Tsunami wave propagation in the Arabian Sea and its implications on run-up/inundation characteristics along the Kerala coast

Transcription

1 Indian Journal of Geo-Marine Sciences Vol. 39 (4), December 2010, pp Tsunami wave propagation in the Arabian Sea and its implications on run-up/inundation characteristics along the Kerala coast N P Kurian & S S Praveen Centre for Earth Science Studies, Thiruvananthapuram , India [ npkurian@gmail.com] Received 20 August 2010; revised 21 December 2010 December 2004 tsunami generated by the M9.3 Sumatra-Andaman earthquake devastated many parts of the Kerala coast of southwest India, though it was in the shadow zone of that tsunami. Post-tsunami field survey and sea level data provided a rare opportunity for the study of wave transformation characteristics along this coast. A review of the available studies on the effects of 2004 tsunami wave propagation like the observed focusing/defocusing, number and arrival times of waves along the Kerala coast is undertaken in this paper. The important processes identified to be responsible for the observed runup and inundation are refraction, diffraction, reflection, and total internal reflection. Water level variations from the third day onwards are attributed to energy trapping on the continental shelf and secondary undulations. In addition, the implications of the wave propagation characteristics on tsunami run-up and inundation along the Kerala coast for the two probable sources of tsunami viz. Sumatra and Makran are examined through numerical modeling. It is found that the central and northern sectors of the coast are more hazard prone to the Makran than Sumatra for the same rupture intensity. [Keywords: Tsunami, Run-up, Inundation, Kerala coast, Wave transformation] Introduction The December 2004 tsunami generated by the M9.3 Sumatra-Andaman earthquake devastated many parts of the Kerala coast. The Kerala coast (Fig. 1) is located in the shadow zone with respect to the direction of propagation of the tsunami, and in that sense its severity was rather unexpected. Nearly 200 people were killed and hundreds injured in addition to the loss of houses and properties worth several crores of rupees (INR 1 crore = ~USD 222,000). The 2004 tsunami appears to be the first of its kind to have affected the Kerala coast. The complex nature of tsunami wave propagation in the shallow seas off the Kerala coast is evident from the non-uniform pattern of wave run-up, inundation and impact along the cost. The characteristics of the tsunami wave like initial withdrawal, number of waves, arrival time, etc. presented striking differences from location to location. There have been efforts by various authors 1-8 to study different aspects of the tsunami wave characteristics along the Kerala coast by utilizing the post-tsunami field survey data, tide gauge data and data from other sources. The pertinent need to pool the available literature to bring out the salient characteristics of tsunami wave propagation in the Arabian Sea has been felt and is attempted in this paper. Further, results of numerical model studies conducted to understand the tsunami wave propagation for different sources of tsunami and assess their hazard proneness in different sectors of Kerala coast are also presented. Oceanographic Processes that affect Tsunami Propagation Several oceanographic processes are known to individually or collectively affect the observed characteristics of tsunami wave. The processes generally identified 9-14 are: i Quarter wave resonance amplification in bays and gulfs ii Helmholtz resonance in harbours iii Constructive interference iv Boundary reflections v Interaction with astronomical tides vi Coupling with internal waves due to ocean density gradients vii Trapping of long gravity energy on continental shelves through Oscillations of the First Class (OFC) and Oscillations of the Second Class (OSC) via the mechanism of trapped and partially leaky modes.

