Tsunami Earthquake” Based on Macroseismic, Seismological, and Tsunami Observations, and Modeling

This document is downloaded from DR‑NTU (https://dr.ntu.edu.sg) Nanyang Technological University, Singapore. Reassessment of the 1907 Sumatra “Tsunami Earthquake” based on macroseismic, seismological, and tsunami observations, and modeling Martin, Stacey Servito; Li, Linlin; Okal, Emile A.; Morin, Julie; Tetteroo, Alexander E. G.; Switzer, Adam D.; Sieh, Kerry E. 2019 Martin, S. S., Li, L., Okal, E. A., Morin, J., Tetteroo, A. E. G., Switzer, A. D., & Sieh, K. E. (2019). Reassessment of the 1907 Sumatra “Tsunami Earthquake” based on macroseismic, seismological, and tsunami observations, and modeling. Pure and Applied Geophysics, 176(7), 2831‑2868. doi:10.1007/s00024‑019‑02134‑2 https://hdl.handle.net/10356/136833 https://doi.org/10.1007/s00024‑019‑02134‑2 © 2019 Springer Nature Switzerland AG. All rights reserved. This paper was published in Pure and Applied Geophysics and is made available with permission of Springer Nature Switzerland AG. Downloaded on 07 Oct 2021 21:17:42 SGT Please contact first author for final PDF 1 1 A reassessment of the 1907 Sumatra “tsunami earthquake” based on 2 macroseismic, seismological, and tsunami observations and modelling 3 4 Stacey Servito Martin1†, Linlin Li1, Emile Okal2, Julie Morin3, Alexander Tetteroo4, Adam 5 Switzer1,5, Kerry Sieh1 6 7 8 9 10 11 1 Earth Observatory of Singapore, Nanyang Technological University, Singapore 12 2 Department of Earth & Planetary Sciences, Northwestern University, Evanston, U.S.A. 13 3 Laboratoire Magmas et Volcans, Université Clermont Auvergne, Clermont-Ferrand, France 14 4 Institute for History, Leiden University, Netherlands 15 5Asian School of the Environment, Nanyang Technological University, Singapore 16 17 † Corresponding author: [email protected] 18 Please contact first author for final PDF 2 19 Abstract 20 On 4 January 1907 an earthquake occurred off the west coast of Sumatra, Indonesia, with 21 an instrumental surface-wave magnitude (MS) in the range of 7.5 to 8.0 at periods of ~40s. The 22 tsunami it generated was destructive on the islands of Nias and Simeulue where it killed hundreds 23 and gave rise to the legend of the S’mong. This tsunami also appears in records in other parts of 24 the Indian Ocean basin, and as far as La Réunion island. Relative to its instrumented magnitude, 25 the size of the tsunami was anomalous qualifying it as a “tsunami earthquake.” However, unusually 26 for a tsunami earthquake the shaking on Nias was severe (6-7 EMS). We revisited the 1907 27 earthquake with a multidisciplinary approach by extracting evidence describing shaking effects or 28 the tsunami from written documents and by acquiring new seismograms. Combining these, we 29 discriminated two large earthquakes within an hour of each other with clear differences in 30 seismological character. The first we interpret to be a tsunami earthquake with characteristic low 28 31 levels of shaking and for which we estimate an average seismic moment (M0) 2.5 x10 dynes*cm 32 (MW ≈ 8.2) in the frequency band 6 – 8 mHz. These records document a regular growth of moment 33 with period: at the longest resolvable period (~170 s) the moment magnitude approaches MW ≈ 34 8.4. Our relocations suggest an epicentral location close to the trench for the mainshock. For the 35 second earthquake that was damaging on Nias, we estimate MS ≈ 7 based on seismograms and 36 phase data from Manila and Osaka. We also identified two MS ≈ 6 aftershocks within 24 hours of 37 the mainshock. We propose a model of seismic rupture which provides an acceptable fit to our 38 new dataset of tsunami run-up and inundation values from 88 local and far-field locations in the 39 Indian Ocean basin. 40 41 Please contact first author for final PDF 3 42 Introduction 43 Numerous large earthquakes have occurred off the West coast of Sumatra, Indonesia, 44 during the historical and instrumental eras (e.g. Newcomb & McCann, 1987), and in particular, 45 since 2000. These events include a tsunamigenic earthquake on 4 January 1907 near the islands of 46 Simeulue and Nias (Figure 1), which Newcomb & McCann (1987) associated with shaking of 47 such severity that “people on Nias could not stand.” As many as 370 people were killed on Nias 48 and at least 1818 lives were lost on Simeulue (Koloniaal Verslag, 1907-08); 1,205 in Tapah (nl: 49 Tapak) district, 431 in Simeuloeë Rajou, 130 in Salang and 52 in Leuköon on Simeulue 50 (Bataviaasch Nieuwsblad, 12 February 1907). It was feared 1,000 people were killed at Koela- 51 Deh on Tapa alone (Utrechts Nieuwsblad, 14 February 1907). This led to the disaster being 52 embodied in myth and legend on the island of Simeulue (e.g. McAdoo et al., 2006; Syafwina, 2014; 53 Rahman et al., 2017). As discussed below, the 1907 tsunami was also recorded in the far field, as 54 far away as La Réunion (Bertho, 1910), which gives the event a clearly anomalous character in the 55 context of the comparatively low “Pasadena” magnitude (MPAS = 7.6) assigned to the earthquake 56 by Gutenberg & Richter (1954), and for long, the only available measure of its size. 57 More recently, Kanamori et al. (2010) conducted an extensive seismological study of the 58 1907 earthquake based on a number of historical seismograms. While they did not compute a 59 seismic moment through waveform fitting, they measured a surface wave magnitude MS = 7.8 ± 60 0.25 and estimated a moment magnitude MW = 7.8 by scaling time domain amplitudes of body and 61 surface waves of the 1907 earthquake to those of nearby modern earthquakes with known moment 62 tensors. These results were obtained in the period range 40 − 50 s, but Kanamori et al. (2010) 63 stress that the source was obviously longer, and thus the moment should be larger at longer periods, 64 suggestive of a "tsunami earthquake". Please contact first author for final PDF 4 65 The term “tsunami earthquake” was first used by Kanamori (1972) to discuss the sources 66 of the 1896 Meiji Sanriku and the 1946 Unimak (Aleutian Islands) earthquakes, both of which 67 resulted in anomalously large tsunamis with respect to their instrumental magnitudes. This type of 68 event can be distinguished based on disproportionate relationships between surface wave 69 magnitude (MS) and seismic moment (M0), from the observation of longer than expected process 70 times despite small rupture areas (Sykes, 1971; Kanamori, 1972; Pelayo & Wiens, 1992; Polet & 71 Kanamori, 2000), from the lack of evidence for abnormal stress drops (Pelayo & Wiens, 1992), 72 and from lower than anticipated macroseismic intensities (Kanamori, 1972; Fukao, 1979; 73 Bourgeois et al., 1999). The longer process times result in red-shifting of the source spectrum, and 74 in inconsistencies between deficient seismic magnitudes measured at short-to-moderate periods 75 (thus relevant to macroseismic effects) and enhanced ultra-long period seismic moments 76 (controlling the generation of tsunamis). The ruptures of tsunami earthquakes have been observed 77 to propagate toward the trench axis (Polet & Kanamori, 2000) on very shallow dipping faults or 78 on splays (Fukao, 1979; Pelayo & Wiens, 1990, 1992) located in weakly coupled regions of 79 aseismic convergence (Pelayo & Wiens, 1990, 1992; Bourgeois et al., 1999), or at the very top of 80 the plate interface under conditions of sediment starvation leading to a jagged rupture (Tanioka et 81 al., 1997). Rupture velocities for such earthquakes are also less than expected for typical 82 earthquakes owing to the low rigidity of materials proximal to the trench axis (Fukao, 1979; Pelayo 83 & Wiens, 1990, 1992; Heinrich et al., 1998; Ihmlé et al., 1998). In addition, tsunami earthquakes 84 can occur as mainshocks, which Okal & Saloor (2017) qualified as “Primary Tsunami 85 earthquakes” (PTEs), or as “Aftershock Tsunami Earthquakes” (ATEs), following a larger, regular 86 megathrust event. The exceptional tsunami of the 2011 Tohoku event may be the result of the 87 combination of a regular megathrust event, and of a lower-frequency rupture of the shallowest Please contact first author for final PDF 5 88 portion of the interface, which might have qualified as an ATE, but for its occurrence only 3 89 minutes after the initial nucleation (Satake et al., 2013). Enhanced tsunami generation following 90 moderate-to-large earthquakes can occasionally be traced to ancillary phenomena, such as 91 triggered submarine landslides (Synolakis et al., 2002) or volcanic processes (Satake & Kanamori, 92 1991; Fukao et al., 2018). However, these are generally not labelled as “tsunami earthquakes.” 93 With the exception of the 2010 Mentawai event (e.g. Newman et al., 2011a; Hill et al., 94 2012), no other tsunami earthquakes have been conclusively identified off the Sumatran coast 95 during the modern or historical period. Although Kanamori et al. (2010) concluded in more 96 general terms that the 1907 Sumatra earthquake held all the hallmarks of a tsunami earthquake, 97 key aspects of this event remain unaddressed, including (i) conclusive, quantitative evidence of its 98 nature as a “tsunami earthquake”; and (ii) an estimate of the geometry and slip parameters of the 99 source supporting the reported distribution of the tsunami. In addition, perplexing observations 100 clearly in need of further study include (iii) the anomalously violent ground motions (Newcomb & 101 McCann, 1987; Kanamori et al., 2010) in comparison to other tsunami earthquakes; (iv) the lack 102 of aftershocks; and (v) the lack of land level changes comparable to those identified for other large 103 earthquakes in the Simeulue-Nias region, e.g., Meltzner et al., (2012; 2015). 104 In this article we employ a multidisciplinary approach to tackle these points. Through a 105 scrutiny of original macroseismic reports and the systematic analysis of a number of seismograms, 106 we separate the mainshock from a previously unsuspected large aftershock, occurring only ~53 107 minutes later.

Load more