INTRODUCTION
Changes in seismic style along the San Andreas fault system appear to correlate with strike-parallel variations in regional heat flow. Large earthquake ruptures are associated with colder segments of the fault zone, while comparatively high levels of micro earthquake activity and small to moderate shocks occur along the hotter segments, often in association with a seismic creep. These profiles suggest that the transition from frictional to quasi-plastic behavior with depth should increase by a few kilometers passing from the hotter to the colder portions of the fault zone, giving rise to long-wavelength concentrations of distortional strain energy near the base of the seismogenic zone in the cold segments. Some support for this hypothesis comes from the observed deepening of microseismic activity passing northwest and southeast from the central Californian active region to the currently locked, ~400 km segments of the San Andreas fault that ruptured in 1906 and 1857.
Observations gathered over the past 150 years or so suggest that alternating segments of the San Andreas fault system persistently exhibit different types of fault activity or seismic style (Fig. 1), and it seems probable that these different behavioral modes are long term characteristics of the fault system (Allen, 1968, 1981; Wallace, 1970; Scholz, 1977). A seismic slip rates reach 32 mm/yr along one 55 km section of the central California segment where geodetic studies indicate large scale rigid block motion without significant accumulation of elastic strain (Thatcher, 1979; Burford and Harsh, 1980). The complex activity northwest of the triple junction at Cape Mendocino results from transform faulting along the Mendocino Fracture Zone combined with the subduction and internal disruption of the Gorda plate (Smith and Knapp, 1980), and will not be discussed further. Several explanations have been put forward to account for these variations in fault behavior. On geometrical grounds, Allen (1968) noted that both M8 ruptures were associated with "restraining" bends where the San Andreas fault trace lies oblique to the interplate slip vector (see also Scholz, 1977). This paper explores in some detail a further possibility, suggested by Sibson (1982), that the changes in seismic style may be related to variations in regional heat flow along strike.
Measurements of heat flow in the dominantly strike-slip regime close to the San Andreas fault have been compiled by Lachenbruch and Sass (1980) into a strike-parallel profile which exhibits a broad maximum coincident with the central California segment of intense microearthquake activity and a seismic slip (Fig. 2). Comparatively few heat flow determinations are available for the southern active region southeast of the 1857 rupture, but a seismic slip and a high level of microseismic activity are also known in the region of high heat flow and geothermal activity southeast of the Salton Sea (Goulty et al. 1978). It is also interesting to note that the probable epicenters (rupture nucleation sites) for the 1906 and 1857 earthquakes (Boore, 1977; Sieh, 1978) linear the ends of the high heat flow region associated with the creeping segment.
