1RE-EVALUATED INTENSITIES FORTHE GREAT ASSAM EARTHQUAKE OF 12 JUNE 1897, SHILLONG, INDIANicolas Ambraseys(1) and Roger Bilham(2)Abstract: The great Assam earthquake of 1897 (8<M<8.1) is the largest known Indianintraplate earthquake. It raised the northern edge of the Shillong Plateau by more than10 m, resulting in the destruction of structures over much of the Plateau andsurrounding areas, and causing widespread liquefaction and flooding in theBrahmaputra and Sylhet floodplains. Shaking intensity data for the earthquake arecrucial for estimating future earthquake hazards in NE India and Bangladesh sincesimilar earthquakes will no-doubt recur. Yet despite the availability of numerous feltreports, no evaluation of isoseismal contours has been attempted since Oldham's (1898)approximation. We have re-evaluated 365 accounts of the earthquake and quantified287 on a simplified version of the "MSK 1981" Intensity scale. The re-appraisedisoseismals are consistent with the geodetic mechanism for the earthquake and aresmaller, less regular, and less elliptical than those inferred by Oldham, and suggest thatOldham's intensities were unsuspectingly inflated by 1.5-3 intensity units. The revisedintensity data provide new quantitative constraints on the attenuation of perceivedintensity as a function of distance in northeastern India.Introduction The earthquake of 12 June 1897 in the Shillong Plateau in north-eastern India is thelargest intraplate event in the last two centuries to have occurred in the Indian subcontinent.Although contemporary geodetic data have been used recently to infer the rupture parametersof the earthquake (Bilham & England, 2001), much of what is know about the intensitydistribution of the event comes from Oldham's detailed report and his subsequent estimatesof aftershock activity (Oldham, 1899; 1901; 1904; 1920). Over the years his report andisoseismal maps have been used to assess the seismic hazard of this part of India, to evaluatethe magnitude of the event, and to constrain the location and dimensions of the associatedrupture zone (Figure 1). In view of the importance of the 1897 earthquake, a review of the intensities assigned byOldham is warranted, especially since we may now incorporate data from Tibet and frompublished data not utilized by Oldham. This paper is concerned solely with the reappraisal ofthe intensity distribution of the Assam earthquake, drawing attention to potential pitfallsassociated with indiscriminate use of intensity data. The earthquake The earthquake occurred beneath the Shillong Plateau in what was then the district ofAssam, in northeast India. For many years the earthquake was considered to have occurredon a gentle north-dipping rupture beneath the Shillong Plateau, linked structurally to thestyle of Himalayan thrust faulting (Seeber et al., 1981, Molnar 1987; Molnar and Pandey,1989, Gahalaut and Chander, 1992). This interpretation has recently shown to be untenable.According to recent analyses of geodetic data acquired in 1860, 1897 and 1936, theearthquake occurred on a steep SSE-dipping reverse fault beneath the northern edge of thecentral Shillong Plateau (Bilham and England, 2001). Rupture is inferred to have extendedfrom 9 km to more than 30 km depth on a 110-km-long, SSE steeply-dipping reverse fault2that slipped approximately 15 m. The Chedrang fault, a north-south secondary fault, rupturedthe surface at the western end of, and above the main rupture.LEFT Fig. 1a. Location map for the 1897 Assam earthquake showing isoseismal approximations assignedby Oldham (1899) as dashed incomplete curves. (Approximate radii are: r5 = 960 km; r4 = 704 km, r3 =544 km; r2 = 304 km; r1 = 160 km) Bold lines enclose Oldham's (a) felt area, (b) "area of extensivedamage" and (c) "the probable limits of the epicentre", beneath and north of the Shillong Plateau. Fig.1b.Locations of observational data (dots) used to evaluate MSK intensities.RIGHT Fig. 2a. Epicentral location map showing Oldham's (1899) epicentral area centered on theShillong Plateau, and dashed rectangle indicating the inferred 1897 subsurface rupture. Fig. 2b Same areawith evaluated MSK intensities, rejected observational data (x), and isoseismal contours evaluated usingkriging methods described in the text.The Assam earthquake holds a prominent place among the great earthquakes of the world,not only because of its large magnitude, but also because of the large area over which itcaused damage, liquefaction, and landslides. Destruction was widespread on the ShillongPlateau, (the present state of Meghalaya), and in surrounding areas (Figure 1). Much of whatis known about the event is exclusively due to Oldham's report, (Oldham 1899). The earthquake almost totally destroyed settlements and small towns on the westernpart of the Plateau, and caused heavy damage in surrounding districts, chiefly due to theextensive liquefaction of the ground. The Plateau, which is the only high ground between3the Himalaya and the Bay of Bengal, is about 250 km long and 80 km wide and includes,from west to east, the Garo, Khasi and Jainti hills. It stands more than 1,500 m above theplains of the Brahmaputra River which flows along the north and west sides of the Plateau.The Sylhet flood plains bound the Plateau's southern margin where much additional damageoccurred. The Plateau receives an average of 11.4 m of precipitation, the second highestrainfall in the world, much of it during the monsoon period.Instrumental The earthquake occurred at 17h 15m local time (11h 09m GMT) and wasrecorded by 12 primitive seismographs in Europe at distances between 64° and 72°(Germany, Italy, France and UK). Instrumental readings, however, are not good enough to beused to establish its position, even approximately, and the macroseismic location of the eventhas been estimated by various authors at 26°N, 91°E. The mean geodetic location is 25.7°N.91.1°E. Although the 1897 event occurred too early to be located instrumentally, its surface-wave magnitude MS can be estimated from the trace amplitudes recorded by six calibratedinstruments operating in Italy (Agamennone, 1897; Cancani, 1897), one in Russia (Kortazzi1900), and two in the UK (BAAS 1898). Using Abe's method we obtain a value of MS 8.0±0.15). This value is very close to the estimates of 8.2 made by Kanamori and Abe (1979),revised by Abe (1994)