2018 술라웨시 지진: 팔루의 주향이동 쓰나미와 파국적 액상화

18:02 WITA: The Palu-Koro Fault Ruptures at Supershear Speed

On the evening of September 28, 2018, the inhabitants of Palu, the capital of Central Sulawesi province in Indonesia, were ending their week in the warm dusk of an equatorial Friday. The city of approximately 340,000 sat at the head of a narrow bay — Teluk Palu — flanked by steep hills and traversed by the Palu-Koro Fault, one of Indonesia's most active Strike-Slip FaultA fault where blocks of rock move horizontally past each other. The San Andreas Fault and North Anatolian Fault are major strike-slip faults that produce destructive earthquakes. systems.

At 6:02 PM local time, the fault broke. The earthquake lasted approximately 30 to 35 seconds. Its MagnitudeA single number that quantifies the total energy released by an earthquake. Each whole number increase represents roughly 31.6 times more energy released. was 7.5. The EpicenterThe point on the Earth's surface directly above the hypocenter (focus) where an earthquake originates underground. Often reported as the earthquake's location in news reports. was approximately 77 kilometres to the north of the city, and the rupture propagated southward at extraordinary speed — a speed that would become one of the defining scientific features of this event.

In the minutes that followed the shaking, multiple catastrophes unfolded simultaneously. A TsunamiA series of ocean waves generated by sudden displacement of the seafloor during an underwater earthquake. Tsunamis can travel across entire ocean basins at jet speed (700+ km/h). inundated Palu Bay. [[Liquefaction]] flow slides engulfed entire neighbourhoods. Buildings collapsed across the city and the surrounding region. The total death toll, when finally compiled, exceeded 4,340 people, with thousands more missing and presumed dead beneath the flow slides. More than 170,000 people were displaced.

The Sulawesi earthquake and its cascading consequences challenged several established assumptions in seismology and tsunami science, providing a dataset that required textbooks to be updated and hazard models to be fundamentally reconsidered.

Use Earthquake Energy Calculator to understand the energy characteristics of the M7.5 event. Use Seismic Risk Checker to assess comparative seismic risk for different structural types in environments similar to Palu.

Supershear Rupture: When Earthquakes Break the Sound Barrier

The speed at which an earthquake rupture propagates along a fault has important consequences for the pattern of ground shaking it produces. Typical rupture speeds are 70 to 85 percent of the shear wave velocity of the surrounding rock — a speed called the Rayleigh wave velocity — which for crustal rocks averages around 2.5 to 3.5 kilometres per second.

Analysis of Seismic WaveAn elastic wave generated by an earthquake or explosion that propagates through the Earth. Seismic waves carry the energy released at the earthquake source to distant locations. recordings from the September 28, 2018 earthquake revealed that the Fault RuptureThe breakage of rock along a fault during an earthquake, releasing stored elastic energy as seismic waves. Rupture length can range from meters (small quakes) to 1,000+ km (great earthquakes). propagated southward from the EpicenterThe point on the Earth's surface directly above the hypocenter (focus) where an earthquake originates underground. Often reported as the earthquake's location in news reports. toward Palu at speeds that exceeded the shear wave velocity of the crustal rock — a phenomenon called supershear rupture. The rupture velocities estimated from various analyses range from approximately 4 to 5 kilometres per second, substantially faster than the theoretical threshold for conventional rupture propagation.

Supershear rupture is analogous to supersonic motion in air. Just as an aircraft exceeding the speed of sound creates a sonic boom — a concentrated pressure wave at the Mach cone — a supershear earthquake rupture creates a "Mach wave" of intensely concentrated seismic energy along a cone extending from the rupture front. Material within this Mach cone receives far more destructive energy than would be predicted by conventional ground motion attenuation models. The city of Palu, situated directly along the path of the southward-propagating rupture, received ground motions consistent with this directivity effect.