2 532 INDIAN J. MAR. SCI., VOL. 39, NO. 4, DECEMBER 2010 viii ix x xi xii xiii Interaction with the strong tidal current gradients near regular and degenerate semidiurnal and diurnal tidal amphidromic points Extraction of energy from opposing ocean currents, through Reynolds eddy stresses. Interaction with the wind wave setup Focusing and defocusing of tsunami energy due to ocean bathymetric features such as ridges and trenches. Phase or frequency dispersion and amplitude dispersion (non linear effects). Breaks in the continental shelves through which tsunami waves traveling through the deeper water interact with tsunamis in the shallow water Tsunami wave characteristics along the Kerala coast The 2004 tsunami wave characteristics along the Kerala coast are summarized from the available studies 1-8 and presented below. Run-up heights The observed run-up along the Kerala coast is reproduced in Fig. 2 from Kurian et al. 1. The run-up distribution shows wide variations along the coast. In the Pozhiyur to Vizhinjam (southernmost) sector, the run-up level was only up to 1.5 m, whereas in the Vizhinjam Varkala sector, it was m. In the southern sectors of the Quilon district, the run-up was up to 3 m. The run-up increased further in the northern sectors of Quilon district. In the Cheriyazhikkal area, run-up up to 4.5 m was reported. Fig. 1 Kerala coast and the southeastern Arabian Sea. The complex bathymetry of the sea due to the Lakshadweep and Maldive group of islands can be seen.

3 KURIAN & PRAVEEN : TSUNAMI WAVE PROPAGATION IN THE ARABIAN SEA 533 Fig. 2- Run-up along the Kerala coast; panels 1 5 cover the whole coast starting from Manjeswaram in the north to Pozhiyur in the south; locations of field visit are also marked (after Kurian et al. 1 ).

4 534 INDIAN J. MAR. SCI., VOL. 39, NO. 4, DECEMBER 2010 In Azhikkal and up to Kayamkulam inlet, severity of attack of the tsunami was further intensified, with the highest run-up of up to 5 m. In the sector immediately to the north of Kayamkulam inlet also, the tsunami onslaught was severe with run-up level up to 5.0 m. Further north, in the Arattupuzha region and up to Thottappally, the run-up level reduced to 3.5 m. From Thottapally onwards, there was further decrease till south of Anthakaranazhi inlet. In the zone around Anthakaranazhi inlet, there was an increase in the run-up level reaching up to 3.5 m. Further north, in the Chellanum Puthuvype region around Cochin, run-up level decreased to 3 m. However, in the Edavanakkad region, the run-up level increased drastically and went upto 4.5 m. There was reduction in the run-up level further north, with a drastic reduction in the zone immediately north of the Munambam inlet. However, in the sector further north, the level increased showing up to 3 m around Vadanapally. There was again a drastic decrease in the sector south of the Ponnani inlet. An increase in the level was found north of Ponnani inlet and run-up level up to 2.5 m was found in Beypore inlet, south of Calicut. In the northern parts of Kerala coast comprising of the Calicut, Cannannore and Kasargod districts, the run-up levels were generally low in the range 1.0 to 2.5 m. However, a short sector around Choottad was notable for a high run-up level of m, which was not reported anywhere in the northern Kerala. Inundation The extent of horizontal inundation along the Kerala coast is reproduced in Fig. 3 from Kurian et al. 3. In the Trivandrum Quilon sector of southern Kerala, where the run-up was of medium value, the inundation was very less with values less than 50 m. The relatively lower inundation here not extending beyond the berm for most part of this sector is due to the higher elevation of the backshore. However, further north, in the northern sectors of Quilon district, the inundation increases commensurate with the higher run-up and lower backshore terrain level. The highest inundation of 2350 m along the Kerala coast was observed in the sectors adjoining the Kayamkulam inlet where the highest run-up also was observed. Further north, the inundation decreased, but it was not as low as in the Trivandrum Quilon sector, in spite of the lower run-up values; in a couple of locations inundation of 1000 m or more were also observed. Arrival times Fig. 4 which is reproduced from Baba et al. 6 gives the arrival times of tsunami waves at different locations of the Kerala coast. All these waves were not necessarily significant at all the locations. In the Trivandrum Quilon sector of southern Kerala, three waves were reported in most of the locations while a fourth wave also was reported in a couple of cases. However, further north in Fig. 3 Extent of Horizontal Inundation along the Kerala coast due to the 2004 tsunami (after Kurian et al. 3 )

