Correlating variations of b values and crustal deformations during the 1990s may have pinpointed the rupture initiation of the Mw=7.4 Izmit earthquake of 1999 August 17

Summary The b value increases from 1.1 along straight segments to 1.7 along a releasing bend of the North Anatolian Fault System between 30.1°E and 30.8°E. The tensional nature of the bend is corroborated by extension rates of 0.3 μstrain yr−1 as obtained from Global Positioning System (GPS) data. Fault plane solutions in this area change from abundant right-lateral strike-slip to oblique slip with normal components in the central part of the bend, compatible with the formation of a pull-apart basin. We suggest that this crustal volume is highly fractured, corresponding to a short mean crack length, due to the superposition of horizontal and normal faulting. As a consequence of low normal stress on the existing cracks, frictional sliding is assumed to be the predominant mechanism for the generation of seismic events, resulting in high b values. The lowest b values (b∼ 0.8) in the area investigated are found at the junction between the fault bend and the adjoining segment, which runs through the epicentre of the 1999 Izmit earthquake. At the junction, a localized stress concentration is expected from numerical models of seismicity along geometrical barriers. Thus, the site of lowest b has been considered to be the most likely place for a major earthquake, a conclusion that is confirmed by the Izmit earthquake, with epicentre located about 13 km from the anticipated site. In early 1992, the spatially averaged b value started to increase along the central and northwestern parts of the fault bend, and there was a corresponding period of intensified horizontal extension. Since no large earthquake was recorded in 1992, it is assumed that the anomalous extension and the associated change in stress have been caused by aseismic fault creep. In view of the available seismotectonic information, surface displacements due to a slow dislocation at depth were calculated. The GPS data are reasonably explained by a source of normal fault type at a depth of 10 km and equivalent magnitude MW = 5.9. Stress changes at seismogenic depths are calculated under the assumption that an increase in b is most likely to be observed along those segments where both deviatoric and mean stress are reduced, while the Coulomb stress change is positive; that is, the fault is brought closer to failure. For right-lateral strike-slip, the stress condition is fulfilled at the sites of the largest b-value increase. The surface rupture of the Izmit earthquake is associated with known fault traces only in the west and in the east of the study area. Along the fault bend, it detaches itself from mapped fault traces and runs through the regions of positive Coulomb stress change caused by the inferred 1992 slow event. Maximum surface slip (5 m) is observed in the immediate vicinity of sites of maximum Coulomb stress change (∼3 bar), indicating that the 1992 event could have partially unclamped the rupture plane of the subsequent Izmit earthquake. We propose that the strong differences in b observed along the complex fault bend can be used to characterize different states of tectonic deformation in space and time.