You are on page 1of 5

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/6035779

The potential for giant tsunamigenic earthquakes in the northern Bay of


Bengal

Article  in  Nature · October 2007


DOI: 10.1038/nature06088 · Source: PubMed

CITATIONS READS

148 1,844

1 author:

Phil R. Cummins
Australian National University
152 PUBLICATIONS   3,287 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Tsunami modeling View project

Field Survey: Ambient Noise Tomography (ANT) in Western Part of Java Region View project

All content following this page was uploaded by Phil R. Cummins on 01 January 2014.

The user has requested enhancement of the downloaded file.


Vol 449 | 6 September 2007 | doi:10.1038/nature06088

LETTERS
The potential for giant tsunamigenic earthquakes in
the northern Bay of Bengal
Phil R. Cummins1

The great Sumatra–Andaman earthquake and Indian Ocean tsu- subduction zones, which have at most a 2-km-thick sediment layer
nami of 2004 came as a surprise to most of the earth science on the upper surface of the subducting plate7, the northern Bay of
community. Although it is now widely recognized that the risk Bengal is unique because it hosts the world’s largest submarine fan
of another giant earthquake is high off central Sumatra1, just east system, the Bengal Fan, consisting of sediments that have been shed
of the 2004 earthquake, there seems to be relatively little concern off Tibet and the Himalayas since the Early Miocene. The thickness of
about the subduction zone to the north, in the northern Bay of the Bengal Fan sediments reaches up to 20 km (ref. 8). Because even a
Bengal along the coast of Myanmar. Here I show that similar 1-km-thick sediment cover can insulate the underlying plate enough
indicators suggest a high potential for giant earthquakes along to cause significant up-dip extension of the thermal regime required
the coast of Myanmar. These indicators include the tectonic envir- for seismogenesis9, it seems reasonable to expect this effect to be
onment, which is similar to other subduction zones that experi- particularly pronounced for the extreme thickness of the Bengal
ence giant megathrust earthquakes, stress and crustal strain Fan sediments. Such considerations support an alternative hypo-
observations, which indicate that the seismogenic zone is locked, thesis for subduction along the coast of Myanmar: that subduction
and historical earthquake activity, which indicates that giant tsu- is active and ‘‘rare great earthquakes may occur along a basal decolle-
namigenic earthquakes have occurred there in the past. These are ment’’4. This hypothesis is supported by studies in the 1990s that
all consistent with active subduction in the Myanmar subduction place the deformation front well offshore of the Myanmar coast,
zone and I suggest that the seismogenic zone extends beneath the not coming ashore until the Mehgna estuary at 23u N (refs 8, 10–
Bengal Fan. I conclude therefore that giant earthquakes probably 12). Marine seismic surveys12 demonstrate unequivocally that the
occur off the coast of Myanmar, and that a large and vulnerable deformation front is offshore of Ramree island, and borehole
population is thereby exposed to a significant earthquake and breakout data show that the stress field in the sedimentary pile of
tsunami hazard. the Bengal basin is oriented east–west13, suggesting that the
The tectonic setting of west Myanmar has been marked by Chittagong–Tripura fold belt is an active feature forming in response
considerable uncertainty and some controversy2,3. Research in the to ongoing subduction of the Indian plate beneath southeast Asia12.
1980s4,5 inferred that the surface expression of the boundary between This extension of the seismogenic zone beneath the Bengal Fan,
the Indian plate and southeast Asia follows the northward extension 100–200 km westward of the previously inferred plate boundary, has
of the Andaman trough, coming ashore on the coast of Myanmar just important implications for seismic and tsunami hazard. First, the
south of Ramree island at roughly 20u N latitude (Fig. 1a). This convergence is much less oblique than it would be along the north-
coincides with the extrapolation of the Wadati Benioff zone beneath ward extension of the Andaman trench (Fig. 1b), suggesting that
western Myanmar to the western flank of the Indo-Burma ranges, recurrence times of large thrust earthquakes might be only a few
and is the plate boundary indicated in most widely used plate models hundred years14. Second, it means that the shallow earthquakes in
(Fig. 1b). Most authors3–5 favoured the interpretation that eastward the Indo-Burma ranges with focal mechanisms exhibiting north–
subduction of the Indian plate has ceased and the slab beneath Burma south pressure axes are actually occurring in the forearc, 100 km
is being dragged northward along with the Indian plate. This con- landward from the deformation front. Both of these factors would
clusion seems to be supported by the fact that the focal mechanisms make the Arakan subduction zone similar to the Cascadia and
of shallow earthquakes near the presumed plate boundary have pres- Nankai subduction zones, which have oblique, partitioned subduc-
sure axes oriented north–south, implying no east–west compression tion, arc-parallel orientation of the maximum horizontal stress in the
and hence no active subduction. Even if subduction were active, this forearc15, and exhibit very little megathrust earthquake activity (apart
position of the plate boundary suggests that convergence is highly from the relatively infrequent occurrence of giant megathrust earth-
oblique (,5u), implying a very long recurrence interval for thrust quakes). Thick trench sediments, which in the Bay of Bengal far
earthquakes, and that the offshore segment of the plate boundary exceed those of any other trench, have been shown to have a positive
north of the 2004 Sumatra–Andaman earthquake’s rupture area correlation with subduction zone earthquake occurrence16. Finally,
would be only 300 km in length; hence, the likelihood of a giant extension of the seismogenic zone to offshore Myanmar means that
tsunamigenic earthquake in the northern Bay of Bengal would be the entire 900-km-long seismogenic zone, from the northern tip of
low6. the Andaman Trench to the northern tip of the Bay of Bengal, is
The up-dip limit of the seismogenic zone should not, however, be submarine, and therefore capable of hosting a giant tsunamigenic
presumed to lie where the slab begins to descend beneath the upper earthquake.
plate. Rather, the seismogenic zone extends as far out along the Along with the interpretation that there is no active subduction
decollement as necessary to encompass all of the significant strain between the Indian plate and southeast Asia, it has been suggested3
accumulation that might be released in a megathrust earthquake. that all of the relative motion between the Indian and Eurasian plates
Although this distinction may not be important for most other is accommodated along the Sagaing fault in central Myanmar.
1
Geoscience Australia, GPO Box 378, Canberra ACT 2601, Australia.
75
©2007 Nature Publishing Group
LETTERS NATURE | Vol 449 | 6 September 2007

