the mw6.2 western Marmara sea earthquake of 23 April 2025: a preliminary assessment of ground motion, building vulnerability, societal response and Istanbul’s readiness for the anticipated major event

<jats:p>The Mw 6.2 Western Marmara Sea earthquake of 23 April 2025 ruptured the Kumburgaz segment of the Main Marmara Fault at shallow depth, generating peak ground accelerations of up to 0.22 g (NS component of Station 3415, Küçükçekmece) – approximately half of the design code levels at that location. One person died, 359 injuries were recorded across the Marmara region, and 4,295 buildings were classified as slightly damaged in Ministry assessments (Kurum, 2025a). Major lifelines were unaffected. Based on the level of ground shaking and losses, it was classed as a moderate event. This report argues that the significance of the 2025 event lies not only in its direct physical and societal impacts – which were, in absolute terms, modest – but in what those impacts revealed about Istanbul's preparedness for the substantially larger event that the seismological evidence indicates is pending on the Princes' Islands segment of the Main Marmara Fault. The earthquake ruptured a segment that had accumulated approximately 260 years of tectonic strain and released approximately 8% of it – equivalent to roughly 0.3 m of slip – leaving approximately 3.4 m of accumulated deficit unreleased on the Kumburgaz segment and transferring stress toward the locked Princes' Islands segment south of Istanbul's historic peninsula – the identified source of an anticipated major Marmara earthquake in the future. Aftershock migration tracked eastward toward this locked segment within 24 hours of the mainshock (Eken et al., 2025). Martínez-Garzón et al. (2025) place the 2025 event within a fifteen-year sequence of eastward-migrating moderate ruptures converging on this segment. The 2025 earthquake has not relieved the seismic hazard facing Istanbul; the available evidence indicates it has incrementally increased the probability of a major rupture on the Princes' Islands segment. Ground motion, structural assessment and tsunami The 212-station AFAD-TADAS and KOERI strong-motion network produced one of the densest sets of urban ground motion records in the world for a moderate earthquake. Chapter 2 presents a unified analysis of this dataset, including a comparative evaluation of the predictive performance of four established ground motion prediction equations applicable to shallow crustal earthquakes in the Europe and Middle East region – BSSA14, ASB14, KAAH15 and CB14. Overall, the dataset is characterized by high peak ground motion ordinates on the European side of Istanbul and all ground motion parameters typically fall within the median ±2σ range for all evaluated models. Despite this, the median PGA estimations from KAAH15 exhibit consistent underestimation corroborating findings from previous studies. European-side stations recorded systematically higher peak ground accelerations and spectral values than Asian-side stations at comparable epicentral distances. This observation is independently corroborated by the official building damage report geography (80% of reports from European-side districts) and by the 4,302 EMSC crowdsourced felt intensity reports (mean intensity 4.42 on the European side versus 3.97 on the Asian side). The prevalence of soft soil conditions in these districts (predominantly ZC and ZD site classes) is consistent with this pattern, though a formal site amplification analysis lies beyond the scope of this report. The building stock experienced ground shaking well below the Turkish code design-level loading: recorded ground motions remained below DD-2 design levels at all stations, with ratios of recorded to design PGA ranging from approximately 20% at Avcılar (Station 3428) to approximately 52% at Küçükçekmece (Station 3415). Aydoğdu and İlki (2026) demonstrate that the limited damage observed in vulnerable pre-1999 buildings reflects the short 13-second ground motion duration rather than structural adequacy, and that buildings of this type showing light damage under the 2025 record would not survive a longer-duration record of comparable PGA. Station 3428 (Avcılar) recorded both the largest long-period spectral accelerations and the longest strong motion duration (D5-95 up to 40.68 s) of the four stations analysed in Chapter 5 – a combination consistent with known basin resonance effects in Avcılar and consistent with its anomalously high damage rates in the 1999 earthquakes. ESDOF analyses (Chapter 5) confirm that inelastic demand was already reached in vulnerable RC frame typologies in Küçükçekmece, Bakırköy, Avcılar and Silivri under this sub-design loading. These districts correspond closely to those repeatedly identified as highest-risk in Istanbul scenario loss studies since 2002 – confirming, under real-world moderate loading, a spatial vulnerability pattern that engineering analysis had already predicted. The earthquake also generated a minor tsunami: a maximum wave height of approximately 6 cm was recorded at Erdek, producing no damage, but confirming that the fault is capable of co-seismic seafloor displacement. Worst-case scenario models (Hébert et al., 2005) estimate waves of 3–4 m at the European coastline under a major rupture. The same European-side districts that recorded the highest ground motion ordinates and damage report concentrations in 2025 face a compounded hazard in the major event. Engineering systems Following the 1999 