5 KURIAN & PRAVEEN : TSUNAMI WAVE PROPAGATION IN THE ARABIAN SEA 535 Fig. 4 Arrival times of the tsunami waves at various locations on the Kerala coast; panels 1-5 cover the whole coast staring from Manjeswaram in the north to Pozhiyur in the south (after Baba et al. 6 ) the Alleppey - Cochin sector only two waves were reported though at Andhakaranazhi the third wave also was reported in the late afternoon. In the northern Kerala the third wave was observed in most of the locations. It is significant that the second and third waves correspond to either late evening of 26 th or early morning of 27 th December. It is also evident that the first few direct waves from the tsunami did not arrive at all the locations with significant amplitudes, to be noticed by eye witnesses. Tsunami wave transformation processes in the Arabian Sea The post-tsunami field survey data, tide gauge data and visual observations for the Kerala coast were used by Murty et al. 5 to identify the different wave transformation processes that contributed to the observed tsunami characteristics along the Kerala coast. From the tide gauge data for Neendakara and Cochin, they identified the different wavelets. The plot of amplitudes of wave crests against arrival time obtained by them is reproduced in Fig. 5. They identified 4 sets of waves. The first set of waves was explained as direct waves travelling by multiple paths, subject to all local shallow water effects, such as diffraction, refraction, scattering and local resonances (and dissipation). The second set of waves, based on the arrival times (using again multiple paths), is explained as, reflection from the east side of the Lakshadweep-Maldive Ridge (LMR) and the east coast of Africa. Waves that arrived beyond the late hours of 26 th December cannot be explained as due to reflected waves, even invoking the multiple path

6 536 INDIAN J. MAR. SCI., VOL. 39, NO. 4, DECEMBER 2010 Fig. 5 Arrival times of various tsunami wave trains at Cochin (top panel) and Neendakara(bottom panel)(after Murty et al. 5 ). hypothesis and hence water level variations from 27 th December onwards were attributed to energy trapping on the continental shelf and secondary undulations. Murty et al. 5 also found a single crest (like a solitary wave) whose amplitude was the second highest (after the direct waves) in the tide gauge record at Neendakara which was not evident for Cochin. They surmised that the single solitary type wave was due to a succession of total internal reflections on the west side of the LMR. Numerical modeling studies: Materials and Methods In order to understand the hazard proneness of the state for different sources of tsunami, numerical models were set up. The Andaman-Sumatra and Makran subduction zones are the two tsunamigenic sources in the Indian Ocean which can affect this coast 16. The generation and propagation of tsunami waves were modelled using the TUNAMI-N2 model of the Tohoku University which has been set up and calibrated under a project sponsored by the Ministry of Earth Sciences at Centre for Earth Science Studies 7. The calibrated model was run for the two past tsunami relevant for this coast viz. Sumatra 2004 and Makran 1945, and a hypothetical worst case scenario (i.e. Sumatra like earthquake in the Makran Subduction Zone). The major input parameters for the TUNAMI-N2 numerical model are earthquake source parameters, topography and bathymetry data. The seismic parameters used for the 2004 Sumatra and 1945 Makran, earthquakes are from Mohamed Chlieh et al. 17 and Jaiswal et al. 18 respectively. Sumatra like earthquake in the Makran Subduction Zone is assumed for the Hypothetical Worst Scenario case. The TUNAMI-N2 numerical model uses the nested grids (Fig. 6) namely, A, B, C and D. The extent of grids was decided based on the region to be modelled and the computational time. The grid D was created, so as to cover the entire study area on which the run-up and inundation have to be estimated. Starting from grid D, each outer grid was expanded thrice the corresponding inner grid. More or less the entire Indian Ocean was covered in the outer grid, namely grid A. The topographic and bathymetric data for the finer grid i.e. grid D, was obtained from the field survey conducted along the study area and C-Map respectively. The main objective of the field survey was to obtain the general topography of the terrain, the elevation, varying changes in and around the coast, general features of the beach and shoreline, etc. The topographic and bathymetric data for the coarser grids were provided from the sources such as SRTM (Shuttle Radar Topographic Mission), C-Map, NHO (Naval Hydrographic Office) map and GEBCO (General Bathymetric Chart of the Oceans). Results The results of calibration of this model for a 40 km sector of Kerala coast are reproduced in Table 1 from Praveen et al. 7. As can be seen there is very good correspondence between the simulated and observed run-up except at one location viz. Tharayilkadavu. The anomaly at Tharayilkadavu could probably be either due to observational error or model computational error which results from inaccurate bathymetric/topographic data or both. By making use of model outputs, inundation maps have been prepared for the whole coast from Trivandrum in the south to Kasargod in the north. Typical examples of inundation outputs for different segments of the coast are given in Figs 7 and 8. By looking at the maps

