Two earthquakes of Mw 7.8 and 7.7 struck southeastern Turkey nine hours apart on the East Anatolian Fault, rupturing over 500 km of surface in the deadliest earthquake sequence since the 2010 Haiti disaster. The event killed more than 53,000 people, destroyed or damaged nearly 300,000 buildings, and generated a USD 91 billion economic loss — with only a fraction insured. Its defining lesson is not seismological: it is about what happens when a country with a modern, detailed earthquake building code allows that code to be systematically circumvented for decades.
The Kahramanmaraş earthquake doublet of 6 February 2023 is the most important recent case study in the catastrophic consequences of the gap between building code design and building code enforcement. Turkey possessed, at the time of the earthquake, a modern, detailed, and internationally benchmarked seismic building code — one that, had it been consistently applied and enforced, would have substantially reduced the death toll and building collapse rate. A new independent field investigation found that despite established technical know-how, state-of-the-art building codes, and rigorous building regulations, a drive for profit pushed all players within the construction industry to take shortcuts, with building stock primarily made of reinforced concrete structures being the main cause of casualties.
The seismology was extraordinary — a doublet of two Mw 7.7+ earthquakes within nine hours on an interacting fault network, producing more than 500 km of combined surface rupture and rupture dynamics that surprised seismologists. The Kahramanmaraş earthquake sequence ruptured at least six faults, including a large portion of the East Anatolian Fault, following unexpected rupture trajectories that included delayed backward branching, statically and dynamically aided triggering, and a combination of subshear and supershear rupture episodes. But the geophysical complexity of the doublet, while scientifically remarkable, explains only a fraction of the death toll. The dominant driver of fatalities was the systematic failure of multi-storey reinforced concrete buildings that should, under a properly enforced code, have survived.
The earthquake sequence became the deadliest global disaster since 2010, the 11th deadliest overall, and by far the costliest and deadliest in Turkey's modern history. The earthquakes in Turkey and Syria were responsible for nearly half of the total global economic losses in the first half of 2023, estimated at USD 91 billion. Against insured losses of approximately USD 5–6 billion, the protection gap exceeded 93% — an extraordinary figure for a country that has operated a compulsory residential earthquake insurance pool since 2000.
The East Anatolian Fault (EAF) is one of the world's major active transform faults — a left-lateral strike-slip system that accommodates the westward escape of the Anatolian microplate as it is squeezed between the converging Eurasian and Arabian plates. The EAF extends approximately 550 km from the Karlıova Triple Junction in the northeast to the Dead Sea Fault in the southwest, accommodating 6–10 mm per year of relative plate motion.
The fault was well-known and well-mapped. Its seismic hazard was documented in Turkey's national probabilistic seismic hazard assessment. The question was not whether the EAF could generate a large earthquake — it was when, and the answer, based on the fault's slip rate and the strain accumulated since its last major rupture, pointed clearly to a significant seismic deficit. The unexpectedly-large slip on some EAF segments suggests a supercycle lasting ≥900 years — meaning that the February 2023 rupture may have released strain accumulated over nearly a millennium on sections of the fault that had not produced major earthquakes in the modern instrumental record.
The Mw 7.8 mainshock initiated not on the main East Anatolian Fault itself but on a smaller splay fault — the Narlı Fault — before propagating onto the main EAF and rupturing bilaterally northeast and southwest. The analysis reveals three subshear slip episodes during the initial Mw 7.8 earthquake with a delayed rupture initiation to the southwest. This initiation on a secondary fault before cascading to the main system was unexpected and highlights the complexity of rupture dynamics in networks of interacting faults.
