The 2023 Kahramanmaraş Earthquake Sequence: a counterfactual contemplation
On 6 February 2023 a magnitude Mw7.8 and Mw7.6 earthquake doublet occurred in the East Anatolian Fault Zone resulting in catastrophic destruction and loss of life in southeast-central Turkey and northern Syria. Knowing what we know now, how might the events have played out differently?
Dr John Wardman
Chief Commercial Officer
This entry was co-authored by Dr John Wardman (Max Info) and Dr Mark Naylor (University of Edinburgh)
Three months on from the 6 February 2023 earthquakes which impacted Turkey and Syria and several well-written articles have been penned on what is known so far (see Dal Zilio and Ampuero, 2023; Stein et al., 2023; Toda et al., 2023; USGS et al., 2023, TERRA, 2023, among others, and look out for special journal issues covering the Kahramanmaraş sequence such as those from EERI, SSA, and Seismica). A review of these studies reveals some important findings, set out below, and in this blog we explore the downward counterfactual question of ‘what if’ the two events had occurred simultaneously.
Strongest quake in 80+ years: the 6 February 2023 Mw7.8 earthquake shook southern and central Turkey and northern and western Syria in what was the strongest quake to hit the region in >80 years. The Mw7.8 is believed to have nucleated on a secondary (splay) fault, the transfer of stress from which triggered a ~300 km northeast-southwest rupture of the East Anatolian Fault (EAF).
Approximately nine hours later, a Mw7.6 event struck ~100 kilometres north-northeast of the epicentre of the Mw7.8 quake, on the ~160 km-long Sürgü-Çardak Fault (SCF) which traces near to the EAF at its eastern most margins (see the below image). These events and their associated aftershocks are referred to as the Kahramanmaraş Earthquake Sequence.
Map showing the parts of south-central Turkey and north-western Syria which surround the East Anatolian Fault and Sürgü-Çardak Fault, both of which ruptured (shown in bold red) on 6 February 2023. Main population centres, active faults (shown in red, from Styron & Pagani, 2020), surface rupture observations (shown in blue, from Reitman et al., 2023) and approximate epicentres (orange stars) are shown for reference.
Earthquake doublet: The ~9hr time difference between the two quakes classifies them as a doublet, defined by Kagan and Jackson (1999) as pairs of large earthquakes whose centroids are closer than their rupture size and whose interevent time is shorter than the recurrence time inferred from the plate motion. Doublets are relatively rare – accounting for ~20% of all historical large earthquakes – but are interesting phenomena in that they suggest large quakes, like small ones, cluster in both time and space.
The Mw7.8 triggered the Mw7.6: There is ongoing debate around what distinguishes an aftershock from a triggered event. Since the two Kahramanmaraș events ruptured two distinctive faults (as opposed to two segments of one fault system), the Mw7.6 quake could arguably be classified as a secondary main shock which was triggered by the Mw7.8 (together comprising a doublet). Bath’s Law states that, on average, the biggest aftershock is ~1.2 magnitudes smaller than the main shock. If we consider the Mw7.6 an aftershock to the Mw7.8 mainshock, then this sequence is unusual.
Ignoring the debate around what distinguishes an aftershock from a triggered event, both Toda et al. (2023) and the USGS (2023) infer that stress transferred by the Mw7.8 quake along the EAF promoted the Mw7.6 event on the SCF (see image below).
Map from USGS (2023) showing Coulomb stress calculations following the Mw7.8 and Mw7.6 quakes on the EAF and SCF, respectively. Stress imparted by the Mw7.8 quake brought the SCF closer to failure (i.e. increased the hazard), and the final stress calculations suggest increased strain (represented by warmer colours in the above map) on major fault systems such as the northern segment of the Dead Sea Fault (DSF), located south of the EAF.
Multi-segment rupturing: Fault segmentation is a principle based on the observation that fault zones, particularly long ones like the EAF, do not rupture along their entire length during a single earthquake. Rather, one or more segments of the fault zone may rupture simultaneously. Historical observations suggest earthquakes in the EAF zone display a high variability in magnitude, involving moderate (Mw6+) and sporadic large (Mw7+) earthquakes, which typically rupture – partially or completely – different segments of the East Anatolian Fault Zone separately (Dal Zilio and Ampuero, 2023).
The earthquake doublet on 6 February 2023 ruptured different segments on both the EAF and SCFs, producing two complex sequences which led to moment magnitudes which were higher than if any one segment ruptured on its own. To explore this, we can ask the question: “What difference would it make to the estimate of magnitude if two or three similar sized segments rupture together?" The answer comes from the equations that define the seismic moment:
and the definition of the moment magnitude:
where μis the shear modulus of the rocks involved in the earthquake, u is the average slip, and S is the rupture area.
Everything else being equal, if two segments of the same area rupture together in a single earthquake, the magnitude of the earthquake will be 0.2 Mw higher than if either segment ruptured on its own. If three similar sized segments rupture together the compound event will be 0.32 Mw higher, and so on.
Implications for earthquake catastrophe modelling
Looking at the active faults delineated in the above figures, it is safe to assume that faults are not randomly positioned, and their geometries follow similar patterns of orientation and location. The structure of fault networks immediately implies a degree of cooperative behaviour and a degree of connectivity between faults which determines whether an earthquake ‘jumps’ across multiple fault segments or multiple independent faults, or whether the earthquake remains along a single segment.
In complex fault zones such as those found in the Kahramanmaraş region, multiple seemingly disconnected faults can rupture at once, increasing the chance of societal impacts. Recent earthquakes such as those during the 2019 Ridgecrest sequence in California ruptured in this way, and the below image from Stirling et al. (2017) shows the 21 different faults which ruptured simultaneously along a ~180 km long zone during the 2016 Mw7.8 Kaikōura quake in New Zealand. Such extensive and complex faulting had not been witnessed in recorded history, and the implications for seismic hazard assessment were vast since modern modelling techniques, to this day, do not commonly allow for multi-fault rupture scenarios.