Supershear rupture had been documented in a handful of previous earthquakes — including segments of the 2002 Denali fault earthquake in Alaska and the 2001 Kunlun earthquake in Tibet — but never with the combination of proximity to a major city, well-instrumented recording network, and catastrophic consequences that the Sulawesi event provided. The 2018 data will be studied for decades as the best-documented case of supershear rupture with direct observation of its effects on human settlements. In particular, the dataset is being used to validate numerical models of fault rupture dynamics and to improve ground motion prediction equations for sites in the forward directivity zone of fast-rupturing strike-slip faults.

The Palu Bay Tsunami: Strike-Slip Faults Should Not Make Tsunamis

The conventional understanding of tsunami generation holds that large tsunamis require significant vertical displacement of the seafloor — the kind of motion produced by thrust faults, where one block is pushed over another. [[Strike-slip-fault]]s, which move primarily horizontally with minimal vertical displacement, were generally considered poor tsunami generators because they do not efficiently move water up or down.

The Palu Bay TsunamiA series of ocean waves generated by sudden displacement of the seafloor during an underwater earthquake. Tsunamis can travel across entire ocean basins at jet speed (700+ km/h). violated this expectation profoundly. Within minutes of the earthquake, waves of three to seven metres struck the shores of Palu Bay, killing hundreds of people on beaches and in coastal areas. The timing — the first wave arrived perhaps 3 to 6 minutes after the shaking ended — made warning-based evacuation essentially impossible.

How did a Strike-Slip FaultA fault where blocks of rock move horizontally past each other. The San Andreas Fault and North Anatolian Fault are major strike-slip faults that produce destructive earthquakes. generate a significant TsunamiA series of ocean waves generated by sudden displacement of the seafloor during an underwater earthquake. Tsunamis can travel across entire ocean basins at jet speed (700+ km/h). in Palu Bay? Post-event research identified several contributing mechanisms. First, portions of the Palu-Koro fault system passing through Palu Bay are not perfectly strike-slip but have oblique components that created localised areas of seafloor uplift or subsidence. Second, the supershear rupture speed may have produced dynamic effects — including lateral pressure changes in the water column — that contributed to wave generation by mechanisms not present in slower ruptures. Third, and perhaps most importantly, large underwater Earthquake-Triggered LandslideThe downslope movement of soil and rock triggered by earthquake shaking. Landslides can bury entire communities and may cause more casualties than the shaking itself.s triggered by the earthquake shaking on the steep submarine slopes of Palu Bay generated their own Secondary Earthquake HazardsHazards triggered by earthquake shaking rather than the shaking itself — including tsunamis, landslides, liquefaction, fires, dam failures, and chemical releases. Often cause more damage than shaking. tsunami waves, which combined with any tectonic source to produce the observed run-up.

The narrow geometry of Palu Bay — essentially a fjord approximately 30 kilometres long and 5 kilometres wide, oriented perpendicular to the fault — amplified any tsunami waves that entered the bay through a process called "resonance amplification." The bay's geometry concentrated wave energy rather than dispersing it, producing higher run-up at the head of the bay than would have occurred in open water.

The relative contributions of tectonic displacement, submarine landslides, and dynamic supershear effects remain an active area of research and debate. The complexity of the TsunamiA series of ocean waves generated by sudden displacement of the seafloor during an underwater earthquake. Tsunamis can travel across entire ocean basins at jet speed (700+ km/h). source in Palu Bay — involving at minimum tectonic deformation, submarine landslides, and possibly dynamic supershear effects — makes it one of the most scientifically challenging tsunami events in recent history to model and understand, and has motivated new research on how Strike-Slip FaultA fault where blocks of rock move horizontally past each other. The San Andreas Fault and North Anatolian Fault are major strike-slip faults that produce destructive earthquakes. earthquakes can generate tsunamis in geometrically favourable settings.