Global positioning system (GPS) surveys14,17, however, have shown lost, and it is interesting to speculate that local knowledge of the
that only 60% of the oblique component of relative plate motion is tsunami threat caused inhabitants of Cheduba to flee to high ground
accommodated on the Sagaing fault. The remaining deformation when the earthquake occurred. Whether a tsunami affected other
could be explained either by distributed deformation west of the parts of the Bay of Bengal is less clear. There are no known reports
Sagaing fault, or by locking of the Arakan subduction zone. In the from the coast of Bengal (now Bangladesh and the Indian province of
latter case, the Arakan subduction zone ‘‘would be expected to pro- West Bengal). At Dhaka, however, the river rose suddenly and ‘‘hun-
duce a magnitude 8.5 earthquake every century or a magnitude 9 dreds of large country boats were driven ashore or lost, and great
every 500 years’’14. numbers of lives lost with them’’19. It is not clear, however, if this was
Many of the above studies have referred to historical earth- due to a wave from the sea or a change in land level that may have
quakes near the Myanmar coast, but none has considered in detail caused the river to back up.
the historical reports of the 2 April 1762, Arakan earthquake. The geological evidence of uplift along the Arakan coast provides
Descriptions of the effects of this earthquake and associated ground clear evidence of the occurrence of large thrust earthquakes there,
movements along the Arakan coast were recorded during a survey in and Halsted’s account indicates that rupture from the 1762 earth-
1841 by the British ship Childers18, commanded by Captain E. quake extended as far south as Foul island. It is less clear how far
Halsted. Halsted recorded evidence of 3–7 m of uplift along the north rupture from this earthquake extended. On the one hand, there
coasts of Ramree, Cheduba and Foul islands, which lie offshore of are no reports of geologic evidence for uplift north of Ramree island.
and parallel to the Arakan coast of Myanmar. Halsted remarked On the other hand, contemporary accounts19 from Chittagong
that the traces of this sudden change in land elevation were ‘‘as clear describe widespread liquefaction and subsidence. Although in many
as could be wished’’, and that on Cheduba ‘‘the natives are all per- cases submergence due to the expulsion of groundwater is difficult to
fectly aware of the bank having formerly been the limit of their distinguish from inundation by the sea, there are also accounts of
island’’. An interview with a local who experienced the 1762 earth- islands sinking to or below sea level, and a report that 60 square miles
quake, who ‘‘had been accustomed to fish over the now-upraised of land were ‘‘permanently submerged’’20, so it seems clear that at
land’’, confirmed that the uplift occurred during the earthquake. least some parts of the coast near Chittagong subsided.
Halsted also noted the existence of at least one additional upraised This subsidence is a crucial observation, because it implies that
terrace. Of particular concern for hazard assessment is Halsted’s fault rupture extended as far north as Chittagong, and that the up-dip
statement that ‘‘these elevations are considered periodical by the limit of rupture was well offshore, because subsidence would occur
inhabitants, occurring every hundred years, and the next event is only above the down-dip limit of fault rupture. The extension of
expected within the course of the next few years, and would excite the rupture at least as far north as Chittagong is supported by con-
but little surprise’’. temporary accounts describing intense ground shaking, extensive
Did the 1762 earthquake cause a tsunami? Halsted’s interviews damage to buildings, and formation of large fissures and landslides19,
with the inhabitants of Cheduba confirm that there was a local tsu- which are all observations consistent with a Modified Mercalli
nami: ‘‘the sea washed to and fro several times with great fury, and Intensity (MMI) of 9 or greater. Studies of fault rupture associated
then retired from the ground’’. Remarkably, no lives were reported with historical subduction zone earthquakes in Japan21, as well as