The Mw 7.7 aftershock — occurring nine hours later on the separate Çardak-Sürgü Fault approximately 90 km to the northwest — was itself seismologically remarkable. The Mw 7.7 event demonstrated a bilateral supershear rupture of approximately 5.0–6.0 km/s over an 80 km length. Supershear rupture — where the rupture front propagates faster than the shear wave velocity of the surrounding rock — produces a distinctive and particularly damaging pattern of ground motion, concentrating seismic energy in a Mach cone ahead of the rupture front rather than radiating it omnidirectionally. Supershear rupture is relatively rare and was not anticipated for this event; its occurrence on the Mw 7.7 earthquake contributed to the severe near-fault ground motions that exceeded model predictions in several affected cities.
Near-fault damage was significantly underestimated by seismic hazard maps based on empirical ground motion models, as observed ground motion intensities significantly exceeded predictions at distances less than 10 km from the fault. This exceedance was not simply a matter of the earthquake being larger than expected — it reflected genuine deficiencies in the empirical ground motion models used to construct Turkey's seismic hazard maps, which were not well-calibrated for the extreme near-fault environment of a major strike-slip rupture. For structures within the immediate rupture corridor, the ground motion was beyond what any hazard model had predicted for the location.
The supershear rupture of the Mw 7.7 event compounded this problem. Empirical ground motion prediction equations (GMPEs) are calibrated primarily to subshear ruptures — which constitute the vast majority of the historical database. The Mach wave effects of supershear propagation produce a distinctly different and more directionally concentrated radiation pattern that existing GMPEs do not accurately capture, contributing to systematic underestimation of ground motion in the supershear rupture direction.
The death toll of 53,000 people from a pair of Mw 7.7–7.8 earthquakes in a country with a modern seismic building code demands an explanation. The answer is not that the earthquakes were unexpectedly powerful — ground motions in most of the affected cities, while severe, were within ranges that well-engineered buildings should have survived. The answer is that a large proportion of the building stock, particularly the multi-storey reinforced concrete residential buildings that housed most of the victims, was not well-engineered. It was constructed to a substandard that had been permitted, normalised, and in some cases officially sanctioned through amnesty programmes, for decades.
"The disaster in Turkey and Syria is very obviously the result of poor construction. Although the disaster produced, in fact, stronger shaking than seismic intensity maps suggested, it should not have caused 5,500 large buildings to collapse."
— Prof. David Alexander, UCL Department of Risk and Disaster Reduction, February 2023Turkey has progressively updated its seismic building code since the catastrophic 1999 Marmara earthquakes, which killed more than 17,000 people and triggered comprehensive legislative and technical reforms. The current Turkish Building Earthquake Code (TBEC-2018) is a modern, internationally benchmarked standard that would, if consistently applied, produce buildings capable of surviving major earthquake shaking with limited life-safety risk. The code specifies detailed requirements for concrete strength, rebar detailing, column-to-beam connections, soft storey prevention, and foundation engineering for different site conditions and seismic zones.
The problem was that compliance with this code was neither universal nor effectively enforced. Several systemic failures combined to create a building stock that bore little relationship to the code's requirements:
Open ground floor (commercial/parking) dramatically weaker than upper floors. Under lateral seismic loading, ground storey collapses while upper floors pancake downward as a rigid block. Preventable through proper column sizing and moment-resisting frame design.
Partial-height infill walls or window openings constrain columns to act over a reduced height, concentrating shear demand. The constrained column fails in shear — a brittle failure mode — before the frame can develop ductile flexural resistance.
Insufficient stirrups (transverse rebar) in columns reduces confinement of concrete core under axial and seismic loading. Columns lose core integrity and buckle axially. Widely observed across collapsed buildings — often the result of cost-cutting during construction.
Concrete with insufficient cement content, poor water-cement ratio, or inappropriate aggregate lacks the compressive strength required by the code. Under seismic loading, columns and shear walls crush rather than deform, eliminating the ductility that earthquake engineering relies on for survival.
Progressive floor-by-floor collapse when floor-to-column connections fail under seismic loading. Upper floors fall onto lower floors with catastrophic force, compressing the entire building height into a fraction of its original dimension. The most lethal failure mode — survival rate extremely low.