7 KURIAN & PRAVEEN : TSUNAMI WAVE PROPAGATION IN THE ARABIAN SEA 537 Table 1- Comparison of observed and computed tsunami run-up (m) at selected locations on the coast of Kerala (source: Praveen et al. 7 ) Sl. No. Location Latitude Longitude Run-up (m) Simulated Observed 1 Neendakara 8 56' 09.77" 76 32' 09.77" Kovilthottam 8 59' 44.23" 76 31' 23.34" Tharayilkadavu 9 09' 18.44" 76 27' 18.10" Pandarathuruthu 9 02' 12.84" 76 30' 30.92" Srayikkad 9 05' 56.97" 76 28' 50.36" Cheriazhikkal 9 07' 52.57" 76 28' 02.07" Azhikkal 9 08' 06.64" 76 27' 50.09" Valiazhikkal 9 08' 23.79" 76 27' 43.54" Kallikkadu 9 12' 13.69" 76 25' 59.18" Aarattupuzha 9 12' 55.08" 76 25' 38.79" Trikkunnapuzha 9 15' 30.14" 76 24' 26.34" Fig. 6 The nested grids A,B,C & D used in the numerical simulation. Fig. 7 Examples of inundation outputs for different scenarios for the southern Kerala coast

8 538 INDIAN J. MAR. SCI., VOL. 39, NO. 4, DECEMBER 2010 Fig. 8 Examples of inundation outputs for different scenarios for the northern Kerala coast