Liquefaction Flow Slides: Entire Neighbourhoods in Motion

The most dramatic and viscerally disturbing aspect of the 2018 Sulawesi disaster was the occurrence of massive LiquefactionA phenomenon where saturated, loose soil temporarily loses strength and behaves like a liquid during strong shaking. Can cause buildings to sink, tilt, or collapse into the ground. flow slides in two Palu neighbourhoods — Petobo to the south and Balaroa to the northwest. These were not ordinary examples of LiquefactionA phenomenon where saturated, loose soil temporarily loses strength and behaves like a liquid during strong shaking. Can cause buildings to sink, tilt, or collapse into the ground. in which sandy soils lose their bearing strength and buildings tilt or settle. They were something far more extreme: entire neighbourhoods, covering areas of tens of hectares, that began to move as coherent blocks and flowed distances of hundreds of metres before coming to rest as chaotic mixtures of soil, debris, and former buildings.

[[Liquefaction]] occurs when water-saturated, loosely packed sediments lose their grain-to-grain contacts during intense shaking and temporarily behave as a fluid. In the Petobo and Balaroa areas, the local geology consisted of thick sequences of water-saturated alluvial sediments deposited by the Palu River and its tributaries. These sediments, when subjected to the strong ground shaking of the M7.5 earthquake, liquefied extensively and rapidly.

What made the Sulawesi case exceptional was the slope of the terrain. Both Petobo and Balaroa were situated on gently sloping alluvial fans — not steeply inclined hillslopes where landslides are expected, but gently tilted surfaces with gradients of one to two degrees. The liquefied soil, with its shear strength reduced almost to zero, could not maintain itself on even this gentle slope and began to flow downslope.

Petobo and Balaroa: Homes Carried 700 Meters

In Petobo, the flow slide carried an area of approximately 180 hectares — containing thousands of houses, roads, agricultural fields, and the infrastructure of a substantial neighbourhood — downslope by distances of up to 700 metres. Structures that had been standing were first engulfed by the moving soil, then shattered as the differential movement within the flowing mass tore them apart, and finally deposited in a jumbled, compressed mass of soil, timber, concrete, and household contents at the downslope boundary of the slide.

The timing of the flow slide compounded the casualties. Because it occurred in the evening of a weekday, many residents were at home. The liquefaction and flow developed rapidly — within seconds to minutes of the earthquake shaking — giving essentially no time for evacuation. Survivors described the ground shaking, followed almost immediately by a sensation of being on a moving surface that tilted and then carried them with it.

[[Lateral-spreading]] at the scale observed in Petobo and Balaroa was not previously well-documented in the scientific literature. While LiquefactionA phenomenon where saturated, loose soil temporarily loses strength and behaves like a liquid during strong shaking. Can cause buildings to sink, tilt, or collapse into the ground.-induced lateral spreads had been observed in many previous earthquakes, the distances involved — up to 700 metres — far exceeded what geotechnical engineers had typically modelled for gentle slopes. The Sulawesi case forced a revision of the maximum runout distances assumed in flow slide hazard assessments, with implications for land use planning in areas underlain by liquefiable soils throughout the world.

Satellite imagery analysis of the before-and-after state of the Petobo and Balaroa areas provided unprecedented documentation of the spatial extent and character of the flow slides. Researchers were able to track individual buildings from their pre-earthquake positions to their post-slide locations, quantifying the distances and directions of movement throughout the slide area. This spatial analysis revealed that the movement was not uniform — some areas moved much further than others — and that the pattern of movement was controlled by subtle variations in the thickness and properties of the liquefiable layer, the slope gradient, and the presence of drainage channels that influenced where water and liquefied material could escape.

Warning System Failure: 34 Minutes of Missed Opportunity

Indonesia operates the InaTEWS system — the Indonesian Tsunami Early Warning System — which was established after the catastrophic 2004 Indian Ocean tsunami. The system monitors seismic activity in real time, detects tsunamigenic earthquakes, and is supposed to disseminate warnings to coastal communities within minutes of a triggering event.