a b
80º 90º 100º 85º 90º 95º
Sagaing fault

Indo–Bu
Chittagong–Tripura
20º fold System
rma ran

Area of b
ges

10º
20º
Su
m
at
ra
fa
u


lt

36 mm yr–1 relative plate motion


(India versus Sunda plate)

0 500 km 10º
0 250 km

15º
Earthquake rupture areas
2004 (Mw = 9.15)
0 5 10 15 20
1861 and 2005 (Mw = 8.6) Sediment thickness (km)
1797 and 1833 (Mw = 8.5–8.7)

Figure 1 | Tectonic setting of the Bay of Bengal. a, Tectonic setting of the northern Bay of Bengal, showing the rough positions of the
eastern Indian Ocean, showing Sumatra earthquake rupture areas, the Chittagong–Tripura fold belt and Indo-Burma ranges with respect to the
subduction zone plate boundary (line with black triangles) from the plate boundary (line with black triangles) and deformation front (line with
University of Texas Institute of Geophysics plates model, and the red triangles). Sediment thickness is also indicated, along with the relative
deformation front11 (line with red triangles). b, A more detailed view of the motion of the Indian plate versus the Sunda plate14.
76
©2007 Nature Publishing Group
NATURE | Vol 449 | 6 September 2007 LETTERS

a b
85º 90º 95º 80º 85º 90º 95º
Dhaka
Dhaka
Kolkata
Chittagong
Kolkata
Chittagong
20º

NE

SW

20º
15º
Ramree

0 200 km Chennai
Cheduba

SW NE
10 Foul
Vertical displacement (m)

Cheduba
Foul 10º
5 Ramree
Chittagong

0 0 200 km

–5 15º 5º
0 50 100 150 200
Distance from deformation front (km)
0 0.5 1.0 1.5 2.0 2.5
Maximum offshore tsunami height (m)
0 500 1,000 1,500 2,000 2,500 3,000
Bathymetry depth (m)