Many of the collapsed buildings were built relatively recently, with many total collapses of newly built multi-story residential buildings. The failure of recently constructed buildings — which should have been built under the most current code — demonstrates that code provisions alone, without enforcement, provide no safety guarantee.
The Kahramanmaraş disaster was not Turkey's first experience of catastrophic building failure in a major earthquake. The 1999 Mw 7.6 Kocaeli (Marmara) earthquake killed 17,000 people through nearly identical building failure modes — soft storey collapses, pancake failures, substandard concrete. That event produced the same engineering analyses, the same condemnations of the construction industry, and the same political commitments to reform that followed February 2023. The TCIP compulsory insurance system was created directly in response to 1999. The building code was substantially updated. Enforcement was promised.
Twenty-four years later, in a different part of Turkey but with the same construction practices, the same failure modes killed three times as many people. The persistence of the same vulnerability for nearly a quarter century, through two major seismic events and multiple reform commitments, illustrates a fundamental challenge in earthquake risk management: the governance and political economy of building code enforcement are often more determinative of outcomes than the technical content of the code itself.
The earthquake's impacts extended across the border into northern Syria — a country already devastated by more than a decade of civil conflict. The Syrian population in the affected area was among the most vulnerable in the world: a fragile population, including migrants, internally displaced persons, and communities whose infrastructure had already been severely degraded by years of conflict. The pre-existing humanitarian crisis in Syria meant that post-earthquake emergency response was severely hampered by damaged roads and infrastructure, absence of functioning government systems in many areas, ongoing security constraints, and the near-total absence of the insurance and financial recovery mechanisms available in Turkey.
The cross-border nature of the disaster created significant operational and analytical complexity. Cat model loss estimates focused on Turkey, where insurance exposure was concentrated. Syria, despite suffering severe physical destruction and a significant share of the fatalities, generated almost no insured losses — the country essentially had no functioning property insurance market in the affected region. This produced a particularly extreme protection gap: economic losses in Syria were substantial, insured losses were effectively zero.
Turkey established the Turkish Catastrophe Insurance Pool (TCIP, known by its Turkish acronym DASK) following the 1999 Marmara earthquake. The system provides compulsory earthquake insurance for all registered residential buildings in Turkey — a pioneering policy response to the catastrophic uninsured losses of 1999. TCIP is heavily reinsured in the international market, with Munich Re and Swiss Re among the leading capacity providers, providing protection above a retention threshold of approximately USD 300 million.
In principle, TCIP represented exactly the kind of broad-based mandatory insurance system that disaster risk finance experts advocate as a solution to the earthquake protection gap. In practice, its performance in the 2023 earthquake revealed several structural limitations:
The Turkish government estimated the economic impact at approximately TRY 2 trillion (USD 105 billion). The USD 91 billion figure includes total physical damage as estimated jointly by the Government of Turkey, the World Bank, the United Nations and the European Union. Against this, only around 20% of the USD 25 billion economic losses that RMS estimates will result from the earthquake in Turkey will be covered by the insurance sector, highlighting the vast protection gap that exists in the country.
The Insurance Association of Turkey (TSB) estimated total losses to the private insurance sector at TRY 76 billion (USD 4.0 billion). The public insurance scheme facilitated by the TCIP received nearly 600,000 claims, with total payments expected to reach TRY 29.5 billion (USD 1.6 billion). Combined, total insured losses of approximately USD 5–6 billion against economic losses exceeding USD 91 billion represents a protection gap of approximately 93–94% — among the largest recorded for a major earthquake in a middle-income country with a functioning insurance market.
