SPECIAL SECTION: SEISMOLOGY 2000

Using geological data for earthquake studies: A perspective from peninsular India
C. P. Rajendran
Centre for Earth Science Studies, Akkulam, Thiruvananthapuram 695 031, India

Geological methods are being increasingly used to document past seismicity and to characterize earthquake zones. Advances in palaeoseismology, for example, have resulted in the development of various spatial and temporal models that are fundamental to the understanding of earthquake recurrence. Although palaeoseismological techniques are being increasingly used to develop recurrence models for interplate and active intraplate regions, questions do remain on their applicability to the Stable Continental Regions (SCRs). Recent palaeoseismological studies indicate that seismic sources in the SCRs are characterized by very short (hundreds of years) to long (tens of thousands of years) recurrence periods. Whether a seismic zone is rift related or not appears to be one major factor that determines the recurrence pattern. We propose two major groups of SCR earthquakes based on their temporal characteristics.

Introduction
THE application of geological methods provides useful constraints to understand the long-term behaviour of seismogenic faults, and it marks an important deviation from the traditional approaches in seismic hazard assessment. It has been realized that an assessment based on conventional methods may not be complete or sufficient for understanding the long-term seismic activity. Even in countries with longer historical records, useful information on the seismic activities may not exist at all. Palaeoseismologic and geologic data provide useful inputs for recurrence models in different tectonic environments1. In regions characterized by lower strain rate, geologic records as old as Pleistocene (1.6 Ma) are now being used to understand the seismic behaviour of structures (Figure 1). Broadly speaking, the geologic studies aim to determine frequency of seismic events, interseismic intervals and timing of the last event2. Figure 2 gives an illustration of how geological inputs are used for seismic hazard assessment. Over the last couple of decades, this branch of investigation, known as palaeoseismology or earth-

quake geology, has contributed much to the understanding of earthquake processes37. Although palaeoseismology has been effectively used in interplate regions8, its application to stable continental regions (SCRs) has been rather limited. Due to the low strain build-up and longer recurrence periods typical of SCR9, evidence of past earthquakes is poorly developed and preserved. These constraints notwithstanding, palaeoseismological methods have been used successfully to reveal seismic history of many SCR sites. Although the details may vary from one site to another, the available data are suggestive of some discriminatory characteristics of SCR earthquakes. Most importantly, we note that recurrence periods may vary, even within an SCR. We illustrate this point using recent observations from locations of four earthquakes in India: Kutch (1819), Killari (1993), Broach (1970) and Jabalpur (1997). These examples show that maximum magnitude, style of deformation and period of recurrence may vary from one site to another, depending upon their tectonic domain.

Palaeoseismology in SCR: Scope and limitations

Palaeoseismology defines a unique methodology for the interpretation of active tectonics by integrating inputs from various disciplines quaternary geology, geomorphology, structural geology, seismology, geodesy, geo-

PALAEOS EISMIC

e-mail: cp_r@vsnl.com CURRENT SCIENCE, VOL. 79, NO. 9, 10 NOVEMBER 2000

Figure 1. Strain rate versus duration of characteristic seismic cycles for different seismo-tectonic provinces and data input required to determine the earthquake recurrence (after Giardini1). 1251

SPECIAL SECTION: SEISMOLOGY 2000

physics, soil science and geochronology. Mainly, it uses surface and subsurface expressions of seismically-induced features to extract information on the time and size of previous faulting episodes. For a detailed treatment of the subject, the reader is referred to McCalpin5, Yeats et al.6 and Noller et al.7. In this paper, we confine our discussions to the scope of this new tool for assessing the seismic history in SCRs, using examples from India. The primary aim of palaeoseismic analyses is to estimate the size and timing of past events and if possible, to constrain the geometry of faulting, using direct as well as indirect evidence. Deformational features such as fault scarps, fissures and folds provide direct evidence from which details of past seismicity can be inferred. For example, the size of past earthquakes is often estimated based on their rupture length and coseismic offsets, as illustrated in the pioneering studies on the Wasatch fault, Basin and Range Province, United States10. Surface ruptures and other direct indicators of fault activity have been found to be rarely associated with SCR seismicity; only eleven such cases are known worldwide (ref. 11; Table 1). In the absence of fault-specific data, secondary effects or proxy indicators such as palaeoliquefaction features are used, to constrain the magnitude, timing and number of past earthquakes, as in the cases of Charleston, SC (southeast United States), New Madrid, MO (central United States) and Wabash Valley, southern Indiana and Illinois, USA1218. Studies on the Meers fault (southern Oklahoma, south central United States)19 and in Australian interiors11 have also been useful in obtaining fault-specific information. Compared to the faults in interplate and active intraplate regions that are associated with larger slip rates, those in stable continental interiors produce smaller slips, which may not be evident at the surface. For example, a moderate SCR earthquake that generates < 0.5 m vertical deformation may not generally leave any clear evidence of faulting at the surface. Further, in many compressional settings, faults tend to develop as splays or blind thrusts and not reach the surface20. The rupture that reaches the surface tend to develop complex geometries, as in the case of the 1989 rupture in Ungava, Quebec21 and Marryat Creek22 earthquakes. The rupture zone in Killari, defined by two opposite verging thrusts, trending somewhat obliquely to the NW-trending fault, is another example of complex surface rupture in SCR. A factor that is common to most SCR earthquakes is their failure through reverse or strike-slip faulting, suggestive of a compressive stress regime23. This style of deformation results in poor development of fault scarps, compared to

