Leonardo Seeber
Lamont-Doherty Geological Observatory;
Introduction
Most of the world seismicity is concentrated along the plate boundaries.
However, a significant number of earthquakes, including some large and damaging
ones, do occur within the plates. Our understanding of intracratonic
seismogenesis and the hazard it entails is poor, in
part because data are scarce.
The 1993 Killari earthquake in central peninsular
This report summarizes our
observations during a ten-day investigation of the mesoseismal
area of the 1993 Killari earthquake. Several aspects
of the earthquake were investigated, ranging from the surface rupture and
related deformation to the pattern of damage to engineered and traditional
structures. Finally, the rescue and reconstruction efforts following the
earthquake brought out important issues that are generally relevant to
earthquake hazard reduction in traditional rural settings; these are also
briefly discussed.
Key Information
Location
Centered near the
Main Shock Source Characteristics (USGS):
00:03:53 local time,
Main Shock Effects
Widespread death and
destruction in the districts of Latur and Osmanabad, Maharashtra state;
complete destruction of stone /mud structures in about 20 villages covering an
area about 15km wide centered 5 kms west of Killari. death toll about 10000;
highest relative death toll 35% in the villages of Killari
and Chincholi-Rebe; mesoseismal
area intensity VII-IX.
Precursory Seismicity
A very active swarm between august and October
1992 (125 felt events in Killari); the largest event
on
Geologic setting
Located in the ½ km-high plateau of central
peninsular India, within the area covered by the Deccan
traps, a sequence of late cretaceous basalt flows that cover the
central-western portion of the craton; the basement
below the trap rock, where the earthquake probably nucleated, is thought to
consist of high grade metamorphic rocks of early Precambrian age.
Intracratonic Seismogenesis in India
Most of the well-known large and
destructive earthquakes of the Indian subcontinent are related to the Himalayan
and Baluchistan arcs, the topographical expression of
a collosional plate boundary that marks the northern
limit of the shield. Yet, intracratonic seismicity in India is well represented, both in number and
in size (Figure 1) The 1819 Rann of Kuch earthquake (RK in Figure 1) is associated with
a surface rupture about 100 km long and a scarp up to 10m high. It permanently
altered the shape of the land surface over a very large area and was felt
throughout northern
Figure 1: Seismicity (ISC
catalog, 1901-1992; epicenters indicated by a circle and
last two digits of the year; epicenters along the Himalayan and Baluchistan
arcs omitted to avoid clutter) and earthquake hazard zonation
in five
categories (I to V is lowest to highest, respectively, Geological Survey
of
India, 1992). The 1993 Killari epicentral
area is indicated by a box; the
two epicenters within this area are the largest representatives of the
very
active precursory swarm felt in Killari in 1992.
Large and destructive
earthquakes near
plotted (years indicated in full). 1967 is the year of the largest
earthquake
from the very active reservoir-induced source zone near Koyna.
RK =
1819 Rann of Kuch
earthquake; NFZ = Narabada fault zone; GG =
Historic seismicity suggests that casualties and damage from
earthquakes with magnitude of 5.5 or grater occur relatively often in cratonic
Surface Rupture
The 1993 Killari earthquake ruptured
the surface. Three weeks after the main shock, at the time of our field
investigation, discontinuous scarps were discernible along a west-northwest
zone about 1km long and starting 1.5km west of the
Figure 2: Portion
of the surface rupture associated with the 1993 Killari
earthquake.
This rupture was mapped discontinuously in an east-west zone over a
distance of about 1 km. Barbs indicate the uplifted side of the scarps;
the
approximate height of the scarps is also indicated. Plus signs indicate
uplifted area, circled when localized. Note that a 'transform' fault
connects
a south-facing scarp in the center of the map with a north facing scarp
on
the right. This 'tranform' offsets a field boundary
by 53 cm left laterally,
which is a minimum measure of the shortening. The strike of this and
other
'transforms' in this map is north-northeast, the inferred direction of
shortening.
