Stress change calculations provide clues to aftershocks in 2023 Türkiye earthquakes
Since the Feb. 6 magnitude-7.8 Gaziantep-Nurdağı earthquake and magnitude-7.5 Kahramanmaraş-Ekinözü earthquake struck southern Türkiye near the border of Syria, thousands of buildings in both nations have crumpled, killing or trapping thousands of people. The horrific death toll will to continue to rise as rescuers race against both time and freezing temperatures to find people still trapped.
Hundreds of aftershocks of varying magnitudes continue to shake the devastated region, further damaging or destroying buildings, roads, railways and other infrastructure while hampering rescue and relief efforts. Meanwhile, scientists around the world are analyzing data from these events to try to understand what happened, and how we might help.
Our goal with this contribution is to understand the role of stress transfer in promoting Türkiye’s evolving earthquake sequence. This can give us tools and insights that might help in forecasting the distribution of future aftershocks and would enable us to consider the possibility of subsequent mainshocks. So, while this is preliminary, our hope is that it could be timely.
Our principal tools for this analysis include: earthquakes from the U.S. Geological Survey (USGS) ANSS catalog; active fault maps from both the Turkish agency MTA (Emre et al., 2018) and the existing scientific literature (Gülerce et al., 2017; Guvercin et al., 2022); source models for the largest events from the USGS; and Coulomb 3.4 (Toda et al., 2011) — our static stress transfer software that treats Earth’s crust as an elastic halfspace. Simply put, the crust is modeled as though it behaves like a stiff block of rubber. Earthquakes are treated like cuts in the rubber with their two sides displaced, or sheared, according to USGS models that provide insight about the spatial extent, amplitude and duration of rupture (these are USGS finite fault models). We also need to assume the coefficient of friction on the faults that receive stress, and for this we use 0.4, a mid-range value widely used in other studies. A friction of 0 is essentially a perfectly slippery fault and would behave like teflon. A value of 0.75 would apply to a geologically young and rough fault surface — movement is restricted and difficult to overcome.
To understand how common or rare the magnitude-7.8 Gaziantep-Nurdağı and magnitude-7.5 Kahramanmaras-Ekinözü quakes are within the East Anatolian Fault Zone, we used Temblor’s T-GEAR (Temblor Global Earthquake Activity Rate) model, an evolution of GEAR1 (Bird et al., 2015). GEAR1 has been under CSEP test for nine years (Strader et al., 2018), where it has outperformed its parent components, Global CMT seismicity and the SHIFT model (Bird and Kreemer, 2015). T-GEAR uses 50 times the data of GEAR1, and in our assessments, retrospectively tests better.
Did the magnitude-7.8 mainshock trigger or promote the magnitude-7.5 event?
We first calculate the stress imparted to surrounding faults under the simplifying assumption that all faults are parallel to the magnitude-7.8 rupture (figure below). In this scenario, increases in stress beyond the ends of the rupture correspond fairly well to the first several days of aftershocks. The area surrounding the Sürgü and Çardak faults (Balkaya et al., 2021) — the likely faults that ruptured during the magnitude-7.5 shock — is also calculated to be promoted toward failure, but by less than 1 bar. (Bar describes units of stress; for reference, about 30 bars of Coulomb stress is generally thought to accumulate before a fault ruptures.) Meanwhile, the fault end zones have calculated stress increases of more than 3-5 bars. In other words, the ends of the rupture are more stressed than the lobes perpendicular to the East Anatolian Fault.
Further, the Sürgü and Çardak faults do not strike parallel to the East Anatolian Fault, so these calculations may or may not be relevant to triggering those faults. No aftershocks greater than or equal to magnitude-4.1 struck in the Sürgü and Çardak fault lobes during the nine-hour period between the magnitude-7.8 and magnitude-7.5 quakes. There is nothing in these calculations that would have permitted us to foresee the location or the size of the magnitude-7.5 earthquake.