Figure 2. Diagram showing relationship between geological data and seismic hazard evaluation (modified after Schwartz and Coppersmith2).

Table 1. Location of earthquake Kutch, India Accra, Ghana Central Sudan, Africa

Data on historical SCR earthquakes that produced documented surface rupture (modified after Machette et al.11) Date 16 June 1819 22 June 1939 9 October 1966 14 October 1968 11 March 1970 6 June 1979 22 December 1983 30 March 1986 22 January 1988 25 December 1989 29 September 1993 Earthquake magnitude M 7.5 M 6.5 Mb 5.1 Ms 6.8; Mb 6.0 Ms 5.7; Mb 5.7 Ms 6.4; Mb 6.3 Ms 6.2; Mb 6.4 Ms 5.8 Mb 5.7 Ms 6.3, 6.4, 6.7; Mb 6.1, 6.1, 6.5 Ms 6.3; Mb 6.2 Ms 6.4; Mb 6.3 Dominant faulting style Thrust Sinistral? Sinistral Dextral > reverse Sinistral > reverse Dextral > reverse Dextral Reverse > dextral and sinistral Reverse Reverse > sinistral Reverse Rupture length (km) 90 917 6 37 3 28 9.4 13 32 10 ~3 Max. scarp height (m) 4.3 (coseismic) 0.46 0.00 3.5 < 0.4 < 1.4 0.13 ~ 0.9 1.8 1.8 0.8

Meckering, WA, Australia Calingiri, WA, Australia Cadoux, WA, Australia Guinea, Africa Marryat Creek, SA, Australia Tennant Creek, NT, Australia Ungava, Quebec, Canada Killari (Latur), India

1252

CURRENT SCIENCE, VOL. 79, NO. 9, 10 NOVEMBER 2000

SPECIAL SECTION: SEISMOLOGY 2000

normal faulting environments5. Another common problem in SCR is that erosional and weathering processes often outstrip the tectonic processes and seismically generated landforms are destroyed much faster than they are created. Fewer incidences of earthquakes and the subdued nature of deformation have often restricted the use of palaeoseismological methods to their full potential in SCRs. The limitations can be overcome at least partially by developing site-specific methods and documenting discriminatory characteristics of palaeoseismic indicators in SCR settings. As mentioned earlier, the major thrust of worldwide palaeoseismological investigations has been to understand the temporal and spatial behaviour of seismically active faults. One of the basic assumptions on the temporal behaviour of faults is the linkage between the time intervals between major events and the amount of strain released during each event. These assumptions, coupled with field observations, have led to various recurrence models, providing the conceptual framework to interpret the fault behaviour. Of these, the time and slip predictable models24 and the characteristic earthquake model10,25 have been the most influential. However, some of these models have been questioned by subsequent workers26 and these assumptions are subject to revision as gaps in data are being filled. Recurrence models are developed mostly on the basis of data on interplate and active intraplate earthquakes and due to the sparse database, such models are not easily applied to SCR faults. Consequently, the longterm behaviour of seismic sources with respect to their spatial and temporal patterns is a poorly understood aspect of SCR seismicity. rence intervals and magnitudes among and within various SCRs. Some sites are known to be associated with long recurrence intervals, in the range of tens of thousands of years, as those in the Australian shield, for example27. With their recurrence intervals far exceeding the recorded history, these earthquakes appear to have occurred unexpectedly when viewed from the point of human life span. However, there are other seismic zones such as the New Madrid, MO (central US) and Charleston, SC (southeastern US) where M > 7 events are known to have recurred within intervals of hundreds of years28,29. Are there any geologic controls that make the former set of earthquakes different from the latter? The available data seem to suggest that it may be possible to group the SCR source zones on the basis of their relationship to their tectonic environments. The spatial association of SCR earthquakes with extended continental crust was first documented by Johnston9. He reported that about 60% of M > 6.0 and about 50% of M > 5.0 earthquakes are associated with extended crust. This association is stronger in the case of larger earthquakes (e.g. New Madrid and Charleston). The relationship may not be applicable to moderate earthquakes that occurred in the mid-cratonic regions which seem to occur