Results associated with two
north-merging scarps and a south-verging scarp were
exposed in three 1-1/2 m deep hand-dug trenches. Two distinct modes of faulting
could be identified in the weathered basalt exposed in the trenches. In the
regions where the fault could not develop along pre-existing fractures, shear
was accomplished in a zone, typically 10-20 cm wide, where pre-existing and
newly splintered fragments were rotated. These zones had lost cohesion (and
probably density as well) and were easily identifiable on the trench walls. The
evidence for motion was much more subtle where the fault followed pre-existing
fractures. In these cases the movement was often confined to a very small thickness,
possibly a single fracture.
Figure 3 shows fatalities
relative to the total population in each village. These gruesome statistics are
quite effective in delineating the mesoseismical
area. In an area about 10 x 10 centered on the
Figure 3: Number of
officially recorded fatalities as percentage of population in
villages in epicentral area. A surface rupture was mapped
near the
of
Surface deformation features reflect a reverse fault dipping to the
Southwest. Note that the zone of highest relative fatalities (20-30 %) is
centered Southwest of the trace of the thrust, above
the inferred rupture.
Note also that this rupture is centered about 10km from the
Dam.
A new Fault?
Is the 1993 Killari
earthquake rupture one of a sequence of ruptures that have
occurred on the same fault? Except for the precursory activity in 1992, the Killari area is not known to have had prior earthquakes
during the historic period, not there is evidence of Holocene ruptures or
accumulated Neocene deformation in the form of prehistoric scarps or a
topographic rise. No evidence of pre-1993 faulting of the late cretaceous
Induced Seismicity?
The Killari
earthquake was about 10 km from the
Response of Buildings
Engineered structures were relatively scarce in the affected area. A maximum
intensity level of MM VII-IX could be determined by the performance of the
few-constructed brick-and-mortar structures.
The collapse of traditional
stone-and-mud buildings in the mesoseismal area was
nearly total. The wood-plank roofs of these single-story dwellings typically
are topped with a 30-60 cm thick layer of clay to provide protection from rain
and heat. All such constructions behaved very poorly due to the heavy mass at
the roof and the poor strength of the supporting rubble masonry walls; such
houses were the main cause for the high number of casualties.
A number of dwellings in
the affected villages had timber columns connected together with transverse and
longitudinal beams. The roof planks in these houses were supported by the
timber beams and columns rather than the rubble masonry walls. When securely
anchored to the floor and to the roof beams, the posts tended to hold up the
roof and prevent the inward collapse of the walls, thereby saving the
inhabitants.
Some of the poorest people
in the villages lived in thatch-type houses consisting of wooden vertical posts
and rafters connected with coir rope ties. Roofs are thatch, and thatch panels
or a series of small stocks or slit bamboo woven together form the walls.
Mud-plaster provided a wall finish in some of these pictures. These houses
performed extremely well with only minor cracks in the mud-plaster walls.
Figure 4: In some
buildings, the traditional heavy clay roof was supported on a frame
of wood posts and beams. In most cases, the timber columns continued
to
hold the roof together while the walls collapsed, generally outwards
from
the timber frame, saving the lives of many occupants
Some houses were made of
load-bearing walls of burnt-clay bricks in cement mortar supporting either a
reinforced concrete slab or a corrugated sheet roof. These suffered varying
degrees of damage, but none collapsed. In a residential colony at Killari, a house supporting heavy reinforced concrete roof
sustained severe wall cracks, while the adjoining one supporting a light roof
of corrugated asbestos sheets sustained only hairline cracks in the walls.
Surprisingly, a few brick masonry houses in the area were found to have
concrete lintel bands. The Indian building code only recommends lintel bands
for seismic zones IV and V. Such houses performed very well with no damage.