The East Anatolian Fault was a known hazard source in all commercial cat models for Turkey. The probabilistic seismic hazard model for the region correctly identified it as capable of generating Mw 7+ earthquakes — the event was not a blind fault surprise in the manner of Canterbury. The modelling failures were more subtle:
The most significant cat model failure was in the vulnerability domain. Vulnerability functions for Turkish reinforced concrete construction — calibrated to the performance of buildings that were assumed to comply with the building code — substantially underestimated the actual damage rates observed. Despite relatively strict and modern building codes currently in place in Turkey, structural integrity and performance varied widely in the affected regions. Many of the collapsed buildings were built relatively recently.
Vulnerability functions calibrated to code-compliant construction will systematically overestimate structural performance in environments where compliance is low. This is not simply a matter of using the wrong fragility curve for a given construction type — it is a fundamental challenge of modelling in environments where the legal definition of a building type (reinforced concrete frame, 5-storey residential) does not reliably indicate the actual structural quality of buildings bearing that description. The gap between what a building is called and what it actually is, in terms of structural performance, was the dominant source of cat model underestimation in Kahramanmaraş.
The mainshock initiates on the Narlı splay fault before cascading onto the East Anatolian Fault and rupturing bilaterally over 310 km. The 04:17 local time means most residents are asleep — in buildings whose quality will prove catastrophic. Shaking is felt from Cyprus to Iraq. The first reports of massive building collapses emerge within minutes of the rupture.
A major aftershock strikes 11 minutes after the mainshock. Buildings already weakened by the mainshock collapse completely. Search and rescue teams have not yet been deployed. The aftershock delays the organisation of emergency response and traps additional survivors under rubble.
The Çardak-Sürgü Fault ruptures 90 km to the northwest of the morning's epicentre, triggered by stress changes from the Mw 7.8. The rupture propagates at supershear velocities along 140 km of previously unruptured fault. Buildings in Elbistan, already damaged from the morning event, collapse completely. International search and rescue teams arriving in Turkey are now operating across two simultaneous major disaster zones.
Temperatures across the affected region fall below freezing at night. Survivors trapped in rubble face cold as well as injury. International urban search and rescue teams from more than 90 countries arrive. The logistics of operating across 11 provinces simultaneously strain all available resources. Survivors are pulled from rubble days after the earthquakes — but thousands more do not survive. The confirmed death toll climbs daily.
President Erdoğan declares a three-month state of emergency across the 11 affected provinces. The government announces emergency housing programmes, pledging to rebuild within one year — a timeline that engineering experts widely characterise as unrealistic for the scale of damage. AFAD (Disaster and Emergency Management Authority) coordinates the national response.
Engineering surveys document the full extent of the building stock performance. Forensic investigations of collapsed buildings reveal systematic structural deficiencies. Turkish authorities arrest more than 200 construction contractors, site supervisors, and building officials on charges related to substandard construction. The arrests generate significant media coverage but also raise questions about systemic accountability — whether individual prosecutions can address a construction culture problem of this scale.
The Turkish government launches a major reconstruction programme, commissioning prefabricated housing units and temporary shelter for displaced residents. The complexity of rebuilding 11 provinces simultaneously — with code-compliant construction, adequate geotechnical assessment, and proper quality control — proves challenging. The question of whether the reconstruction will replicate the vulnerabilities of the destroyed stock, or genuinely implement the building code, remains the defining challenge of the recovery.