Earthquake recurrence in SCRs

Models for earthquake recurrence can be tested only when reliable data on event chronology and fault displacement are available for several earthquake cycles. The available data, however, seem to suggest drastic variations in recur-

Table 2.

List of significant recent (instrumentally recorded) earthquakes* in SCR India Date 210756 270367 101267 130469 241069 230370 300993 220597 Magnitude 6.0 (Mw) 5.4 6.3 5.7 (Mw) 5.3 5.4 (Mw) 6.1(Mw) 5.8 (Mw) Reference 30 31 32 33 31 33 34 35 Tectonic domain Rift Rift Non-rift Rift Rift (?) Rift Non-rift Rift

Location Anjar Ongole Koyna Bhadrachalam Mt. Abu Broach Killari Jabalpur

Ahmed abad

*1819 Kutch (Mw 7.5) is the most prominent historic (rift-related) earthquake in SCR India. A few other notable historic events like 1900 Coimbatore (M 6) related to unrifted SCR crust are also not listed in the table. CURRENT SCIENCE, VOL. 79, NO. 9, 10 NOVEMBER 2000

Figure 3. a, Seismicity of peninsular India from historical (AD 12th century) to recent times. Circles indicating magnitudes are scaled to size; earthquakes at Koyna and Killari are of magnitude > 6.0 and those at Bhadrachalam are > 5.0 and so on. Earthquake locations discussed in this paper are shown by filled circles (modified from Rajendran and Rajendran36); b, Study area Kutch and Broach; c, Study area Killari. 1253

SPECIAL SECTION: SEISMOLOGY 2000

by the reactivation of Precambrian faults. This basic difference between the tectonic regimes provides the starting point for classification of SCR earthquakes in terms of their recurrence. 1997 have generated new data on the scarp morphology and historical seismicity. Analyses of the data and dating of samples are in progress and here we provide a summary of salient observations. One of the important observations in Kutch concerns the nature of recurrence. Trenching excavations in the area revealed multiple liquefaction events, which crosscut the horizons of human settlement (Figure 5). Trenches reveal liquefaction features generated by the 1819 earthquake and those by a previous event. Based on the radiocarbon age data, we estimate that the penultimate earthquake occurred between cal. AD 8851035 (refs 40 and 41). In addition to obtaining age data on the penultimate event and exposing several potential sites for excavation, the Kutch earthquake provides details on the style of deformation. It is generally believed that the coseismic uplift evidenced by Allah Bund is the expression of a fault, the block south of the bund moving down with respect to the northern block38. Our observations in the shallow trenches across this structure, however, suggest that this scarp is not the trace of a fault, instead it may represent a fold. Lack of visible offsets in trenches across the bund is considered as evidence against the bund being a fault plane. Morphological features such as the progressive migration of streams and formation of a lake at the down dip side of the upthrust block are considered as indications of large-scale folding40,41. Current stress regime in the Kutch basin is predominantly compressive and because of the stress reversal, the Mesozoic normal faults are now reactivated as reverse faults40. The sense of movement on these reverse faults indicates that the hanging wall occurs to the north. These field observations together with the evidence for folding seem to imply that the far field effects of the Himalayan collision may be playing a role in the reactivation of faults in the Kutch rift. In the maps of SCRs by Johnston9, Kutch region falls in the northern fringe of SCR India. This fact,