Figure 2 | Models for the 1762 Arakan earthquake and tsunami. a, A fault is a southwest–northeast (SW–NE) profile of the vertical displacement
model for the 1762 Arakan earthquake, with length 700 km, width 125 km across the fault, with arrows indicating the down-dip positions of the
(indicated by the rectangle along the eastern coast of the northern Bay of observed uplift and subsidence. b, Maximum offshore heights of a tsunami,
Bengal) and slip 10 m. Parameters were chosen to reproduce roughly the calculated using the source in a. This calculation does not take into account
observed subsidence and uplift associated with the 1762 Arakan earthquake, run-up, which can be many times greater than the offshore heights.
with the fault’s upper edge coincident with the deformation front. The inset

studies of macroseismic data in Mexico22, suggest that such high Arakan subduction zone, it seems likely that the number of lives at
MMIs are only observed within 50 km of fault rupture. Finally, if risk may be over a million.
rupture had extended only as far north as Ramree island, the rise The Mw 5 8.8 earthquake proposed here as representative of the
of the river at Dhaka could be attributable only to a tsunami that had 1762 event is a worst-case scenario, and the details of the rupture
propagated along the strike of the subduction zone. Little tsunami model may be open to question. Also, it may be over 200 years before
energy would be propagated in this direction, so it seems unlikely that a similar event occurs, according to GPS observations of 23 mm yr21
a tsunami large enough to travel 100 km upriver to Dhaka would have accumulation of backslip along the Arakan trench14. The next tsuna-
been generated by an earthquake whose rupture stopped at Ramree migenic earthquake in the Arakan subduction zone may not be this
island. large, but it may also occur much sooner. It is also not certain that the
A tsunami was simulated for a moment magnitude Mw 5 8.8 1762 earthquake produced more than a local tsunami, although it
thrust earthquake along the Arakan coast, positioned so that its should be borne in mind that the rapid rate of sedimentation in the
upper edge lies along the deformation front11 and 10 m reverse slip Bay of Bengal could also lead to tsunamis caused by submarine land-
results in the observed pattern of uplift and subsidence (Fig. 2a). As slides. In any case, the evidence of active convergence accompanied
discussed above, this pattern requires the seismogenic zone to extend by thrust earthquake activity along a coastal region with an extremely
offshore, beneath the Bengal Fan, consistent with geologic inter- high population density suggests that the risk of a major tsunami in
pretations and stress observations. Such an earthquake would gen- the northern Bay of Bengal should be taken seriously.
erate a large tsunami that could have a pronounced impact on the
Received 27 February; accepted 10 July 2007.
Chittagong coast and the Ganges–Bhramaputra delta at the northern
tip of the Bay of Bengal (Fig. 2b). The latter region is home to over 1. Natawidjaja, D. H. et al. Source parameters of the great Sumatra megathrust
60 million people living within 10 m of sea level. Chittagong, with a earthquakes of 1797 and 1833 inferred from coral microatolls. J. Geophys. Res. 111,
population of over 6 million, lies just above the presumed fault, and is doi:10.1029/2005JB004025 (2006).
2. Satyabala, S. P. Subduction in the Indo-Burma region: Is it still active? Geophys.
the source of some of the most dramatic accounts of ground Res. Lett. 25, 3189–3192 (1998).
deformation and damage from the 1762 earthquake. Dhaka and 3. Guzmn-Speziale, M. & Ni, J. F. Comment on ‘‘Subduction in the Indo-Burmese
Kolkata, although they are some distance from the presumed rupture region: Is it still active?’’ by S. P. Satyabala. Geophys. Res. Lett. 27, 1065–1066
area, are urban agglomerations of over 10 million each, with a large (2000).
4. Le Dain, A. Y., Tapponier, P. & Molnar, P. Active faulting and tectonics of Burma
proportion of their populations living in structures that are unlikely and the surrounding regions. J. Geophys. Res. 89, 453–472 (1984).
to be seismically resilient. Even if only a small fraction of this popu- 5. Ni, J. F. et al. Accretionary tectonics of Burma and the three-dimensional
lation is vulnerable to a giant tsunamigenic earthquake along the geometry of the Burma subduction zone. Geology 17, 68–71 (1989).
77
©2007 Nature Publishing Group
LETTERS NATURE | Vol 449 | 6 September 2007