Surface fault ruptures (red lines) of the Mw7.8 2016 Kaikōura earthquake. Grey lines show the active faults, the geographic name assigned to each fault rupture zone is also shown (from Stirling et al., 2017)
The latest seismic hazard map for Turkey (released in 2018, shown below) suggests that the likelihood of these faults rupturing and generating strong ground shaking was well captured (see the below figure, where red colours indicate a high probability of experiencing damaging ground shaking in the next 475 years). Although the Mw7.8 and Mw7.6 quakes represent distinct ruptures, insights from recent large earthquakes such as those mentioned above suggest it is not unreasonable to consider that both faults in the Kahramanmaraş sequence could have ruptured simultaneously. Counterfactual analysis considers alternative realities of past events, and is useful in identifying rare events which might not currently be recognised as risks, but which are nonetheless satisfactorily "real” (Lin et al., 2020). If we explore the counterfactual scenario of the two Kahramanmaraş faults rupturing at the same time, then the combined moment magnitude of the event would have been moderately increased to Mw7.91. Would the resulting ground shaking have been different to the events occurring separately? Possibly, as the wavefields interact over the affected area. Would a contemporary Turkey earthquake catastrophe model have considered this Mw7.9 multi-fault propagation a possibility? Maybe, but unlikely.
Seismic hazard map of Turkey, a 2018 product by the Turkey Ministry of Interior Disaster and Emergency Management Presidency. Approximate epicentres of the February 2023 M7+ quakes are shown as white stars and suggest the latest iteration of the hazard model adequately identified these areas as high hazard (i.e. a high probability of experiencing damaging ground shaking over a designated time period, 475 years in this example).
Downward counterfactual thought is increasingly being used by the re/insurance industry to quantify the potential impacts of previously unseen catastrophic events (Philp et al., 2022), and uncertainty arising from the lack of multi segment and multi fault rupture scenarios in most contemporary earthquake cat models represents a major blind spot for the risk transfer industry. Even the most sophisticated hazard models such as the Uniform California Earthquake Rupture Forecast Version 3 (UCERF3) have only recently begun to incorporate multi-segment and multi-fault scenarios in their simulations. Lee et al. (2022) found that long ruptures (i.e. >500 km) achieved through multi fault rupture scenarios generally produced larger aggregate loss estimates – particularly at high return periods which are meaningful for tail risk – for a geographically dispersed portfolio simply due to the wider areas that are affected by strong ground shaking.
Every new earthquake sequence tells us something we didn’t know before and reinforces things that we already do. In this case, we are reminded that fault ruptures are complex physical phenomena which are challenging to reproduce in a modelling framework.
References
Dal Zilio, L., & Ampuero, J. P. (2023). Earthquake doublet in Turkey and Syria. Communications Earth & Environment, 4(1), 71.
Kagan, Y. Y., & Jackson, D. D. (1999). Worldwide doublets of large shallow earthquakes. Bulletin of the Seismological Society of America, 89(5), 1147-1155.
Lee, Y., Hui, Z., Daneshvaran, S., Sedaghati, F., & Graf, W. P. (2022). Impacts on catastrophe risk assessments from multi-segment and multi-fault ruptures in the UCERF3 model. Earthquake Spectra, 38(1), 128-151.
Lin, Y. C., Jenkins, S. F., Chow, J. R., Biass, S., Woo, G., & Lallemant, D. (2020). Modeling downward counterfactual events: Unrealized disasters and why they matter. Frontiers in Earth Science, 8, 575048.
Philp, T. J., Champion, A. J., Hodges, K. I., Pigott, C., MacFarlane, A., Wragg, G., & Zhao, S. (2022). Identifying limitations when deriving probabilistic views of North Atlantic hurricane hazard from counterfactual ensemble NWP re-forecasts. In Hurricane Risk in a Changing Climate (pp. 233-254). Cham: Springer International Publishing.
Reitman, Nadine G, Richard W. Briggs, William D. Barnhart, Jessica A. Thompson Jobe, Christopher B. DuRoss, Alexandra E. Hatem, Ryan D. Gold, John D. Mejstrik, and Sinan Akçiz (2023) Preliminary fault rupture mapping of the 2023 M7.8 and M7.5 Türkiye Earthquakes. DOI: https://doi.org/10.5066/P985I7U2
Stein, R.S.., Toda, S., Özbakir, A. D., Sevilgen, V., Gonzalez-Huizar, H., Lotto, G., Sevilgen, S. (2023). Interactions, stress changes, mysteries, and partial forecasts of the 2023 Kahramanmaraş, Türkiye, earthquakes, Temblor, http://doi.org/10.32858/temblor.299
Stirling, M. W., Litchfield, N. J., Villamor, P., Van Dissen, R. J., Nicol, A., Pettinga, J., ... & Zinke, R. (2017). The Mw7. 8 2016 Kaikōura earthquake: Surface fault rupture and seismic hazard context. Bulletin of the New Zealand Society for Earthquake Engineering, 50(2), 73-84.
Styron, R., & Pagani, M. (2020). The GEM global active faults database. Earthquake Spectra, 36(1_suppl), 160-180.
Toda, S., Stein, R. S., Özbakir, A. D., Gonzalez-Huizar, H., Sevilgen, V., Lotto, G., and Sevilgen, S., 2023, Stress change calculations provide clues to aftershocks in 2023 Türkiye earthquakes, Temblor, http://doi.org/10.32858/temblor.295