Earthquakes in Indian SCR

During the last 50 years, eight moderate size earthquakes have occurred in the Indian SCR, a majority of them spatially coincident with palaeorifts (Table 2; refs 3035). For example, the Jabalpur earthquake was associated with the Narmada rift, expressed well by tectonic signatures as well as moderate seismicity. The Rann of Kutch earthquake was associated with the E-W trending boundary fault of the Mesozoic rift in northwest India. Although the data are quite limited at this stage, we have attempted a classification of these earthquakes, based on the tectonic environment, recurrence period (RP), maximum magnitude (Mmax) and style of deformation. By and large, we believe that the earthquakes associated with rift systems tend to recur at relatively shorter intervals (hundreds of years) in comparison to non-rift earthquakes. Further, the Mmax for the rift-related earthquakes seems to be larger than those in the non-rift environments. Thus, the spatial association with rift or the lack of it, is an important criterion in our classification. We illustrate these points by using the examples from the Rann of Kutch, Broach, Jabalpur and Killari earthquakes in the SCR India (ref. 36, see Figure 3 for locations). The Kutch earthquake: The 1819 Rann of Kutch earthquake (M 7.5) in north-west India (Figure 3 a) has been considered as one of the classic examples of earthquakes in extended crust37, ranking second in size to the 1811 1812 New Madrid earthquakes. The general level of seismicity of Kutch is relatively low for a region that has generated a large earthquake and no significant events have been reported since 1819. The Kutch earthquake generated a 90-km-long scarp known as the Allah Bund or the wall of God. With a maximum coseismic uplift of about 4.3 m and width of about 1520 km, significant features of this scarp are still preserved, although erosion has cut several transverse gullies (Figure 4). Coseismic changes associated with this earthquake are reported to have completely destroyed several fortified townships and permanently altered the local drainage system38,39. Rich documentation of the coseismic effects, reliable historic seismicity data together with well-preserved expressions of ground deformation and a large liquefaction field make this a unique site for palaeoseismologic studies. Political sensitivity due to its proximity to the international border with Pakistan, remoteness and inclement weather conditions have refrained many workers from taking up detailed investigations in this area. Our investigations in this area since
1254

Figure 4. Allah Bund scarp, viewed from the floor of an entrenched valley, possibly incised by the presently defunct Nara (Puran) River. CURRENT SCIENCE, VOL. 79, NO. 9, 10 NOVEMBER 2000

SPECIAL SECTION: SEISMOLOGY 2000

together with the relative closeness to an interplate boundary (in comparison to other SCRs) makes the Kutch seismic zone a unique example of SCR seismicity40. If topographic warping and consequent changes are the only surface expressions at Kutch, how does it compare with other rift/extended margin-related SCR sites, for example, the New Madrid (18111812) and the Charleston (1886) earthquakes? Contrary to the Kutch earthquake, the New Madrid events did not produce any spectacular topographic expressions, apart from localized arching and warping42. Surface expressions of faulting are not evident also at Charleston, although a recent work has reported geomorphological evidence for the existence of a major fault43. Our experience at Kutch suggests that the role of blind thrust in the seismogenic processes in the SCR needs to be studied in detail. Palaeoseismological investigations suggest an average recurrence interval of about 500 years for New Madrid28 and about 500 to 600 years for Charleston29. Although we do not have data for many cycles, the penultimate event at Kutch seems to have occurred about 1000 yr BP. Does this imply that these M > 7 earthquakes are comparable in terms of recurrence periods? It is worth examining whether such a relation can be developed, as we gather more data on these and other rift-related earthquakes. Broach and Jabalpur earthquakes: The Mw 5.4 Broach (1970) and the Mw 5.8 Jabalpur (1997) earthquakes occurred in the Narmada rift, a well-defined continentalscale structure that dissects the Indian subcontinent along the 22 parallel (Figure 3 b). Higher level of background seismicity, occasional moderate events since historic times and an explicit association with a continental-scale fault qualify these events as unique cases of SCR seismicity44,45. Earthquakes of M > 5 have occurred near Jabalpur in 1927, 1957 and 1997, i.e. at an interval of 35 5 years44. The only other site in Indian SCR where moderate earthquakes have recurred during an interval of tens of years is Koyna, believed to be triggered by the reservoir (see papers by Talwani; Mandal et al. in this issue). A notable aspect of the Narmada-rift seismic zone is the lower crustal source, as in the case of the Jabalpur earthquake, which occurred at a depth of 3540 km35. Moderate earthquakes (M > 5) at similar depths have been reported from the Amazonas, an ancient rift system in SCR of South America46. Large uncooled plutonic bodies (rift pillows) at lower crustal regions are likely to localize stresses around them leading to such deep crustal earthquakes46. Rajendran and Rajendran44,45 suggested that a similar mechanism may explain the spatial clustering and possibly, the shorter recurrence (tens of years) of moderate earthquakes observed in parts of the Naramada rift. No surface rupture was produced during the Jabalpur earthquake, presumably due to the deeper source. Another moderate earthquake occurred at Broach47, in the western extremity of the Narmada rift (Figure 3). Due to the hypocentral depth of 11 km (ref. 33) and the presence of the Narmada alluvium, we expect this region to have higher potential for seismically-induced liquefaction. Preliminary investigations along the eastern bank of the Narmada river at village Mandva, near Broach, revealed widespread liquefaction features, confirming our assumption. An exploratory trench at the village Mandva (Figure 3 b) revealed possibility of multiple pulses of liquefaction. Sand vents breaking through sedimentary sequences, traceable to lower horizons were observed. We suspect them to be of seismic origin, particularly due to their occurrence in the epicentral region. The intervening layers of sediments between the liquefaction features are not very thick (~ 15 cm), indirectly suggesting that the interseismic intervals cannot be very long. Radiocarbon dating of samples and processing of other data are in progress; at this stage these inferences are tentative. Based on the available data, we make two preliminary inferences. One, that moderate and shallower earthquakes in the Narmada rift are likely to generate liquefaction features. Two, their recurrence periods may be in the range of 100200 years. There are some weak spots within the rift (like Jabalpur), where moderate earthquakes may recur during smaller intervals of time. Broach will be an excellent site to do further excavations to develop a proper chronology of past seismicity, possibly representing several earthquake cycles. If evidence for possible moderate earthquakes can be obtained from this region, it will make a good case for studying the nature of earthquake recurrence associated with intracontinental rift systems.
1255