| Factor | Kahramanmaraş 2023 (Turkey) | Canterbury 2010–2011 (New Zealand) |
|---|---|---|
| Earthquake magnitude | Mw 7.8 + Mw 7.7 doublet on known EAF | Mw 7.1 + Mw 6.2 sequence on blind faults |
| Primary cat model failure | Vulnerability functions not calibrated to non-compliant construction; near-fault GMPE exceedance | Blind fault source absent from hazard model; liquefaction not modelled |
| Dominant damage mechanism | Building collapse from structural non-compliance | Liquefaction and lateral spreading (suburban); building collapse (CBD) |
| Building code status | Modern, internationally benchmarked code existed — widely circumvented | Modern code — generally complied with; pre-code stock vulnerable |
| Fatality driver | Pancake and soft-storey collapse of occupied multi-storey RC buildings | CTV building collapse (115 deaths); tsunami-equivalent residential losses from liquefaction |
| Protection gap | ~93–94% — TCIP covers ~52% of eligible structures; underinsurance within covered | ~23% — EQC system with high residential penetration |
| Government response | State of emergency; reconstruction pledges; contractor arrests; TCIP reform discussions | Red Zone buyouts; CERA rebuild; EQC reform; geotechnical zoning |
| Key modelling legacy | Non-compliance vulnerability adjustment; near-fault supershear GMPE revision; multi-fault cascading rupture | Blind fault inclusion; liquefaction explicit modelling; aftershock sequence frameworks |
The Kahramanmaraş sequence focused global attention not only on the immediate disaster but on the unresolved seismic risk in Istanbul — a city of 16 million people built largely on similarly non-compliant reinforced concrete construction, sitting above the North Anatolian Fault which has been seismically loaded since the 1999 Marmara earthquake. With only 20% national penetration and approximately one-third of all TCIP policies in Istanbul, the system exhibits signs of adverse selection, inadequate premium structure, and insufficient funding. The annualized earthquake losses in Istanbul are between USD 140–300 million. Even with a raised deductible, the earthquake losses following a large Istanbul earthquake will be approximately USD 2.5 billion — somewhat above the current capacity of the TCIP.
The structural vulnerability exposed in southeastern Turkey — soft storey buildings, substandard concrete, illegal construction amnesties — is replicated at vast scale in Istanbul, where millions of buildings were constructed before the 1999 code reforms, many on soft alluvial and landfill sites with high liquefaction potential, in a city that has not experienced a major earthquake since 1766. The probability of a large earthquake on the North Anatolian Fault segment beneath Istanbul is estimated at approximately 50–70% over the next 30 years by USGS and Turkish seismologists. Kahramanmaraş demonstrated what that probability implies in human and financial terms if building vulnerability is not addressed before the next major earthquake arrives.
Kahramanmaraş established that vulnerability functions calibrated to code-compliant construction are inadequate for environments where compliance is low and variable. Post-event, model vendors have developed approaches to adjust vulnerability based on governance quality, construction sector regulation indices, and building permit enforcement indicators — moving beyond purely physical classification of building types.
The supershear rupture of the Mw 7.7 earthquake, producing ground motions that substantially exceeded empirical predictions, has driven investment in supershear-specific GMPEs and near-fault ground motion models. These are being incorporated into next-generation hazard models for strike-slip fault systems globally — including the North Anatolian Fault, the San Andreas Fault, and the Alpine Fault in New Zealand.
The stress-triggered triggering of the Mw 7.7 on a separate fault nine hours after the Mw 7.8 — itself initiated on a splay fault before cascading to the main EAF — demonstrated the importance of modelling fault interaction and cascading rupture in complex tectonic environments. Stochastic catalogs are being revised to incorporate multi-fault rupture scenarios on interacting fault systems, moving beyond single-fault rupture assumptions.
Turkish authorities are now striving to increase and expand future TCIP cover following the coverage gaps revealed by the 2023 event. Reforms under discussion include mandatory declaration verification of floor areas, integration with building registration systems to close the informal construction coverage gap, and revised premium structures to improve both penetration and adequacy.
Kahramanmaraş has accelerated the incorporation of governance and institutional quality indicators into cat model frameworks. Building code quality alone does not determine structural performance — it is the combination of code quality and enforcement quality that matters. Models that treat Turkish RC construction as a homogeneous class miss the variance in structural quality driven by differential enforcement across regions and time periods.
The Turkey-Syria doublet highlighted the challenge of modelling and responding to earthquakes whose physical impacts cross political boundaries with radically different insurance market structures. The extension of catastrophe modelling frameworks to cover cross-border events — where the insured fraction on one side of a border may be near zero — is a recognised gap that the 2023 event has brought into sharp focus.