Influence of Releasing Step-Overs on Surface Fault Rupture and Fault Segmentation: Examples from the 17 August 1999 Izmit Earthquake on the North Anatolian Fault, Turkey

Abstract

The 17 August 1999 Izmit earthquake broke four distinct structural segments of the North Anatolian fault, from east to west: the Karadere, Sakarya, Sapanca, and Golcuk fault segments. These segments are separated from one another and the adjoining Yalova and Duzce fault segments to the west and east, respectively, by distinct right-releasing step-over and/or gaps in the fault trace. These include, from east to west, the Eften Lake step-over (2-4 km wide), Akyazi gap (6-13 km long), Sapanca step-over (1-2 km wide), Golcuk step-over (1-2 km wide), and Karamursel step-over (4-5 km wide). The earthquake nucleated within or near the Golcuk step-over with bilateral rupture propagation to the west and east. To the west, about 4-5.5 m of surface rupture (up to 7 m of subsurface rupture) on the Golcuk segment was mostly, if not entirely, arrested by the 4- to 5-km-wide Karamursel step-over. To the east, about 3-4 m of surface rupture (about 4 m of subsurface rupture) propagated through the 1- to 2-km-wide Sapanca step-over, triggering up to 4.5-5.5 m of surface rupture (up to 5-6 m of subsurface rupture) on the Sakarya segment. Rupture on the Sakarya segment stepped across the Akyazi gap triggering 1-1.5 m of rupture on the Karadere segment. Rupture on the Karadere segment was arrested by the 2- to 4-km-wide Eften Lake step-over. Modeling of the transient stress field associated with the rupture propagation agrees well with the field observations. The modeled fault rupture of 5 m on the Golcuk segment appears to decrease the transient stress on the eastern tip of the adjoining Yalova segment across the 4- to 5-km-wide Karamursel step-over, even if we allow rupture on the Golcuk segment to extend along the entire southern margin of the step-over. Our model shows that regardless of the amount of slip, in a generic sense, it is difficult to increase stress on the tip of an adjoining fault segment across a 5-km-wide step-over. On the other hand, fault rupture of 3 m on the western tip of the Sapanca fault segment significantly increases the transient stress on the adjoining Sakarya segment across the 1- to 2-km-wide Sapanca step-over, allowing rupture propagation across the basin. An evaluation of 30 historical strike-slip earthquakes involving 59 step-over basins supports our field observations and stress modeling of the Izmit fault rupture. Historical strike-slip events with small to large displacements usually propagate through step-overs less than 1-2 km wide. With increasing displacement, larger and larger step-overs can be ruptured through. The empirical data generally show a ratio ranging from 2:1 to 1:1 between step-over width (in kilometers) and strike-slip displacement (in meters) for a through-going rupture. However, step-overs of 4- to 5-km width appear to always arrest fault rupture, regardless of the amount of displacement. These results, therefore, provide important constraints for evaluating the probability of multisegment ruptures on segmented strike-slip faults. Manuscript received 20 August 2000.