Figure 5. Multiple generation of liquefaction features from a trench excavated within the Vigakot Fort, near the epicentre of the 1819 shock. Note the vertical sand dike on the right hand side of the trench, cutting across the earlier generation of liquefaction feature, horizontal layers of white structureless sand called sill (thickness: 30 cm), which contains clasts of the host sediments. First generation liquefaction features cross-cut, therefore post-date an older occupation level (cal. AD 885 1035), identified on the basis of embedded fragments of pottery, bricks and bones. Base of the fort (brick layers) is seen in the upper level of the trench, affected by the second cycle of liquefaction, generated during the 1819 event. CURRENT SCIENCE, VOL. 79, NO. 9, 10 NOVEMBER 2000

SPECIAL SECTION: SEISMOLOGY 2000

The Killari earthquake: The Mw 6.1 Killari earthquake in the eastern margin of the Deccan plateau (Figure 3) shares many characteristics of mid-cratonic SCR seismicity, including subdued neotectonic expressions and very low background seismicity. This earthquake which nucleated at a shallow depth (5 km), was associated with hundreds of aftershocks. The rupture area was fairly small (38.5 sq km), involving only a fault length of 5.5 km. Choosing sites for trenching would have been difficult in this flat and featureless basalt country, but for the occurrence of a surface rupture that was recognizable for about 3 km near Killari village. Trenching excavations across the rupture zone not only exposed the fault plane but also revealed a displacement of near surface layers through reverse faulting4850. The sense of movement observed in the trenches as well as the fault plane solutions of the main shock define a north-west striking fault, dipping 45SW. Deeper trenches exposed layers of basalt, thrust in similar fashion, suggesting repetition of movements. A wide impact zone comprising minute fragments of rocks and grounded mass with yellowish clay was observed between the fault blocks (Figure 6). We interpret this as an indication of repeated faulting. Evidences for thrusting were observed in a few rock quarries and open wells49,50. With their sense of movement conformable with the 1993 faulting, it is reasonable to assume that faulting episodes, probably separated by thousands of years have occurred in this zone. Lack of datable material is a serious impediment in constraining the ages of previous events. Preliminary observations based on TL dating of the fault breccia from the deeper trenches provided an age of > 20,000 years50. Deep drilling at the Killari fault zone indicates about 6 m of displacement in the deeper basalt layers51. Assuming an average of about 1 m slip for a moderate event, as revealed by the 1993 shock, accumulated slip of 6 m at the deeper level indicates at least six moderate earthquakes at Killari. On a first approximation, this should suggest occurrence of at least six events (assuming constant slip) subsequent to the last episode of basalt flow. We also observed expressions of old thrusting events in lateritized profile overlying the lava flows in some well sections. Based on this premise, we believe that the recurrence period may be several tens of thousands of years, in contrast to the Kutch event. The long recurrence intervals of the order of tens of thousands of years or more for the Killari earthquakes are consistent with the recurrence intervals observed for the surface rupturing midcratonic earthquakes elsewhere22,46. Evidence for the only datable event in the vicinity of Killari was obtained from Ter, about 40 km NW of Killari, where an ancient earthquake seems to have occurred ~ 1500 years ago52. Can this be considered as an indication of regional clustering of events or are they just random events, with no spatial or temporal consistency? With the scanty data from this region, it is difficult to suggest any pattern.