On September 28, 2018, InaTEWS issued a TsunamiA series of ocean waves generated by sudden displacement of the seafloor during an underwater earthquake. Tsunamis can travel across entire ocean basins at jet speed (700+ km/h). warning three minutes after the Sulawesi earthquake — a performance that met the system's technical specifications. The warning indicated expected wave heights of 0.5 to 3 metres — substantially lower than the waves that actually struck Palu Bay. The warning was cancelled approximately 34 minutes after the earthquake, before the TsunamiA series of ocean waves generated by sudden displacement of the seafloor during an underwater earthquake. Tsunamis can travel across entire ocean basins at jet speed (700+ km/h). had fully affected the coastline.

Several factors contributed to this failure. The tide gauge at the mouth of Palu Bay malfunctioned during the earthquake — the shaking likely disrupted the gauge's power or transmission system — providing no real-time confirmation of tsunami wave heights. The initial seismic MagnitudeA single number that quantifies the total energy released by an earthquake. Each whole number increase represents roughly 31.6 times more energy released. estimate used for the warning calculation was lower than the eventual revised magnitude, which would have triggered a higher warning level. And the cancellation of the warning — based on the absence of confirming data from the malfunctioning gauge — came while people had not yet been fully informed of the danger and some were returning to the coast.

The warning system failure had direct lethal consequences. Some residents of Palu's coastal areas had begun evacuating when they felt the Seismic WaveAn elastic wave generated by an earthquake or explosion that propagates through the Earth. Seismic waves carry the energy released at the earthquake source to distant locations.s and heard the Earthquake Early Warning (EEW)A system that detects an earthquake and sends alerts to people and systems before strong shaking arrives. Can provide seconds to tens of seconds of warning, enough to take protective action. alert. When the all-clear was issued, some returned to the beach. These individuals were killed by the waves. The 34-minute window between the earthquake and the warning cancellation, which should have been the period of maximum protective action, became instead a period in which the risk was reduced in official communications even as it was escalating in physical reality.

The post-event review of the InaTEWS failure identified multiple improvements: more robust tide gauges with redundant power and communication systems, faster magnitude estimation using finite-fault algorithms rather than point-source approximations, policies that limit early cancellation of warnings in the absence of confirming negative data, and better integration of warning information with public communication and evacuation protocols. It also reinforced the "self-evacuation" principle: coastal residents should treat strong earthquake shaking as the primary warning for local tsunamis, without waiting for any official alert, because the physics of near-field tsunami propagation give too little time for technology-dependent warning chains to be effective.

Rewriting the Textbooks: New Hazard Models After Sulawesi

The 2018 Sulawesi earthquake and its cascading consequences collectively forced revisions in multiple domains of earthquake and TsunamiA series of ocean waves generated by sudden displacement of the seafloor during an underwater earthquake. Tsunamis can travel across entire ocean basins at jet speed (700+ km/h). science.

In tsunami hazard assessment, the lesson is that Strike-Slip FaultA fault where blocks of rock move horizontally past each other. The San Andreas Fault and North Anatolian Fault are major strike-slip faults that produce destructive earthquakes.s cannot be categorically excluded as tsunami sources. The Palu Bay geometry — a narrow fjord-like bay aligned with the fault — created conditions in which even modest vertical displacement and submarine Earthquake-Triggered LandslideThe downslope movement of soil and rock triggered by earthquake shaking. Landslides can bury entire communities and may cause more casualties than the shaking itself.s could generate locally damaging waves. Tsunami hazard assessments for areas near Strike-Slip FaultA fault where blocks of rock move horizontally past each other. The San Andreas Fault and North Anatolian Fault are major strike-slip faults that produce destructive earthquakes.s must now consider the potential for Secondary Earthquake HazardsHazards triggered by earthquake shaking rather than the shaking itself — including tsunamis, landslides, liquefaction, fires, dam failures, and chemical releases. Often cause more damage than shaking. including landslide-generated tsunamis and for dynamic effects from fast-moving ruptures, not just the tectonic seafloor deformation from thrust faults.