We now make a retrospective calculation of the stress transferred by the magnitude-7.8 Gaziantep-Nurdağı earthquake to the Sürgü and Çardak faults (see the figure below). Here, we are trying to represent the fault receiving the stress more realistically, but we still treat it as an idealized planar surface, ignoring its bends and breaks.
This calculation shows that the Sürgü and Çardak faults may have experienced a 1-2 bar stress increase toward failure. The distribution of the increase in stress roughly coincides with the subsequent rupture, which could be coincidental or corroboratory. Because this calculation was made after the magnitude-7.5 quake struck, it has no predictive value; rather, it allows us to say, narrowly, that the magnitude-7.8 earthquake likely promoted the magnitude-7.5 event. That is not the same as saying that the magnitude-7.8 event caused the magnitude-7.5 quake. The stress increase is no more than 5% of the roughly 30 bars of Coulomb stress thought to accumulate before a fault ruptures, as mentioned above.
Where will future aftershocks strike?
Next, we consider the combined stresses imparted by the magnitude-7.8 and magnitude-7.5 earthquakes, and their relationship to the aftershocks that have occurred since the magnitude-7.5 struck (figure below). One can see that the stress lobes changed significantly after the magnitude-7.5 event because of the interaction of the two large earthquakes. Note that the red band trending east-west is now much stronger than the northeast-striking band of the magnitude-7.8 event. Also visible is a fair correlation of aftershocks to the zones of calculated stress increase (the red stress trigger lobes). This provides some confirmation that the stresses do indeed influence the distribution of aftershocks, and so the stress trigger lobes are likely sites of continuing aftershocks. These calculations do not enable us to forecast whether another large aftershock will strike; they only guide our understanding of where shocks of all sizes are most likely.
How rare are magnitude-7.8 and magnitude-7.5 shocks along the East Anatolian Fault?
We used Temblor’s T-GEAR model to assess the likelihood of these events, considering the entire East Anatolian Fault Zone. Here, we treat the two earthquakes as independent, which is not likely the case. Nevertheless, this exercise can provide some insights. The East Anatolian Fault has a slip rate of about 10 millimeters per year — less than half the rate of the North Anatolian Fault and California’s San Andreas Fault, and about the same as California’s Hayward Fault. Since 1795, the East Anatolian Fault has experienced about eight magnitude 6.7-7.2 events (Güvercin et al., 2022). What we find is shown in the table below.
We think these estimates are consistent with the relatively brief historical record, as well as the low slip rate on the fault and the amount of slip in earthquakes of this size. Thus, even though the East Anatolian Fault is shorter in overall length, slips more slowly, and has not had a similar progression of earthquakes in the last century as the North Anatolian Fault, it still is capable of very large earthquakes, like the two investigated in this study. So, our answer is that these events are rare, but not improbable.
Did the 2020 magnitude-6.8 Elazığ earthquake on the East Anatolian Fault earthquake influence this sequence?
One possible explanation for why the magnitude-7.8 Gaziantep-Nurdağı earthquake didn’t — or hasn’t yet — ruptured to the northeast on the East Anatolian Fault is the occurrence of the magnitude-6.8 Elazığ earthquake in 2020 (also called the Doğanyol-Sivrice earthquake). But for two reasons, it seems unlikely to us that this earthquake (shown in the figure below) plays a significant role. First it is small, with an energy release (“seismic moment”) of only about 3% of the magnitude-7.8 event. The second reason is that it is well to the northeast of the magnitude-7.8 rupture, and there is still a gap of about 50 kilometers between the northeast end of the magnitude-7.8 break and the southwest end of the magnitude-6.8 rupture.
As more data arrive from the Turkish networks, geological reconnaissance, satellite imagery and international collaborations, we should be able to enhance these calculations and assess their true uncertainties. But tentatively, we infer that the magnitude-7.8 earthquake promoted the magnitude-7.5 event, and we find that the distribution of smaller aftershocks is consistent with these admittedly crude but widely used calculations of stress transfer. The most likely future scenario is that aftershocks will remain clustered in the stress trigger lobes (the red lobes), where they will undergo an Omori (or power law) decay over time of frequency — but not necessarily magnitude.