Discussion
A majority of earthquakes in the SCR India appears to be spatially clustered around the pre-existing structural features. Studies suggest that these seismogenic zones are characterized by varying repeat periods and deformation styles. The size of the earthquakes generated by these faults also differs. Thus it is not reasonable to treat all the SCR earthquakes as belonging to one class. Based on its association with an ancient rift and shorter recurrence interval (in comparison to Killari type), we classify Kutch earthquake as Type I. Other rift-related earthquakes, which are smaller in size, are grouped as a subclass in Type I. The Killari type earthquakes that recur over much longer intervals of time are grouped as Type II. At this point, the classification is based only on a few examples, but this can be refined, as more cases are available. The seismogenic fault at Kutch, probably a throughgoing boundary fault of the rift system, appears to be capable of generating larger earthquakes (M > 7). The fault itself is not exposed at the surface, but is buried under a thick sediment cover. Age data from the seismically-induced features of the seismic zone indicate a moderate recurrence period of about 1000 years. Deformation at Kutch appears to be higher than what is observed at Killari or Broach and this is one aspect that can be resolved by GPS measurements (see paper by Bilham and Gaur, in this issue). What are the factors that could lead to a higher deformation and consequently, faster recurrence? Relative proximity of the collision boundary may be considered as a possible cause of higher deformation. Detailed geological, geophysical and geodetic studies have to be conducted for a proper assessment of the seismic hazard in this region. Earthquakes in the Narmada rift seem to recur at much shorter intervals, but they are smaller in size compared to
CURRENT SCIENCE, VOL. 79, NO. 9, 10 NOVEMBER 2000

Figure 6. Log of eastern trench wall, excavated across the rupture zone at Talni, near the epicentre of the 1993 Killari (Latur) earthquake. 1256

SPECIAL SECTION: SEISMOLOGY 2000

the Kutch earthquake. While the recurrence interval of Jabalpur or Broach type of earthquakes (M 56) may be 100200 years, smaller earthquakes (M < 5) recur at shorter recurrence intervals44. In comparing the Narmada and Kutch earthquakes, which are related to rift systems, it is important to relate recurrence period (RP) with the maximum magnitude (Mmax). Although the RP in the Narmada rift is of the order of 100 to 200 years, the earthquakes are smaller and the Mmax is only about 6. We do not have good quality data to calculate the RP and Mmax based on b-value estimates. However, it should be noted that the shorter recurrence interval of moderate earthquakes in the Narmada rift and its spatial and temporal clustering are not commonly observed in SCRs. In particular, the regional seismicity associated with the Narmada rift appears to be higher compared to the rest of SCR India and other SCRs. Although these are rift-related and should be placed under the same category as Kutch, due to their shorter return periods and lower range of Mmax, we group them as a subclass of Type I. In contrast to the Jabalpur type of earthquakes, the Killari event occurred in a region, which is not associated with significant historic or recent seismicity. Geologic studies suggest that the region around Killari has been subjected to previous episodes of deformation, separated by intervals of the order of tens of thousands of years. Recurrence interval of Killari type of events is comparable to other moderate mid-cratonic earthquakes, e.g. the Cadoux (Ms 5.1), Marryat Creek (Ms 5.8) and Tennant Creek (Ms 6.3, 6.4 and 6.7) earthquakes in the Australian interiors and Ungava (Ms 6.3), Quebec. Recurrence interval in the range of tens of thousands of years, poor surface expressions of faulting and low background seismicity make these events rather typical of mid-cratonic seismicity, which are grouped under Type II events. The Killari event is an example of a typical midcratonic earthquake (unrifted SCR crust). The region may not have undergone any major rifting since the Precambrian (~ 570 Ma). The low strain rate is evident from the subdued topography, long interval between the earthquakes and almost nonexistent background seismicity. Results from other parts of the world, sharing similar tectonic characteristics suggest that the recurrence time between surface rupturing events may be tens of thousands of years or even hundreds of thousands of years at any single source53. We prefer to call these as passive seismic sources because they are located far from influences of plate boundaries. Seismic cycle may be poorly defined for such regions, particularly because the age of the penultimate earthquake is not known. We believe that potentially hazardous faults, some of which are hidden, may exist in the cratonic hinterlands and they need to be recognized on the basis of geological evidence. Developing recognition criteria for such faults and assessing their seismogenic potential will be crucial for refining seismic hazard and risk estimates in cratonic regions.
CURRENT SCIENCE, VOL. 79, NO. 9, 10 NOVEMBER 2000