In geotechnical hazard, the Petobo and Balaroa flow slides extended the known envelope for LiquefactionA phenomenon where saturated, loose soil temporarily loses strength and behaves like a liquid during strong shaking. Can cause buildings to sink, tilt, or collapse into the ground.-induced runout distances, requiring revision of hazard zone maps for gently sloping alluvial terrain in seismically active areas worldwide. The observation that flow slides can develop on slopes as gentle as one to two degrees — far gentler than the five to fifteen degrees typically required for dry landslides — means that alluvial fan and delta environments near active faults must be reassessed.

The 2018 Sulawesi earthquake is, in the language of earthquake science, a "surprising" event — one that revealed new aspects of fault behavior, Secondary Earthquake HazardsHazards triggered by earthquake shaking rather than the shaking itself — including tsunamis, landslides, liquefaction, fires, dam failures, and chemical releases. Often cause more damage than shaking. interaction, and system failure that had not been adequately considered in existing models. These surprises are simultaneously the most tragic and the most scientifically valuable outcomes of destructive earthquakes. Palu paid an enormous price. The knowledge purchased from that price is now embedded in hazard models, warning system designs, and land use policies that will, over time, reduce casualties from future events in similar settings around the world.

Indonesia's Seismic Risk Context

The 2018 Sulawesi earthquake occurred in a country that has one of the highest concentrations of seismic risk in the world. Indonesia straddles the boundary between the Eurasian, Australian, Pacific, and Philippine Sea plates, and experiences approximately 7,000 to 8,000 earthquakes per year, including dozens that are felt and several that cause damage. The archipelago's position at the intersection of multiple Subduction ZoneA region where one tectonic plate dives beneath another into the mantle. Subduction zones produce the world's largest earthquakes (M8.5+) and are associated with deep ocean trenches and volcanic arcs.s and Strike-Slip FaultA fault where blocks of rock move horizontally past each other. The San Andreas Fault and North Anatolian Fault are major strike-slip faults that produce destructive earthquakes. systems means that virtually every major island is exposed to severe seismic hazard, and the diversity of the hazard — subduction earthquakes with their associated TsunamiA series of ocean waves generated by sudden displacement of the seafloor during an underwater earthquake. Tsunamis can travel across entire ocean basins at jet speed (700+ km/h).s on Sumatra, Java, and Sulawesi's western coast; strike-slip earthquakes on internal faults throughout the archipelago; volcanic earthquakes throughout the volcanic arc — makes comprehensive hazard management extraordinarily complex.

Indonesia's rapid urbanization — the country's urban population has grown from approximately 20 percent in 1970 to over 55 percent today — has concentrated millions of people in coastal cities exposed to earthquake and tsunami hazard. The Building Code (Seismic)A set of legal requirements governing the design and construction of buildings to ensure minimum levels of earthquake safety. Updated after major earthquakes reveal new vulnerabilities. framework has improved progressively, incorporating lessons from each destructive earthquake, but enforcement remains inconsistent and the pace of informal urban growth frequently outstrips the regulatory capacity of local governments to ensure compliant construction.

The 2018 Sulawesi earthquake added to a long sequence of destructive Indonesian earthquakes that includes the 2004 Indian Ocean tsunami (which devastated Aceh), the 2006 Yogyakarta earthquake (which killed 5,700 people), the 2009 Padang earthquake (which killed 1,100), and the 2018 Lombok earthquakes (which killed over 500 just two months before Sulawesi). This pattern of recurrent destruction reflects the fundamental exposure of the Indonesian archipelago to seismic hazard, and the persistent gap between the hazard that science can characterize and the risk reduction that governance can implement at the scale and speed required to protect a rapidly growing urban population.

