Field investigation of an enigmatic 2018 tsunami in Indonesia
Indonesia is arguably one of the world’s countries most frequently hit by tsunamis. Indonesia is torn apart by tectonic forces that have dispersed its landmass into a maze of islands. The faults that crisscross the country have regularly reshaped its contours. Sudden displacements of the seafloor near active faults and volcanoes generate deadly tsunamis that have already devastated the country’s shorelines three times this century. The most infamous of these was the humongous 2004 Boxing Day tsunami, which killed 230 000 people. Two deadly tsunamis occurred in 2018. The latest one occurred in December, when volcano Anak Krakatoa collapsed into the waters of the Sunda Strait. Another one occurred on September 28th, when a major earthquake struck the city of Palu, on the island of Sulawesi.
Our scientific team, led by Jean-Philippe Goiran, researcher at Archéorient, was deployed there to gather first-hand observations within a month and a half of the catastrophe. The goal of the campaign was to understand how this tsunami formed, and why so many tsunamis hit the bay of Palu over the past few centuries.
The magnitude 7.5 September 28th earthquake occurred during a quiet, warm Saturday evening. The Palu-Koro fault, which runs below the city, ruptured suddenly (Fig.1). The rupture started in the north, rapidly unzipping the fault southwards at a velocity of more than 250 m per second. The rupture rushed through the city after about 25 seconds, and vanished 30 seconds later 130 km south of its starting point. By then, and within a few seconds, the city of 335,000 inhabitants had been cut into two parts, its streets and homes displaced laterally 4 to 5 meters across the fault trace (Fig.2). People died by the thousands. Yet very few people were actually killed by the collapse of buildings directly induced by ground shaking. The danger came from elsewhere. Within minutes, three vast landslides had carried away entire villages, croplands, and urban neighbourhoods, while tsunami waves up to 11m high had battered the beach front, penetrating 400m inland. These indirect effects claimed more than 4,400 lives and damaged more than 4,000 buildings.
There is no simple relationship between the magnitude of an earthquake and destruction to buildings, because buildings are sensitive to different shaking rhythms (seismic wave frequency) depending on their size. The city of Palu is mostly made up of small houses and these were little affected by direct ground shaking. Instead, large buildings such as universities and shopping malls were badly damaged. We now know that the Palu earthquake rupture unzipped at supershear velocities (Bao et al., 2019; Socquet et al., 2019), that is, that the rupture propagated very quickly, on a very smooth fault plane. As a result, most of the energy was radiated in low frequencies, which affected the largest buildings. High frequencies were also probably filtered out by the loose soils that underlie the city. In the end, the destruction turned out to be remarkably limited, considering the size of the quake. On the other hand, the low frequencies might have contributed to the widespread occurrence of soil liquefaction. During liquefaction, ground becomes loose, moving away from loads, such that houses sink and topple over their own ground.
This process damaged many houses, but was usually not deadly. Unfortunately, here and there vast panels of shallow sloping land completely yielded, decomposing into earthflows that displaced districts by hundreds of meters.
Meanwhile, an enigmatic tsunami struck the bay. A series of large waves battered its shorelines, stripping houses from their foundations, and dumping their debris hundreds of meters farther inland.
The tsunami came as a surprise for two reasons. First, the earthquake accommodated horizontal motion between two blocks (strike-slip displacement). Such motion does not produce a lot of up or down movements. Therefore, when it occurs under the seafloor, strike-slip displacement rarely displaces large volumes of water, and the resulting tsunami waves are fairly small. Second, in the case at hand, most displacements occurred on land. The only place where the seafloor was traversed by the fault was under the bay of Palu. A tsunami wave seemed therefore unlikely, and if one was triggered, its effects would remain confined to the bay. For these reasons the National Climatology and Geophysics Agency (BMKG) soon revoked the tsunami warning. Tsunami waves wrecked the coastline within minutes of the quake, affecting an already panicked population. Within such a short spell, the only efficient warning was the quake itself.
Setting off to Palu
The origins of the tsunami and of the soil liquefaction have been a matter of intense debate (Liu et al., 2018; Williamson et al., 2018). Local universities and governmental agencies teamed up to rapidly measure the run-up distance of the tsunami (Muhari et al., 2018; Fritz et al., 2018), and remap the sea-floor to detect any substantial changes following the quake. Many research teams worldwide used satellite images to produce maps of surface displacements and ground liquefaction within a days of the quake (e.g.: see here and there ). French ambassador Jean Charles Berthonnet visited Palu a few days after the catastrophe and delivered drinkable water processing facilities to the affected areas. Scientific field investigations might not be considered a priority right after a disaster, yet they are, because critical clues on how the tsunami and soil liquefaction unravelled must be gathered before they are effaced by early recovery efforts. It was therefore important to be there as soon as possible.