9 KURIAN & PRAVEEN : TSUNAMI WAVE PROPAGATION IN THE ARABIAN SEA 539 from the south through the north it can be seen that while the Sumatra tsunami has a tremendous impact in the southern Kerala coast, the hypothetical case of Makran has higher inundation in the central and northern parts of the Kerala coast. The Makran 1945 has little impact on any part of the coast except the inlets, with the inundation line right on the shore particularly in the southern parts of Kerala coast. Fig. 7(a, b) presents the scenario for two locations in the southern Kerala which are typical for the dominance of Sumatra source. Incidentally, Fig. 7 (b) pertains to the sector just north of the Kayamkulam inlet where the 2004 tsunami wreaked havoc due to high run-up and inundation. Fig. 7(c) presents the scenario for a typical location where the Sumatra and Makran sources have more or less the same impact, in other words it is like a transition point for the switchover of the dominance of Makran over Sumatra. This sector comes more or less midway between Alleppey and Cochin. Fig. 8 presents three cases further north. Fig. 8(a) presents the scenario for a sector in the Trichur district north of Cochin while Fig. 8(b) and (c) presents cases for the Malappuram and Calicut districts in northern Kerala. It can be seen that the impact of the Makran becomes more dominant as we go further north. Discussion The Indian Ocean is different from the other oceans in that the boundary reflections significantly influence the tsunami propagation 4. Kowalik et al. 9 depicts the multiple reflections from the Indian peninsula and Sri Lanka which results in a reverse wave towards Indonesia. The south eastern Arabian Sea is particularly so where in addition to the reflections from the continental mass the chain of islands constituting the Lakshdweep-Maldive Archipelago plays a prominent role. The bathymetry in the Archipelago region is so complex that in addition to reflection, refraction and diffraction total internal reflection 5 also can occur. Thus the Kerala coast which borders the southeastern Arabian Sea can expect extended periods of high waves as seen by the high waves in the late evening of 26 th and morning of 27 th for the 2004 Sumatra. Oscillations of smaller amplitudes will prevail for several days depending on the latitude as has been shown by Murty et al. 5. It is interesting to note that while the Sumatra 2004 tsunami generated run-ups as high as 5 m along the Kerala coast, the Makran 1945 with the source in the Arabian Sea itself and more proximate to the Kerala coast did not produce any significant impact along the coast. Even the hypothetical scenario of Sumatra like rupture intensity at Makran does not produce the type of run-up along the Kerala coast as seen for Sumatra This probably has to do with the energy directivity of the tsunami. Though the hypothetical Makran source does not generate run-up as high as the Sumatra along the Kerala coast, it indeed appears to be more hazardous for a major part of the Kerala coast in terms of inundation. This anomalous behaviour in the inundation in spite of lower run-up can be attributed to the beach morphology. In general, the beach elevation is relatively less for the central and northern Kerala coast when compared to the southern coast. This in turn could be linked to the lower wave energy of the central and northern Kerala coast when compared to the southern coast of higher wave energy 15. Thus the same wave amplitude and run-up can generally cause more inundation in the central and northern Kerala coast than the southern coast. The Vizhinjam-Sakthikulangara sector in southern Kerala did not undergo any inundation for the 2004 tsunami in spite of run-up of the order of 2-3m 1,3. The higher inundation along the northern Kerala coast is an important point from the point of view of tsunami hazard mitigation as Makran is considered to be the other potential tsunamigenic source in the Indian Ocean 16,19. Rajendran et al. 20 observed a significant discrepancy between the predicted and the observed arrival time of 1945 Makran tsunami and they attributed this disparity to submarine landsides triggered by the earthquke. Any future earthquake from Makran, particularly of higher intensities, might generate submarineslide generated tsunami which could add another dimension to the tsunami hazard from Makran source. This has not been considered in the numerical modeling carried out in the present study and has to be dealt with separately in order to have realistic pictures of inundation due to tsunami from Makran source. Acknowledgements Authors are grateful to M. Baba to former Director and Drs T.S. Shahul Hameed and K.V. Thomas, scientist, Centre for Earth Science Studies for useful discussions. Thanks are due to Director, CESS for permission for publication of the paper. Part of the work was done under a project funded by the Ministry of Earth Sciences.