6. Kamesh Raju, K. A., Murty, G. P. S., Amarnath, D. & Mohan Kumar, M. L. The west 16. Ruff, L. Do trench sediments affect great earthquake occurrence in subduction
Andaman fault and its influence on the aftershock pattern of the recent zones? Pure Appl. Geophys. 129, 263–282 (1989).
megathrust earthquakes in the Andaman-Sumatra region. Geophys. Res. Lett. 34, 17. Vigny, C. et al. Present-day crustal deformation around Sagaing fault, Myanmar.
doi:10.1029/2006GL028730 (2007). J. Geophys. Res. 108, doi:10.1029/2002JB001999 (2003).
7. Scholl, D. Sediment influx to subduction zone and great magnitude earthquakes; 18. Halsted, E. P. Report on the Island of Cheduba. J. Asiatic Soc. Bengal 114, 319–446
an observational appraisal. In Tsunami Sources Workshop Proceedings (US (1843).
Geological Survey, Menlo Park, California, 21–22 April 2006). 19. Hirst, W. An Account of an Earthquake at Chattigaon: translated from the Persian
8. Alam, M., Alam, M. M., Curray, J. R., Rahman Chowdhury, M. L. & Royhan Gani, M. by Mr. Edward Gulston, in the Service of the Honourable East India Company, and
An overview of the sedimentary geology of the Bengal Basin in relation to the Communicated by Him to the Reverend Mr. Hirst. Phil. Trans. R. Soc. Lond. 8, 251
regional tectonic framework and basin-fill history. Sedim. Geol. 155, 179–208 (1763).
(2003). 20. Oldham, T. A catalogue of Indian earthquakes from the earliest time to the end of
9. Wang, K., Hyndman, R. D. & Yamamoto, M. Thermal regime of the southwest A.D. 1869. Mem. Geol. Surv. India 19, 163–215 (1883).
Japan subduction zone: effects of age history of the subducting plate. 21. Ando, M. Source mechanisms and tectonic significance of historical
Tectonophysics 248, 53–69 (1995). earthquakes along the Nankai Trough, Japan. Tectonophysics 27, 119–140
10. Kahn, A. A. Tectonics of the Bengal Basin. J. Himal. Geol. 2, 91–101 (1991). (1975).
11. Acharyya, S. Break-up of the greater Indo-Australian continent and accretion of 22. Chavez, M. & Castro, R. Attenuation of Modified Mercalli Intensity with distance
blocks framing south and east Asia. J. Geodyn. 26, 149–170 (1998). in Mexico. Bull. Seismol. Soc. Am. 78, 1875–1884 (1988).
12. Nielson, C., Chamot-Rooke, N., Rangin, C. & the ANDAMAN Cruise Team. the
Acknowledgements Among the many colleagues at Geoscience Australia who
ANDAMAN Cruise Team. From partial to full strain partitioning along the Indo-
offered support and advice, I especially thank C. Collins for discussions about the
Burmese hyper-oblique subduction. Mar. Geol. 209, 303–327 (2004).
1762 Arakan earthquake and Bengal during the colonial era, and M. Hollow for help
13. Gowd, T. N., Srirama Rao, S. V. & Gaur, V. K. Tectonic stress field in the Indian locating many of the historical references. I am also grateful to K. Sieh and C. Vigny
subcontinent. J. Geophys. Res. 97, 11879–11888 (1992). for their thorough and constructive reviews.
14. Socquet, A. et al. India and Sunda plates motion and deformation along their
boundary in Myanmar determined by GPS. J. Geophys. Res. 111, doi:10.1029/ Author Information Reprints and permissions information is available at
2005JB003877 (2006). www.nature.com/reprints. The authors declare no competing financial interests.
15. Wang, K. & He, J. Mechanics of low stress forearcs: Nankai and Cascadia. Correspondence and requests for materials should be addressed to P.R.C.
J. Geophys. Res. 104, 15191–15206 (1999). (phil.cummins@ga.gov.au).

78
©2007 Nature Publishing Group

You might also like