The February 6, 2023 Kahramanmaraş earthquake sequence in Turkey and Syria included a magnitude 7.8 main shock on the East Anatolian Fault and a subsequent magnitude 7.5 triggered event on the neighbouring Sürgü-Çardak fault (see Figures 1a,b). This event tragically resulted in more than 53,000 deaths, heavy damage or collapse of more than 230,000 buildings, and the displacement of more than 2 million people across the affected region as of May 6, 2023, according to the Earthquake Engineering Research Institute’s (EERI) Learning from Earthquakes program.
Though insured losses account for only a fraction of total economic losses, the (re)insurance industry locally and internationally contributes to the rebuilding effort.
On February 6, the same day as the events, Verisk released Similar Stochastic Events (SSEs), denoted by the bold circles in Figure 1d. Insured loss estimates from those same-day SSEs were similar to those released a week later from Verisk’s observations-informed ALERT custom event set (Figures 1e, f). Further, both SSE and ALERT modeled insured estimates were consistent with external estimates—on the order of low-single-digit billions of U.S. dollars.
Verisk’s analysis yielded five key findings:
- Similar stochastic events exist in Verisk’s catalog. The main event and wider sequence occurred around the well-known and well-studied part of the East Anatolian Fault, with a significant seismic gap captured by Verisk’s time-dependent model. Such events should already be considered in existing catastrophe models.
Much larger events are also included in this region, with multi-segment rupture scenarios of magnitude 7.9-8.2.
- Initial Similar Stochastic Events, ALERT simulated events, and published estimates show good agreement. Due to a 72-hours clause present in some Turkish policies, Verisk combined the shaking footprints of these events when generating the ALERT event set. Verisk’s modeled losses at the time of publication are in the single-digit billions of U.S. dollars, similar to most industry loss estimates. Important considerations are the difficulty of estimating losses in Turkey with recent high inflation, the length of time to rebuild, and the exchange rate between the U.S. dollar and the Turkish lira, which can further complicate these estimates.
- The countrywide insured occurrence return period of loss is around 50-60 years and shouldn’t be considered as being in the tail of the loss distribution. From a regional point of view, though, these events are quite extreme, as seen in the bottom right of the graph below with return periods of hundreds to thousands of years for individual CRESTA zones near the epicenter. Nevertheless, even for this region of Turkey, the figures below reveal the potential for larger loss-causing events.
- Most of the insured 100-year return period and greater events (100 year TVaR) affect Istanbul, and the industry shouldn’t lose sight of this; insured losses from future events can be much greater.
- Modeled damage compares well with observations, and Verisk already explicitly accounts for lack of code adoption and enforcement by increasing the theoretical vulnerability class for each region to make structures weaker than assumed in the code. Reliable damage surveys and reports (e.g. EERI1, Earthquake Engineering Field Investigation Team2, World Bank3) were published relatively quickly, providing great data to enable model vs. observed comparisons as below.
Model damage ratio at regional scale compares well with loss ratios inferred from observations
As shown, Verisk’s model demonstrated good performance. Ground motion models implemented within Verisk’s EU earthquake model performed broadly well in the areas of main damage. The large number of recordings from Turkey’s well-established seismic network may improve future intensity and/or uncertainty representations.
Damage data is also expected to help improve buildings’ seismic performance, with design hazard values potentially increasing for mid-high-rise buildings. Lack of building code adoption and enforcement has been widely reported and is expected to improve. Verisk believes the insurance industry that adversely bears this risk should support better code adoption and enforcement both for existing and future buildings—especially with lives at stake.
Now that this event has happened, the U.S. Geological Survey believes there is reduced risk of similarly large earthquakes (greater than magnitude 7.5) on the Kahramanmaras segment of the East Anatolian Fault. However, there are elevated seismicity areas in other locations, including the Northern Dead Sea Fault Segment (primarily impacting Syria and Turkey) and other more northerly segments of the East Anatolian Fault Zone.
The February 6 event will have important implications across a broad area. It should highlight that large earthquake risk in Turkey is very real, can produce significant losses, and should encourage the industry to resist urges to reduce modelled risk estimates without thorough consideration.
- February 6, 2023 Türkiye Earthquakes: Report on Geoscience and Engineering Impacts, Earthquake Engineering Research Institute, Geotechnical Extreme Event Reconnaissance Association (USA), Earthquake Engineering Association, Earthquake Engineering Foundation of Türkiye, May 6, 2023, <https://learningfromearthquakes.org/images/2023_02_06_nurdagi_turkey/GEER_2023_Turkey_Earthquake_FullReport_ReducedSize.pdf >, accessed on June 16, 2023.
- EEFIT Mission: February 2023 Turkey earthquake, The Institution of Structural Engineers, March 29, 2023, <https://www.istructe.org/resources/case-study/eefit-mission-to-6-february-2023-kahramanmaras-tur/ >, accessed on June 16, 2023.
- Global Rapid Post-Disaster Damage Estimation (GRADE) Report, February 6, 2023 Kahramanmaraş Earthquakes, Türkiye Report, The World Bank and the Global Facility for Disaster Reduction and Recovery, February 20, 2023, <https://documents1.worldbank.org/curated/en/099022723021250141/pdf/P1788430aeb62f08009b2302bd4074030fb.pdf >, accessed on June 16, 2023.