10 540 INDIAN J. MAR. SCI., VOL. 39, NO. 4, DECEMBER 2010 References 1 Kurian, N.P., Pillai, A.P., Rajith, K., Murali Krishnan, B.T. & Kalaiarasan, P., Inundation characteristics and geomorphological impacts of December 2004 tsunami on Kerala coast, Current Science, 90 (2006) Kurian, N.P., Baba, M., Rajith, K., Nirupama, N. & Murty, T.S., Analysis of the tsunami of 26 December 2004 on the Kerala coast of India Part I: Amplitudes, Marine Geodesy, 29 (2006) Kurian, N.P., Rajith, K., Murali Krishnan, B.T., Nirupama, N. & Murty, T.S., Analysis of the tsunami of December 26, 2004, on the Kerala Coast of India-Part III: Inundation and Initial Withdrawal, Marine Geodesy, 29 (2006) Murty, T.S., Kurian, N.P. & Baba, M., Trans-oceanic reflection of tsunamis: the Kerala example, Disaster & Development, 1(2006) Murty, T.S., Kurian, N.P. & Baba, M, Roles of reflection, energy trapping and secondary undulations in the tsunami on Kerala coast, Int. Journ. Ecol. & Dev., 10 (S08)(2006) Baba, M., Kurian, N.P., Murali Krishnan, B.T., Nirupama, N. & Murty, T.S., Analysis of the tsunami of 26 December 2004 on the Kerala coast of India - Part II: Arrival times, Marine Geodesy, 29 (2006) Praveen, S.S., Reshmi, A.K., Dhanya, P., Arjun, S., Kalarani, Kurian, N.P., Ramana, Murthy, M.V, Shahul, Hameed, T.S. & Prakash, T.N., Numerical Simulation of 26 December 2004 tsunami on the Southwest coast and Lakshadweep islands of India, Marine Geodesy, Special Issue on Tsunamis, (in print). 8 Arjun, S., Kalarani, Dhanya, P., Praveen, S.S., Reshmi, A.K., Kurian, N.P., Ramana, Murthy, M.V., Shahul, Hameed, T.S. & Prakash, T.N, Numerical simulation of the Makran 1945 tsunami on the southwest coast and Lakshadweep islands of India., Marine Geodesy, (in print). 9 Kowalik, Z., Knight, W., Logan, T. & Whitmore, P., Numerical modeling of the global tsunami: Indonesian tsunami of 26 December 2004, Science of Tsunami Hazards, 23 (1)(2005) Murty T S, Nirupama N, Nistor I & Rao A D, Role of trapped and leaky modes around Andaman and Nicobar islands: Tsunami of 26 December 2004, Proc. National Workshop on Tsunami Effects and Mitigation Measures (Allied Publishers Pvt. Ltd.), Chennai, Murty, T.S., Nirupama, N. & Rao, A.D., Why the Earthquakes of 26 December 2004 and the 27th March Differed so Drastically in their Tsunamigenic Potential? Voice of the Pacific, 21(2)(2005) Murty, T.S., Nirupama, N., Nistor, I. & Rao, A.D., Leakage of the Indian Ocean tsunami Energy into the Atlantic and Pacific Oceans, J. Canadian Association of Exploration Geophysicists, CSEG Recorder, (2005) Nirupama, N., Murty, T.S., Rao, A.D. & Nistor, I., Numerical Tsunami Models for the Indian Ocean Countries and States, Indian Ocean Survey, 2 (1) (2006) Nirupama, N., Murty, T.S., Nistor, I. & Rao, A.D., The Energetics of the tsunami of 26 December 2004 in the Indian Ocean: A Brief Review, Marine Geodesy, 29 (1)(2006) Baba, M. & Kurian, N.P., (Ed.), Ocean Waves and Beach Processes, Centre for Earth Science Studies, Trivandrum, 1988, pp Chadha, Tsunamigenic sources in the Indian Ocean: factors and impact on the Indian landmass, in: The Indian Ocean Tsunami, edited by T.S. Murty, U. Aswathnarayana & N. Nirupama, (Taylor & Francis) 2007, pp Mohamed, Chlieh., Jean-Philippe, Avouac., Vala, Hjorleifsdottir., Teh-Ru, Alex, Song., Chen, Ji., Kerry, Sieh., Anthony, Sladen., Helene, Hebert., Linette, Prawirodirdjo., Yehuda, Bock. & John, Galetzka., Coseismic slip and afterslip of the Great Mw 9.15 Sumatra-Andaman Earthquake of 2004, Bulletin of the Seismological Society of America, 97(2007) Jaiswal, R.K., Singh, A.P. & Rastogi, B.K., Simulation of the Arabian Sea tsunami propagation generated due to 1945 Makran Earthquake and its effect on western parts of Gujarat (India), Nat. Hazards, (2008), doi: /s Rastogi B. K. A historical account of the earthquakes and tsunamis in the Indian Ocean, in: The Indian Ocean Tsunami, edited by T. S. Murty, U. Aswathnarayana & N. Nirupama, (Taylor & Francis) 2007, pp Rajendran, C.P., Ramana, Murthy, M.V., Reddy, N.T. & Kusala, Rajendran., Hazard implications of the late arrival of the 1945 Makran tsunami, Current Science, 95 (2008)