Bangladesh Meteorological Department, 1989) This tornado moved east and eventually northeast from Daultapur to Saturia. The path was 13 km long and 1.6 km wide. The affected area had been in a 6 month severe drought prior to the tornado. All houses in a six sq km area were completely destroyed. Thousands of trees were uprooted and blown away. According to the Bangladesh Observer, "The devastation was so complete, that barring some skeletons of trees, there were no signs of standing infrastructures".

9) 13 May 1996

This was certainly one of the most devastating tornadic storms ever to strike Bangladesh. Over 700 people were killed. Hail the size of softballs pummeled the affected area (Ono 1997). This was a family of tornados that moved to the south-southeast at 45 km/hr for 80 km. Sixty-five bodies were found suspended in trees. 30,000 houses were destroyed and 1600 cattle were lost. A 600 kg pump machine was thrown 56 m. The Tangail district was the hardest hit. Many people were blown long distances (Ono 1997). One person was blown 1.5 km. 

i. Other deadly thunderstorms

Many events were not classified as tornados in this study. This does not mean that tornados did not occur, just that there is doubt as to whether the destruction or loss of life was caused by a tornado. Table B lists 88 such events in chronological order, along with a brief description when available. Some of the events were counted as tornados in other studies and are denoted by a star to the left of the date.

5. Conclusions

A comprehensive climatology of violent tornados was developed for the Bengal region. The primary data souces included the Bangladesh Observer (Pakistan Observer prior to 1972), previous Bengal tornado studies, and world-wide web databases. Although a few tornados were documented for Bengal in early to mid-March and in mid to late-May, they are most likely in early to mid-April. Most violent tornados occur in the afternoon and evening, andmoved to the south-southeast or southeast.All tornados occured between 21N and 26.5N and 85.7E and 91.5E. Indian tornados are less common, with most occurring close to the Bangladesh border. The highest concentration (20 out of 43) of the most violent tornados (30 or more deaths) occurred in a small area (4000 sq. km.) in central Bangladesh, generally along and north of 23N and along and south of 24N and between 89.5E and 90.6E.About 72% of the most violent tornados occurred in a 8000 sq mi area of central, south central and southeast Bangladesh.