Dual classification proposed for the Indian SCR is based on well-known inland seismic source zones. We may expand or modify this classification, as more data are available from other types of tectonic settings. For example, Johnston54 identifies three types of the eastern North American SCRs, namely (1) continental oceanic boundary (COB), (2) unrifted SCR crust and (3) rifted SCR crust. We do not have any reliable historical Indian examples of COB earthquakes or any information on their recurrence period. From the seismic hazard point of view, an important observation from the Indian SCR is that most earthquakes are temporally and/or spatially associated with certain pre-existing weak zones. All the source zones that we investigated suggest indications of previous seismic activity. An outstanding problem is to understand the mechanisms that localize seismic activity. As a first step towards resolving this problem, spatial and temporal patterns of earthquakes in SCRs need to be understood in greater detail. Palaeoseismological studies were initiated in India in the early nineties and many potential regions remain to be examined. The challenge before the scientific community is to expand the scope of these studies further and use the data to identify probable future nucleation points. This insight can be gained by the clever use of palaeoseismic information in tandem with data generated from advanced imaging techniques of fault zones and high precision GPS studies.

1. Giardini, D., in Proc. International School of Solid Earth Geophysics, Erice-Sicily, Italy, 1995, pp. 114. 2. Schwartz, D. P. and Coppersmith, K. J., in Active Tectonics, National Academy Press, Washington DC, 1986, pp. 215 230. 3. Active Tectonics, National Academy Press, Washington DC, 1986, p. 266. 4. Crone, A. J. and Omdahl, E. M. (eds), US Geological Survey Open-file Report 87673, 1987, p. 456. 5. McCalpin, J. P. (ed.), Palaeoseismology, Academic Press, New York, 1996, p. 588. 6. Yeats, R. S., Sieh, K. and Allen, C. R., The Geology of Earthquakes, Oxford University Press, New York, 1997, p. 569. 7. Noller, J. S., Sowers, J. M. and Lettis, W. R. (eds), Quaternary Geochronology Methods and Applications, Am. Geophys. Union, 1999, p. 581. 8. Sieh, K., in Proc. Natl. Acad. Sci. USA, 1996, 93, 37643771. 9. Johnston, A. C., in Earthquakes at North Atlantic Passive Margin: Neotectonics and Post-glacial Rebound (eds Gregersen, S. and Basham, P. W.), Kluwer Academic Publishers, 1989, pp. 299 327. 10. Schwartz, D. P. and Coppersmith, K. J., J. Geophys. Res., 1984, 89, 56815968. 11. Machette, M. N., Crone, A. J., De Martini, P. M. and Okumura, K., in Proc. Hokudan Int. Symp. and School on Active Faulting, Japan, 2000, pp. 243247. 12. Talwani, P. and Cox, J., Science, 1985, 229, 379381. 13. Obermeier, S. F, Gohn, G. S., Weems, R. E., Gelinas, R. L. and Rubin, M., Science, 1985, 227, 408411. 14. Obermeier, S. F., US Geological Survey Open-File Report 94663, Chapter A, 1995, pp. 158. 1257