We were approached by Prof. Umar Suykur, from Tadulako University, to contribute to these early investigations. At the time, our team, composed of Dr Jean-Philippe Goiran (Archéorient), Dr Gilles Brocard (University of Grenoble-Alpes) and PhD candidate Stoil Chapanski (University of Paris 1- Sorbonne) was coring the coastal plain of Banda Aceh, in Northern Sumatra, where the 2004 Boxing Day tsunami rushed five kilometers inland. In coordination with Dedy Alfian and Dr Ella Meilianda from the Center for Tsunami mitigation (TDMRC) of Banda Aceh, we were then collecting sediments deposited by previous tsunamis known to have devastated ancient harbors and cities. The TDMRC was in charge of coordinating international efforts to gather field observations in Palu. Our team, supported by the French embassy in Indonesia conducted fielwork together with researchers from the Singapore Earth Observatory (EOS) and the University College of London (UCL).
Campaign objectives and design
JP Goiran has a long experience in studying tsunamis that struck antique settlements around the Mediterranean Sea (Rebemarichal et al., 2012; Goiran, 2012). This campaign offered the opportunity to observe fresh deposits, study their composition, and see how this relates to their immediate environment. This information is lacking on ancient sites and where the environment has been altered by centuries of natural and anthropogenic modifications. This campaign was also an opportunity to observe phenomena associated with the tsunami and the earthquake and that we could look for at archaeological sites. Another objective was to identify areas best suited for tsunami preservation, in order to locate sites ideal for future coring so as to assess the recurrence and origin of tsunamis in the past.
We collected sediment deposited by the tsunami (rubble, sediment of marine origin, Fig. 3) along the entire rim of the bay. We made laser measurements of tsunami height, run-up distance and run-up height at our sampling sites to relate deposit characteristics to tsunami power and height. We used a sonar to acquire bathymetric profiles that showed us the shape of the sea floor within a few hundred meters from the coastline (Fig. 4). Upon landing, tsunami wave height is strongly influenced by the shape of the sea-floor near the coast: this can either suppress or amplify their devastation power. We wanted to understand how coastal morphology influences variations in sediment texture and composition.
In the meantime Gilles Brocard and Dedy Alfian made important observations along the head scarps of the large earthflows that had struck the city and sampled soils in areas extensively affected by liquefaction (Fig.5).
Conducting field investigations during the immediate aftermaths of a catastrophe is not an easy task, because logistics is impaired by the damages imparted to facilities. At the time of arrival, the airport was experiencing extensive rebuilding, and landing was allowed only for a few hours each day. The road network, power grid and reticulated water had all been severely damaged; many buildings, bridges and roads had collapsed. It was therefore difficult to find places providing water and electricity devoid of structural damage. Such damage was a concern due to the many replicas that rattled the city during the months following the main shock. Mapping the seafloor also proved challenging because all boats anchored inside the bay had been destroyed by the tsunami (Fig. 6) forcing us to drive 40 km every morning to hire ships located at the entrance of the bay. These ships had to be large enough to withstand potential attacks by large sea crocodiles that had flocked to the area when tsunami waters receded to feed on the rafts of debris dragged into the sea.
Early results and moving forward
The bay of Palu is very unusual. It is a sort of marine Loch Ness, very long, narrow and deep, rapidly reaching depths in excess of 700m. It was commonly thought, therefore, that the source of the tsunami was one or several large submarine landslides triggered by the earthquake. Yet, the remapping of the deep sea floor after event did not reveal any new landslide large enough to generate the observed waves. Instead, it was found that the seafloor had dropped by several meters (Kongko et al., 2018). The bay of Palu has been hit by other tsunamis several times during the 20th century. These were not triggered by the Palu-Koro fault. Some were produced far out at sea and subsequently rushed into the bay. Because the bay is deep and narrow, it funnels the tsunami energy towards the beach front of Palu. Some tsunamis might also have originated within the bay. Our field investigations indeed show that large submarine landslides occurred in the recent past and were large enough to trigger deadly tsunamis.