Marmara earthquake, Istanbul made sustained investments in earthquake engineering infrastructure: the Istanbul Earthquake Rapid Response and Early Warning System (IERREWS), the IGDAŞ natural gas automatic shut-off network (IGRAS), and a programme of seismic isolation for critical public buildings, with mandatory requirements for hospitals with capacities exceeding 100 beds enacted in 2013. In the 2025 earthquake, these systems performed largely as intended: IERREWS and IGRAS both triggered correctly, and eight base-isolated hospitals remained fully operational throughout the event and its aftershock sequence. These outcomes represent genuine and measurable returns on the post-1999 investment programme and demonstrate that targeted engineering intervention in critical infrastructure can deliver reliable performance under moderate seismic loading. Building stock and hidden vulnerabilities Chapter 3 documents why Istanbul's building stock carries the risk it does, and why that risk cannot be read from the regulatory record alone. Approximately 70% of Istanbul's buildings predate the post-1999 regulatory transition; a substantial proportion of the pre-1975 stock was constructed under codes with no ductility requirements, using low concrete grades, smooth reinforcing bars and inadequate stirrup detailing. Scenario loss studies conducted since 2002 – with fatality estimates ranging from 14,000 to 87,000 depending on scenario and methodology – consistently identify the same seven European-side districts as highest-risk: Fatih, Küçükçekmece, Bağcılar, Bahçelievler, Zeytinburnu, Avcılar and Esenyurt. The 2025 damage report geography confirms this pattern under real moderate loading. Hidden vulnerabilities compound the regulatory picture. Approximately 317,000 buildings in Istanbul have been retroactively legalised through construction amnesties (imar barışı), of which roughly 90,000 are considered at serious collapse risk. The condition of this older stock was made concrete by the 2025 inspections: post-earthquake surveys in Küçükçekmece documented RC columns containing marine shell fragments (midye kabukları) – concrete made with unwashed beach sand, porous, weak and entirely non-compliant with basic material standards – in buildings representative of the pre-regulatory era. This deficiency was detected under sub-design loading. Separately, the third-party building inspection system (yapı denetim), introduced after 1999 to govern new construction, is structurally compromised by its fee and contracting model – inspection firms are contracted and paid by building owners, with fee structures that incentivise cursory sign-off over rigorous site inspection. That 199 buildings in Avcılar confirmed as moderately damaged in the 1999 earthquakes were still standing and occupied on 23 April 2025 – their legally-mandated remediation deadline having passed in 2020 – illustrates how the governance failures documented in this chapter are current, not historical. Seismic risk mitigation and urban transformation Since 2012, Istanbul has been carrying out a large-scale urban transformation programme under Law No. 6306, officially aimed at replacing seismically deficient residential buildings. Chapter 4 examines whether the programme is working as intended. The regulatory framework provides two instruments: TBEC 2018 Chapter 15, a comprehensive performance-based assessment that can identify whether structural intervention could bring a building to adequate seismic performance; and the RBTE-2019, a simplified screening procedure that produces a binary risky/not-risky designation. In practice, the RBTE-2019 designation – which yields a risky outcome for virtually all pre-1999 buildings regardless of actual structural condition – is frequently treated as the basis for demolition decisions without triggering the more rigorous TBEC 2018 assessment that should follow. Only around 10% of designated buildings are structurally strengthened rather than demolished (Kayaalp, 2025). This bias is not only technically unjustified in many cases – structural intervention is feasible for a significant proportion of the vulnerable stock, at 20–40% of the embodied carbon of full replacement – it is being actively exploited: construction companies have systematically purchased a single apartment in pre-1999 buildings, commissioned a risk assessment confident of the designation outcome, and used the resulting classification to force demolition and secure the reconstruction contract and development rights uplift – a practice known colloquially as deprem korkutmacılığı (earthquake scaremongering). The financial incentive structure of Law 6306 subsidises demolish-and-rebuild but not structural intervention, leading to the prioritisation of areas of high land value over areas of highest seismic risk, and generates gentrification that progressively displaces lower-income households toward the urban periphery where seismic risk is concentrated and programme activity lowest. The 2025 earthquake has not resolved questions about the programme's replacement stock. Whether replacement buildings – built by the same industry, subject to the same inspection regime, and designed to the same codes that produced vulnerable post-1999 stock documented in the 2023 Kahramanmaraş sequence – are genuinely more resilient than the stock they replace remains unresolved. Following the 2025 event, the programme was extended through the Yarısı Bizden initiative, with 61,000 slightly damaged units declared eligible for