MODELING OF PERIODIC GREAT EARTHQUAKES ON THE SAN-ANDREAS FAULT - EFFECTS OF NONLINEAR CRUSTAL THEOLOGY

We analyze the cycle of great earthquakes along the San Andreas fault with a finite element numerical model of deformation in a crust with a nonlinear viscoelastic theology. The viscous component of deformation has an effective viscosity that depends exponentially on the inverse absolute temperature and nonlinearly on the shear stress; the elastic deformation is linear. Crustal thickness and temperature are constrain ed by seismic and heat flow data for California. The models are for an ti plane strain in a 25-km-thick crustal layer having a very long, ver tical strike-slip fault; the crustal block extends 250 km to either si de of the fault. During the earthquake cycle that lasts 160 years, a c onstant plate velocity v(p)/2 = 17.5 mm yr(-1) is applied to the base of the crust and to the vertical end of the crustal block 250 km away from the fault. The upper half of the fault is locked during the inter seismic period, while its lower half slips; at the constant plate velo city. The locked part of the fault is moved abruptly 2.8 m every 160 y ears to simulate great earthquakes. The results are sensitive to crust al theology. Models with quartzite-like theology display profound tran sient stages in the velocity, displacement, and stress fields. The pre dicted transient zone extends about 3-4 times the crustal thickness on each side of the fault, significantly wider than the zone of deformat ion in elastic models. Models with diabase-like theology behave simila rly to elastic models and exhibit no transient stages. The model predi ctions are compared with geodetic observations of fault-parallel veloc ities in northern and central California and local rates of shear stra in along the San Andreas fault. The observations are best fit by model s which are 10-100 times less viscous than a quartzite-like theology. Since the lower crust in California is composed of intermediate to maf ic rocks, the present result;suggests that the in situ viscosity of th e crustal rock is orders of magnitude less the rock viscosity determin ed in the laboratory.