The fault beneath Istanbul doesn't behave the way scientists once thought.
New research from USC shows that variations in underground temperature and sediment thickness segment the Main Marmara Fault in ways that control where earthquakes start, how far they spread and where they stop — findings that could reshape risk assessments for one of the world's most vulnerable megacities.
The study, published in Nature Communications Earth & Environment , focuses on the North Anatolian Fault beneath the Sea of Marmara that hasn't produced a major earthquake since 1894. Using physics-based simulations that model more than 10,000 years of seismic activity, researchers found that the fault is unlikely to rupture in a single, catastrophic event. Instead, it will likely break in segments, with maximum earthquake magnitudes reaching about 7.3.
"Fault geometry tells us where earthquakes are possible, but rheology — how rocks deform under stress — tells us how they actually unfold," said Sylvain Barbot, the study's principal investigator and associate professor of Earth sciences at the USC Dornsife College of Letters, Arts and Sciences. "Variations in temperature and rock type along the Main Marmara Fault act as barriers that can stop ruptures or cause the fault to creep instead of breaking in a large earthquake."
How scientists simulated thousands of years of earthquakes
The Main Marmara Fault is part of the North Anatolian Fault system, which has produced devastating earthquakes throughout Turkish history. While previous studies mapped the fault's geometry and slip rates, researchers hadn't fully understood why earthquakes stop where they do — a critical question for estimating maximum possible magnitudes.
The answer lies beneath the seafloor. In the central Sea of Marmara, thick sedimentary basins sit above the warmer crust, creating what the researchers call a strong rheological barrier. Frictional properties of sedimentary rocks under specific temperature and pressure conditions show that they deform slowly and stably at shallow depths rather than breaking suddenly. Meanwhile, elevated temperatures at greater depths weaken rocks in ways that prevent large ruptures from growing.
"The main takeaway is that temperature and sediment thickness fundamentally change how the fault behaves," said Sezim E. Guvercin, a postdoctoral researcher at USC Dornsife and first author of the study. "These variations create zones that resist rupture, particularly beneath sedimentary basins in the central Sea of Marmara."
To test this, the research team built a three-dimensional earthquake-cycle model combining realistic fault geometry, frictional properties of rocks and thermal structure based on regional heat-flow measurements. The simulations used Unicycle, an open-source code that can model thousands of years of seismic cycles.
Different segments, different earthquake patterns
When the model incorporated both sedimentary layers and temperature variations, it reproduced key features of the historical record, including the large 1766 and 1912 earthquakes. Over the simulated period, no earthquake exceeded magnitude 7.3.
Different parts of the fault showed distinct patterns. The western Ganos and Tekirdağ segments, which are cooler and geometrically simpler, produce more regular earthquakes — including magnitude 7.2 events recurring roughly every 150 years. The eastern segments, Kumburgaz and Princes' Islands, generate smaller, more frequent earthquake doublets, typically between magnitude 6.2 and 6.8 roughly every 100 years and a magnitude 7.0 earthquake roughly every 500 years.
The models also predict shallow creep — slow, continuous slip that releases stress without breaking — in parts of the fault near the Central Basin. That behavior matches geodetic observations and clusters of small, repeating earthquakes recorded over the past two decades.
"Earthquakes tend to nucleate near bends in the fault, where stresses are highest," Barbot said. "But whether a rupture keeps going or stops is largely controlled by rheology."
Models that ignored sedimentary basins or thermal structure consistently overestimated earthquake sizes and missed behaviors such as creeping segments. Only by incorporating the physical complexity of underground geology did the simulations match observed patterns.
What this means for Istanbul
The findings don't reduce Istanbul's earthquake risk. Moderate to large earthquakes occurring closer to the city, or in rapid succession, could still cause catastrophic damage. Instead, the research provides a more accurate picture of how the fault actually behaves — information essential for building codes, emergency planning and infrastructure decisions.
"Our work shows that what's underground — heat, rocks and structure — matters enormously for earthquake behavior," Guvercin said. "Integrating these factors is essential for improving seismic hazard forecasts in regions like Istanbul."
The locked segments of the Main Marmara Fault on both sides have now gone more than 100 years without a major rupture. If these segments follow the patterns observed in simulations, the region may experience major earthquakes in the coming decades. Understanding exactly how the fault will break when it does could help identify which parts of Istanbul face the greatest risk.