support. At the current rate of programme delivery, the 600,000+ high-risk units across Istanbul will not be substantially reduced before the next moderate event, let alone before the major rupture. Societal response An estimated 100,000 people spent at least one night outdoors following the earthquake despite the absence of structural collapse in inhabited buildings. Intercity coach fares tripled within hours. Rental prices in stable-ground districts rose over 100% within the first day. The Fear of Earthquake Scale (FES) data collected by EMSC in near-real time – 186 respondents, mean score 22.60/35, 42% in the high-fear range – indicate elevated seismic fear levels consistent with a population carrying the accumulated psychological weight of decades of seismic risk awareness, amplified by the direct experience of the 2023 Kahramanmaraş sequence. This is the first application of the FES to a real-time post-earthquake dataset. Outdoor sheltering was spatially sorted: European-side residents were significantly more likely to remain outdoors than Asian-side residents, consistent with rational risk assessment of their buildings' structural quality. The FES data, together with the behavioural evidence, indicate that the scale of the societal response reflects not disproportionate panic but a calibrated population response to a widely recognised existential risk – a distinction with direct implications for emergency planning assumptions about behaviour in the major event. Institutional performance Four institutional frameworks nominally designed to protect Istanbul's population against earthquake were tested at moderate scale and found deficient – in each case replicating failures documented after previous events without substantive reform. The DASK compulsory earthquake insurance system paid out as little as 6,400 TL for multi-room damage. The deductible and coverage ceiling structures render the instrument near-ineffective for the light-damage category produced by moderate events. This is not a new finding: the same structural inadequacy was documented after the Elazığ (2020), İzmir (2020) and Düzce (2022) earthquakes without reform. In the 2023 Kahramanmaraş sequence, DASK paid approximately 147,000 TL for a completely destroyed dwelling and then sought to claw back 75,000 TL of that amount – illustrating the consequences of the same coverage structure at catastrophe scale. The housing market sorted by perceived seismic safety within hours, with stable-ground district prices rising over 100%. This accelerates the progressive concentration of lower-income households in the highest-risk districts, compounding physical and social vulnerability in the same spatial pattern that scenario loss studies have documented since 2002. The assembly area system failed operationally: Formally designated areas were blocked, encroached by construction sites, or simply absent. The average per-capita assembly area provision across Istanbul of 2.98 m² falls below the Sphere Project minimum of 3.5 m², and 419 of 496 post-1999 designated areas have been converted to commercial development. These failures were not caused by the earthquake: locked gates, construction encroachment and prior commercial conversion are pre-existing, knowable conditions that an operational audit would have identified. Mobile telecommunications saturation – network congestion preventing voice and SMS communication within minutes of the mainshock across European-side districts – has been documented after every significant Turkish earthquake since 1999 and remains unaddressed 26 years later. Official communication also failed: the Education Minister's statement on 23 April that "none of our schools have any issues" was followed within a week by the announcement that four schools had been evacuated on structural grounds. Compound spatial vulnerability The districts generating the highest building damage report concentrations in 2025 – Bahçelievler, Esenler, Bağcılar, Zeytinburnu, Avcılar, Esenyurt – are also those with the lowest per-capita assembly area provision, the highest proportions of pre-1999 RC building stock on soft alluvial ground, the highest rates of lower-income household concentration following seismic-risk-driven outmigration, the lowest urban transformation programme uptake, and – critically – the populations with the fewest resources to absorb uninsured repair costs when DASK fails them. This co-location reflects sixty years of urbanisation dynamics, regulatory non-enforcement, construction amnesties and housing market responses to seismic risk. It is not coincidental, and it will determine the spatial distribution of casualties and displacement in the major event. The preparedness gap The failures documented in this report are regulatory in nature. Assembly area legal protection and operational audit, DASK deductible and coverage reform, urban transformation programme reorientation directed toward the highest-risk districts, telecommunications resilience standards, and building inspection enforcement are all within the existing competence of the relevant authorities. They have been available as policy options since 1999. Their persistence across successive moderate events – Elazığ in 2020, İzmir in 2020, Düzce in 2022, Marmara in 2025 – without substantive resolution is a matter of record. This report provides a specific, documented and dated account of what was known and what remained unaddressed as of May 2026. The eight key messages below summarise the principal findings.</jats:p>