The Reconstruction of Palu: Navigating Multiple Hazard Zones

The reconstruction of Palu after the 2018 disaster presented Indonesian planners with an extraordinarily complex challenge. The earthquake had revealed multiple overlapping hazard zones: the narrow coastal strip exposed to TsunamiA series of ocean waves generated by sudden displacement of the seafloor during an underwater earthquake. Tsunamis can travel across entire ocean basins at jet speed (700+ km/h). inundation, the gentle alluvial fans susceptible to LiquefactionA phenomenon where saturated, loose soil temporarily loses strength and behaves like a liquid during strong shaking. Can cause buildings to sink, tilt, or collapse into the ground. flow slides, the steep hillslopes prone to earthquake-triggered Earthquake-Triggered LandslideThe downslope movement of soil and rock triggered by earthquake shaking. Landslides can bury entire communities and may cause more casualties than the shaking itself.s, and the fault trace itself — which runs through the centre of the city — where surface rupture and severe near-fault shaking can be expected in any future event.

The Indonesian government's initial response was to designate large areas of Palu as uninhabitable and to plan relocation of affected communities to new sites away from the most severe hazard zones. Implementation proved far more difficult than declaration. Land tenure in the relocation sites was contested. Communities resisted being moved from their social networks, livelihoods, and cultural connections to specific places. Building temporary shelter rapidly enough to protect displaced families through the rainy season competed with the need to develop permanent solutions that actually reduced future risk.

The Petobo and Balaroa flow slide areas were formally prohibited from reconstruction, and a plan for their conversion to green space — similar to Christchurch's Avon River Corridor — was developed. But informal rebuilding began in parts of these areas almost immediately, as families with no other options returned to the land they owned even when it had been devastated. Enforcing exclusion zones in the absence of viable alternatives for displaced residents is a persistent failure mode in post-disaster reconstruction worldwide, and Palu was not immune to it.

The Palu-Koro Fault: A Hazard Long Identified

One of the most sobering aspects of the 2018 Sulawesi disaster is how well known the Palu-Koro Fault had been to seismologists before the earthquake occurred. The fault had been mapped and characterized as highly active. GPS measurements had documented slip rates of several centimetres per year — among the fastest measured on any onshore Strike-Slip FaultA fault where blocks of rock move horizontally past each other. The San Andreas Fault and North Anatolian Fault are major strike-slip faults that produce destructive earthquakes. in the world. Historical records documented multiple large earthquakes on the fault over the preceding centuries, including events in 1927 and 1938 that had damaged Palu. The fault's trace ran directly through the city, and Seismic Hazard MapA map showing the probability of earthquake shaking exceeding specified levels over a given time period. Used by engineers, planners, and insurers to assess earthquake risk.s based on this knowledge had been produced by the Indonesian government and international partners.

What was missing was the translation of this hazard knowledge into practical risk reduction. The LiquefactionA phenomenon where saturated, loose soil temporarily loses strength and behaves like a liquid during strong shaking. Can cause buildings to sink, tilt, or collapse into the ground. susceptibility of Petobo and Balaroa had been identified in geotechnical studies before 2018. Tsunami inundation models for Palu Bay had been run and published, showing that the narrow bay geometry would focus and amplify waves from local fault ruptures. Early warning system gaps — including the vulnerability of coastal tide gauges to earthquake damage — had been discussed by Indonesian and international experts.

The gap between knowing about a hazard and acting to reduce the risk it poses is not unique to Indonesia or to the 2018 Sulawesi earthquake. It is arguably the central problem of applied earthquake science worldwide. The 2018 disaster added a case study of unusual scientific richness to the already substantial evidence for why closing this gap requires not just better hazard information but better governance, stronger enforcement of land use regulations, more investment in resilient warning infrastructure, and sustained public education — none of which is primarily a scientific problem, but all of which depend critically on the scientific foundation that events like the 2018 Sulawesi earthquake help to build.