Complexities of the Turkey-Syria doublet earthquake sequence

What, where, and when

In the early morning of February 6^th, 2023, an M7.8 earthquake occurred in southeastern Türkiye near the northern border of Syria. The event initiated a complex sequence of aftershocks, including an M7.6 earthquake about 9 h later and 90 km to the north (Figures 1A and 1B). The earthquake sequence is also referred to as a strong doublet earthquake sequence. Aftershocks of the two strong earthquakes occurred along two separate branches of the East Anatolia Fault, with lengths of up to 300 km, and some aftershocks occurred in Syria (NEIC/USGS, 2023). The earthquake sequence caused significant losses to Türkiye and Syria, including over fifty thousand human lives and tens of billions of US dollars in damage to the economy, social infrastructure, and valuable historical sites,¹ along with abundant evidence of surface ruptures, ground deformation and liquefied soils. The earthquake sequence also triggered extensive subaerial landslides, with about 100 identified by NASA from Planet Lab’s high-resolution satellite imagery, including several in the valley near Sarıseki. Very strong ground motion was observed at station 3126 in Antakya during the M7.8 earthquakes (Figure 1D), with horizontal acceleration greater than 1 g (normal gravitational value).

Figure 1.

Seismo-tectonics of the 2023 earthquake sequence along with seismological and geodetic observations

(A) Hypocenter (stars) and centroid location (beach balls) of the M7.8 and M7.5 earthquakes. The beach balls show focal mechanisms of the two events. Active geological faults are displayed as dashed red lines, and purple dashed lines. The inset panel show active tectonics of a larger area.

(B) Location of earthquakes in the first 2 days after Feb 6^th along with the magnitude-time distribution of earthquakes in the inset.

(C) Ground displacements caused by the earthquakes as measured with point tracking methods.

(D) Fourier amplitude spectra of strong motion recording at station 3126 for the M7.8 event and response spectra (with 5% damping) in comparison with the most recent building code, adopted from Cetin et al.¹

(E) The two M7+ earthquakes as observed on seismographs at HIA station at a distance of about 7,000 km.

(F and G) Near-field observation of ground displacements observed with GPS for the two M7+ earthquakes.

(H) Gravimeter (Wuhan Jiufeng station) observation at a distance of about 8,000 km.

(I) Magnitude-time distribution of the earthquake sequence in the first 4 weeks.

How and why

The earthquake sequence took place in a region characterized by intricate geological fault networks, resulting from the tectonic extrusion of the Anatolian block, caused by the convergence of the Arabian and Eurasian plates in a north-south direction. Seismological and geodetic observation networks that record ground displacements, velocities, and accelerations of the Earth’s surface are crucial for understanding earthquakes. The rupture of the M7.8 earthquake may have commenced on a splay branch of the eastern Anatolian fault (EAF) system and then propagated to a 300-km segment of the EAF, where many on-fault aftershocks subsequently occurred. The EAF slip caused significant stress changes on neighboring faults and triggered off-fault aftershocks. The centroid moment tensor solutions of the two M7+ earthquakes (Figure 1A) show clear strike-slip mechanisms, albeit with substantial non-double-couple components. This suggests that ruptures occurred on complex fault plane geometry or fault networks.

Two co-seismically ruptured faults can be clearly observed on Figure 1C, which is obtained via pixel tracking of Sentinel-2 optical image using COSI-Corr. There are hints of deformation associated with rupture on the initial splay fault. The M7.8 earthquake lasted for more than 60 s and had a complicated source time function, as revealed by teleseismic P waves, in contrast to the much shorter duration and simpler rupture processes of the M7.6 earthquake (Figure 1E). Near-field GPS station MLY1, located more than 100 km away from the centroid of the earthquakes, recorded both dynamic and static displacements of tens of centimeters (Figures 1F and 1G), demonstrating the large magnitude of the earthquakes. The Wuhan superconducting gravimetry station recorded the earthquake sequence and solid Earth tide signals (Figure 1H), providing critical insights into the ultra-long period behavior of the earthquake rupture processes.

The earthquake sequence that occurred in this particular location presents a challenging task to answer the questions of how it unfolded and why it happened. Previous geological field investigations in this region have identified complex networks of active faults that have been responsible for historical strong earthquakes. It is supposed that this earthquake sequence ruptured quite a few segments of the fault system. As for why the M7.6 event occurred, stress change from the M7.8 earthquake offers a plausible reason, but a quantitative model is required to clarify the 9-h delay, which presents a challenging task for modeling stress transfer along faults. As for why the earthquake sequence caused so much damage, two aspects need to be investigated in detail. The first aspect is the relationship between strong ground motions and building code design levels, and the second is the design and construction quality of the buildings and infrastructure.

What is needed now

To better understand the complexity of this earthquake sequence, it is essential to have coordinated efforts among various experts such as seismologists, geodesists, and geodynamicists as well as engineering geologists and earthquake engineers. Their collaboration can help address many critical questions about the earthquake sequence and promote strategies for more effective earthquake hazard mitigation.

• 1.Performance of earthquake early warning systems for the case when an initial rupture on a minor splay fault jumps onto a major fault system. Does the first 3-s algorithm work well for this Türkiye-Syria earthquake sequence? Other algorithms that do not rely on initial motions like PLUMB or FinDer may work well in this scenario. Investigation of these alternative early warning algorithms for a case like this earthquake sequence is needed.
• 2.To construct a reliable and detailed model of the rupture process of the two M7+ earthquakes. Such a model is feasible as there are abundant geological, geodetic, and seismological data, including strong ground motion recordings in the heavily damaged zones. These rupture models are crucial for constructing a reliable prediction of the ground motions generated by such events, allowing for a better understanding and assessment of building and infrastructure performance in response to shaking.
• 3.More quantitative models of triggered earthquakes. Pairs of large earthquakes, also known as doublets of strong earthquakes, are not rare, but the intervals between them vary greatly. Typically, the dynamic stress changes resulting from a strong earthquake are much stronger than the static ones at distances a few wavelengths away. However, only qualitative models have been proposed to explain delayed dynamic triggering. With a better understanding of crustal stress, fault friction properties as well as the photoelasticity and rheological structure in Türkiye-Syria region, physics-based geomechanical models can be developed to explain the 9-h delay quantitively.
• 4.History provides clues to the future. It is necessary to expand the geodetic network and conduct further geological field investigations. Valuable data can be obtained through geological field mapping of the surface rupture, as well as by measuring of co-seismic fault offsets and considering the distributed nature of the fault zone. This information can be used to improve empirical fault displacement hazard models, which play a critical role in informing the engineering design of infrastructure situated near faults or those that make fault crossings.
• 5.Seismically induced soil liquefaction caused ejecta, lateral spreading, and subsidence in Holocene sediments over a wide area and resulted in excessive settlements and bearing capacity failures that damaged building foundations.¹ Preliminary mapping of liquefaction phenomena using remote sensing data showed that these phenomena were mainly located along or near the coast, fluvial valleys and drained lakes and swamps, confined to a small fraction of Holocene deposits.² It is important to use these data to assess the capability of using remote sensing data to assist the much more laborious field work that is needed to identify locations of high liquefaction potential prior to future earthquakes.
• 6.A dense seismograph network not only detects earthquakes but can also assist in early warning of spontaneous or earthquake-induced landslides.³ Combining physics-based numerical simulations of landslide processes with deep learning analysis of a large database of landslides caused by historical earthquakes can enhance the potential for early warning systems.

Published Online: April 25, 2023