SPECIAL SECTION: SEISMOLOGY 2000

15. Talwani, P., Amick, D. C. and Scheffer, W. T., NUREG/CR-6619, 1999, p. 109. 16. Obermeier, S. F., Eng. Geol., 50, 227254. 17. Tuttle, M. P., Lafferty, R. H., Guccione, M. J., Schweig, E. S., Lopinot, N., Cande, R. F., Dyer Williams, K. and Haynes, M., Geoarchaeology, 11, 451480. 18. Kelson, K. I., VanArsdale, R. B., Simpson, G. D. and Lettis, W. R., Seismol. Res. Lett., 1992, 63, 349356. 19. Crone, A. J. and Luza, K. V., Bull. Geol. Soc. Am., 1990, 102, 117. 20. Wheeler, R. and Johnston, A. C., Seismol. Res. Lett., 1992, 63, 491505. 21. Adams, J., Wetmiller, R. J., Hasgava, H. S. and Drysdale, J., Nature, 1991, 3562, 617619. 22. Machette, M. N., Crone, A. J. and Bowman, J. R., US Geol. Surv. Bull., 1993, 2032-B, 29. 23. Talwani, P. and Rajendran, K., Tectonophysics, 1991, 186, 1941. 24. Shimazaki, K. and Nakata, T., Geophys. Res. Lett., 1980, 7, 279 282. 25. Wesnousky, S. G., Bull. Seismol. Soc. Am., 1994., 84, 19401959. 26. Kagan, Y. Y., Bull. Seismol. Soc. Am., 1996, 86, 274285. 27. Crone, A. J., Machette, M. N. and Bowman, J. R., Aust. J. Earth Sci., 1997, 44, 203214. 28. Tuttle, M. P. and Schweig, E. S., Geology, 1995, 23, 253256. 29. Amick, D. and Gelinas, R., Science, 1991, 251, 655658. 30. Chung, W. Y. and Gao, H., Tectonophysics, 1995, 242, 281292. 31. Chandra, U., Bull. Seismol. Soc. Am., 1977, 67, 13871413. 32. Gupta, H. K. and Rastogi, B. K., Dams and Earthquakes, Elsevier, Amsterdam, 1976, p. 229. 33. Chung, W. Y., Tectonophysics, 1993, 225, 219230. 34. Gupta, H. K., Rastogi, B. K., Indra Mohan, Rao, C. V. R. K., Sarma, S. V. S. and Rao, R. U. M., Tectonophysics, 1998, 287, 299318. 35. Singh, S. K., Dattatrayam, R. S., Shapiro, N. M., Mandal, P., Pacheco, J. F. and Midha, R. K., Bull. Seismol. Soc. Am., 2000, 89, 16311641. 36. Rajendran, K. and Rajendran, C. P., in Proc. Second Int. Conf. Harbour and Ocean Eng., 1997, pp. 305314. 37. Johnston, A. C. and Kanter, L. R., Sci. Am., 1990, 262, 6875. 38. Bilham, R., in Geol. Soc. London, Spl Publ., 1998, 146, 295319. 39. Rajendran, C. P., Rajendran, K. and John, B., Curr. Sci., 1998, 75, 623626. 40. Rajendran, C. P. and Rajendran, K., Bull. Seismol. Soc. Am., 2000 (in review). 41. Rajendran, C. P. and Rajendran, K., in Proc. Hokudan Int. Symp. and School on Active Faulting, Japan, 2000, pp. 387391. 42. Johnston, A. C. and Schweig, E. S., Annu. Rev. Earth Planet. Sci., 1996, 24, 339384. 43. Marple, R. T. and Talwani, P., Geol. Soc. Am. Bull., 2000, 112, 200220. 44. Rajendran, K. and Rajendran, C. P., Curr. Sci., 1998, 74, 168 174. 45. Rajendran, K. and Rajendran, C. P., Tectonophysics, 1999, 305, 335370. 46. Zoback, M. L. and Richardson, R. M., J. Geophys. Res., 1996, 101, 54595475. 47. Gupta, H. K., Indra Mohan and Narain, H., Bull. Seismol. Soc. Am., 1972, 62, 4761. 48. Seeber, L., Ekstrom, G., Jain, S. K., Murthy, C. V. R., Chandak, N. and Armbruster, J. G., J. Geophys. Res., 1996, 101, 8543 8560. 49. Rajendran, C. P., Rajendran, K. and John, B., Geology, 1996, 24, 651654. 50. Rajendran, C. P. and Rajendran, K., Tectonophysics, 1999, 308, 6781. 51. Gupta, H. K., Rao, R. U. M., Srinivasan, R., Rao, G. V., Reddy, G. K., Dwivedy, K. K., Banerjee, D. C., Mohanty, R. and Satyasaradhi, Y. R., Geophys. Res. Lett., 1999, 26, 19851988. 52. Rajendran, C. P., Curr. Sci., 1997, 72, 750755. 53. Crone, A. J., Machette, M. N. and Bowman, J. R., US Geol. Surv. Bull., 1992, 2032-A, p. 51. 54. Johnston, A. C., in Proc. 7th Canadian Conf. Earthquake Eng., Montreal, Quebec, 1995, pp. 8187.

ACKNOWLEDGEMENTS. I thank the Department of Science and Technology, Govt of India for funding. The manuscript benefited from the comments from Pradeep Talwani and Kusala Rajendran.

1258

CURRENT SCIENCE, VOL. 79, NO. 9, 10 NOVEMBER 2000