A one-room 8x8 m school
building in the
Figure 5: A brick
masonry school building in Rajegaon collapsed. The
roof of
precast panels lacked adequate connections
Some of the factors that
contribute to the poor performance of the traditional structures are
unavoidable, given the materials available within the economic constraints.
Some patterns of failure, however, point to improvements that probably could be
made within these limitations. For example, stone walls tended to separate
along axial planes, with the two faces collapsing in opposite directions. Large
stones for cross-wall ties are difficult to procure from the native weathered
basalt, but suitable substitutes should be sought out. The heavy roofs used for
thermal insulation contributed heavily to destruction and casualties. Vertical
wooden posts that supported some roofs saved many lives. Because timber is a
scarce resource, some effective alternative should be developed. Ironically,
the poorest houses with thatched roofs and walls were often the only ones to
survive unscathed.
Response of Engineered
Structures
Figure 6: While
most elevated water tanks performed well, this tank of Kautha
collapsed straight down into its crumpled supports. Circumferential
displacement of about 0.5m suggests that rotational vibration led to its
collapse.
For water supply, most of the villages had an elevated water
tank consisting of a reinforced concrete container supported on a concrete
moment-resisting frame about 10 m high, a standard design of the public health
engineering department. Tank capacities ranged from 40 kl
to 200 kl. Most of these tanks survived the
earthquake with little or no damage. The one tank that suffered a complete
collapse came straight down, burying the remains of six supporting columns
directly under the bottom dome of the tank. That, and evidence of a
circumferential displacement of about 0.5 m, suggest torsional
vibrations were the primary cause of the damage.
Several structures in the
Several bridges sustained
minor damage. The bearings on one of the piers of a six-span overpass were
damaged when the pier moved away from the abutment by about 4 to 5 cm.
Emergency Response
The affected area does not suffer from floods and was
considered Aseismic. The only natural disaster known
in the area is drought. Hence, the earthquake took the people and
administration by surprise. It took the administration 2 to 4 days to
effectively organize rescue and relief operations. What then followed appeared
to be well-organized and effective. Civilian as well as military authorities
cooperated in the effort. Since the reconnaissance team did not visit the
affected areas of Karnataka, the discussion here is based on the response in
the districts of Latur and Osmanabad
only.
Of all the villages
devastated by the quake, only the
By the afternoon of the day
of the earthquake, many curious onlookers had arrived in the area. This led to
traffic jams and hampered the task of search and rescue. Beginning Oct. 2,
entry to the affected area was controlled to allow entrance only to government
personnel and members of volunteer organizations.
The task of search and
rescue became extremely difficult due to the heavy rains, which immediately
followed the earthquake and the enormous quantities of the rubble. In places
2-3 m of rubble has to be removed to extricate the bodies. The narrow village
streets were choked by fallen rubble, which further hampered rescue operations.
Mass cremations that were
held on the first two days after the earthquake were done without adequate
record keeping. This led to confusion about the actual number of deaths. At one
time, the newspapers were reporting up to 30000 dead. Later it was discovered
that about 9000 fatalities actually occurred.
Recovery
Medical Aid: The number of
injured in
Food, water, Provisions: Cooked food was provided in the affected
villages during the first fortnight by large number of voluntary agencies and
the army. After 15 days the community kitchens were closed. Enough rations and
other provisions to last a month were provided to the people, requiring them to
cook their own food.
Figure 7: After the
earthquake, water was trucked into the region. Generally, each
village was assigned a truck requisitioned from unaffected neighbouring
districts
Drinking water was supplied
regularly by mobile water carrying units, generally one for each village,
requisitioned from the neighbouring districts. Water
storage tanks (5000 l), donated by UNICEF/CARE, were
being used for storing water in villages. In some villages, these tanks were
lying upside down while quite a few were still resting in storage yards,
indicating excess availability of the tanks. A total of 227 water hand pumps in
the earthquake-affected areas of the Latur district
were found operative.