We found that the rapid lateral variations in tsunami strength along the shoreline were due to changes in the submarine slope close to the shore. Gentler underwater slopes favoured deeper tsunami penetration. At places were the coast is very steep, or protected by a coral reef, the tsunami was forced upwards, rather than landwards, then rise vertically and collapse, or bounce back seawards. We further observed systematic ground subsidence along the coastline. In places it was triggered by the Palu-Koro fault itself (Fig. 7), but more often it was caused by coastal soil liquefaction leading to diffuse (Fig. 8), or sharp landsliding of the coast into the sea (Figs. 9 and 10).
On land, we found a strong relationship between the distribution of large deadly earthflows and places where soils were soaked with water, either voluntarily for irrigation, or accidentally through the leakage of reticulated water. This finding was confirmed using remote sensing (https://eartharxiv.org/tey6f/).
Samples are currently being studied at Archéorient using infrared spectroscopy (in the Commanderie of Jales, CNRS). As part of an upcoming study, we will take deeper samples as far down as 5 to 10 meters below the ground surface. Studying these deep sedimentary deposits will enable us to retrace the chronology of past tsunamis much farther back in time, probably over a few thousands of years. With the help of mathematical models, we can generate a relatively reliable estimate of the number of tsunamis that are likely to be produced in the area over the next century. The accuracy of the results to be obtained will be in accordance with other possible scientific approaches, such as the study of historical archives reporting earthquakes and tsunamis in the region. The results will be handed to stakeholders and hopefully will enter the public policy arena, both at the regional and national levels.
Funding was provided by the French embassy of Jakarta, through program PHC Nusantara, by LabEx IMU and by the CNRS.
Bao, H., Ampuero, J.-P., Meng, L., Fielding, E. J., Liang, C., Milliner, C. W., Feng, T., and Huang, H. 2019. Early and persistent supershear rupture of the 2018 magnitude 7.5 Palu earthquake. Nature Geoscience 12/3, 200-205
Fritz, H., Synolakis, C., Kalligeris, N., Skanavis, V., Santoso, F., Rizal, M., Prasetya, G., Liu, Y., and Liu, P., Field survey of the 28 September 2018 Sulawesi tsunami, in Proceedings AGU Fall Meeting Abstracts 2018.
Goiran, J.-Ph. 2012. Caractérisation d’un dépôt de tsunami dans le port antique d’Alexandrie par l’étude exoscopique des quartz : apports et limites de la méthode. Archéosismicité & tsunamis en Méditerranée, 157-190.
Kongko, W., Prihartanto, P., Yudhicara, Y., Putra, P., and Santoso, B. Rapid Survey on the Aftermath of Palu-Indonesia Tsunami 28 Sept. 2018 and the Plausible Genesis of Tsunami, in Proceedings AGU Fall Meeting Abstracts 2018.
Liu, P., Barranco, I., Fritz, H., Haase, J., Prasetya, G., Qiu, Q., Sepulveda, I., Synolakis, C., and Xu, X., What we do and don’t know about the 2018 Palu Tsunami-A future plan, in Proceedings AGU Fall Meeting Abstracts 2018.
Muhari, A., Imamura, F., Arikawa, T., Hakim, A. R., and Afriyanto, B. 2018. Solving the puzzle of the september 2018 palu, indonesia, tsunami mystery: Clues from the tsunami waveform and the initial field survey data. Journal of Disaster Research, v. 13.
Rebemarichal, I., Laurenti, A., Boehm, I., Goiran, J.-Ph.2012. Archéosismicité et tsunamis en Méditerranée, approches croisées. France.
Socquet, A., Hollingsworth, J., Pathier, E., and Bouchon, M. 2019. Evidence of supershear during the 2018 magnitude 7.5 Palu earthquake from space geodesy. Nature Geoscience, p. 1.
Williamson, A., Melgar, D., Xu, X., and Milliner, C., Tsunami Generation From Coseismic Deformation During the 2018 Mw 7.5 Palu Earthquake, in Proceedings AGU Fall Meeting Abstracts2018.
Gilles Brocard is a visiting research scientist in geoscience at ISTerre, Observatoire des Sciences de l’Univers (OSUG), University of Grenoble.
Jean-Philippe Goiran is a research scientist in coastal geoarchaelogy at Archéorient / University of Lyon 2. He is working on ancient harbour cities and landscape evolution.
Umar Syukur is a Professor at the University of Palu, Indonesia. He did his PhD in environmental science at the University of Paris in France.
Dedy Alfian and Ella Meilianda are research scientists at the Center for Tsunami mitigation (TDMRC) in Banda Aceh, Indonesia.