Predicting the Endpoints of Earthquake Ruptures Steven G. Wesnousky Nature, 444, , 2006 doi: /nature05275

Introduction Active fault traces are: Generally not continuous Segmented by discontinuities Stress concentrations resulting from slip at discontinuities may slow or stop rupture propagation and hence play a controlling role in limiting the length of earthquake rupture.

Data Examined mapped surface ruptures of: 22 historical strike-slip earthquakes km rupture lengths

DateLocationTypeLengthMw 1859 Jan 09San Andreas, CaliforniaSSR Oct 28Neo-Dani, JapanSSL Nov 02Kita-Izu, JapanSSL Dec 25Erzincan, TurkeySSR May 19Imperial, CaliforniaSSR Dec 20Erbaa Niskar, TurkeySSR Sep 10Tottori, JapanSSL Nov 26Tosya, TurkeySSR Feb 01Gerede Bolu, TurkeySSR Jul 22Mudurnu, TurkeySSR Apr 08Borrego Mountain, CaliforniaSSR Oct 15Imperial, CaliforniaSSR Jul 29Sirch, IranSS Nov 23Superstition Hills, CaliforniaSSR Jul 16Luzon, PhilippinesSSL Jun 28Landers, CaliforniaSSR Mar 14Fandoqa, IranSSN Aug 17Izmit, TurkeySSR Oct 16Hector Mine, CaliforniaSSR Nov 12Duzce, TurkeySSR Nov 14Kunlun, ChinaSSL Nov 03Denali, AlaskaSSR3027.6

Mapping Fault Geometry Began circa 1968 (Borrego Mtn) Clear that active fault traces are discontinuous Discontinuities in fault traces often associated with endpoints of earthquake ruptures

Focus on continental SS ruptures of length greater ~15 km Depth dimension of brittle failure during SS earthquakes limited to ~15 km owing to physical and frictional constraints Direction of rupture propagation may be viewed as principally horizontal for each event

Figure 1. Map of 1968 Borrego Mountain earthquake surface trace. Adjacent and continuing traces of active faults that did not rupture during the earthquake are shown as dotted lines. Also annotated are the dimensions of fault steps measured approximately perpendicular to the fault strike and the distance to the nearest-neighboring fault from the 1968 rupture endpoints. The star is the earthquake epicenter.

Figure 2. Synopsis of the observations bearing on relationship of geometrical discontinuities along fault strike to the endpoints of historical earthquake ruptures.

Approximately 2/3 of terminations of SS ruptures are also associated with geometrical steps in fault trace or the termination of the active fault on which they occurred

Figure 3. Geometrical discontinuities as a function of size.

Discussion Histogram gives a statistical idea as to whether or not an earthquake rupture will be associated with particular fault step or fault terminus A transition exists between 3 and 4 km, above which rupture has not been observed to propagate through

Transition seems largely independent of rupture length Suggests that magnitude of stress changes and the volume effected by those stress changes at the leading edge of propagating ruptures are largely invariable during the rupture process

Recent theoretical work implies that releasing steps should be easier to rupture through than restraining steps although it is releasing steps that are observed more frequently at the endpoints of earthquake ruptures Releasing steps outnumber restraining steps by six to one in the data set

Observations indicate that deformation processes attendant on cumulative SS displacements are relatively more efficient in the creation and maintenance of extensional steps and analogously in the linkage and removal of restraining steps