Simultaneously, a program
for installation of tube wells in the temporary settlements was initiated. In
the Latur district, 83 deep-bore wells at the
temporary shelter sites needed to be dug. 47 of these had been completed by Oct
19. For this task, eleven deep-bore rigs were mobilized from the other
districts of
Figure 8: Distribution
of relief supplies was systematized through the issuance of
'family cards' and 'individual cards' to residents of the affected areas
The government and several
voluntary agencies distributed clothes and household goods like utensils,
stoves, storage cans, and provisions for daily needs, either through the
administration or directly. A team consisting of an officer from the
administration, an officer from the police, and a local head of the village
issued every affected family a “family card” and every affected person an
“individual card” to record and regulate the goods distributed as a relief aid.
This, to a large extent, systematized the relief distribution. Spontaneous
offers of aid were received from within the country and outside. Bombay International Airport made special arrangements to receive
and clear planeloads of clothes, medicine, tents and food. Within about
3 weeks, the relief material available was somewhat in excess of the needs.
Figure 9: Temporary
shelters of galvanised iron sheet were rapidly
erected. Each
provides shelter for five families
Temporary Shelter: Immediate shelter was provided by erecting
temporary galvanized iron sheet houses. Each shelter consisted of 5 units of
about 15 m2 areas each; each affected family was offered one such
unit. Common toilets and hand operated bore wells were provided. Some families
chose to take tents and tin sheets instead and made their own shelters.
Resettlement and Housing: A large number of villages which were totally destroyed are
being relocated to places not far away. Most of the sites for relocation have
been identified.
Every affected family will
be provided a housing unit with a covered area of 20, 40 or 60 m2
(depending on the needs of the family) and an open area of 15 to 45 m2
for animal shelters and future extensions. The government will bear the cost of
only 20 m2 of the housing per family. For houses with a larger
constructed area, the remaining cost will be treated as a long-term
low-interest loan payable on terms similar to those for World Bank loans.
Transportation
The
primary mode of transport in the area is bus. The affected area has no rail
network. The nearest railway stations are at Latur
and
Figure 10: Solar
connect the larger villages to the outside world
Communications
An
excellent communication network was set up after the earthquake. Initially, a
number of HAM sets were called in. Later, solar powered satellite communication
towers were installed in most of the bigger villages. The district collectors
were provided with a hotline to the chief secretary at
Economic activity
The
economy of the area is primarily farm based with very few industries. The only
major industry in the region was a sugar factory at Killari
Pati, which had closed quite some time prior to the
earthquake. It suffered no apparent damage and was being used as a relief
distribution center.
A large chicken farm about
15km north of the epicentral area reported that egg
production fell substantially after the quake. Production was still down about
7% three weeks later. (No fluctuations in egg laying preceded the quake.)
Though it is harvesting
season for the Kharif crop and the sowing season the
Rabi crop, the farmers have been unable to resume regular farming activity.
This is due to incessant rains, unexpected at this time of year, and the
non-availability of farm laborers, who do not seem concerned about daily wages
due to the availability of free relief provisions. There are reports that the
government may have to help in harvesting by bringing in labor from outside the
area.
Earthquake Hazard Maps and the 1993 Killari Earthquake
Earthquake hazard maps for intracratonic areas are usually based on the distribution
of historic seismicity. Geologic features which are
thought to be preferentially associated with large earthquakes are also a
factor in modern maps because the relatively short historic period is likely to
give an incomplete picture of the possible earthquake sources. The hazard zonation in Figure 1 conforms to this state-of-art
approach and is based on both seismicity and geology.
For example, the earthquakes in 1843 and 1900 as well as the Narabada fault zone (NFZ) and the
Is there ground for optimism?
The Killari
earthquake offers new opportunities to improve the resolution of hazard maps
and the effectiveness of hazard-reduction measures. A current controversy in
Acknowledgements
The
financial support provided by the Earthquake Engineering Research Institute
(EERI),