Near Parkfield in central California, the San Andreas Fault changes character. To the northwest, much of the fault slips gradually without producing large earthquakes. To the southeast, it is predominantly locked and stores elastic strain. The short transitional reach between those regimes does both: parts creep, while a deeper patch has repeatedly broken in earthquakes near magnitude 6.
This combination has made Parkfield one of the most closely examined fault zones on Earth. Its movement shows that creep and earthquakes are not opposing states shared uniformly by an entire fault. They can occupy neighboring depths and strands, interact through changing stress, and shift their rates during an earthquake cycle.
Motion Without a Mainshock
Fault creep is slow slip that occurs without the rapid rupture responsible for an ordinary earthquake. Along the central creeping section of the San Andreas, the two sides move right-laterally: viewed from either side, the ground across the fault shifts toward the right. Northwest of Parkfield, measured rates approach roughly 25 to 30 millimeters per year in some places, comparable to a large share of the long-term plate-boundary motion accommodated by the fault.
The movement is not always perfectly steady. Some surface stations record gradual displacement interrupted by small creep events, while other instruments detect transient accelerations or pauses. Creep also varies along the trace and with depth. A marker at the surface may be moving while a stronger patch several kilometers below remains coupled.
Parkfield lies near the southern end of the approximately 150-kilometer-long central creeping zone. Between the town and Gold Hill, the fault transitions toward the locked Cholame segment, which forms part of the rupture area of the great 1857 Fort Tejon earthquake. This change in frictional behavior concentrates scientific interest into a relatively compact area.
A Locked Patch Surrounded by Slip
Small earthquakes occur frequently around Parkfield, and some repeatedly rupture nearly the same tiny patches. Their highly similar waveforms indicate that they are recurring on the same structures while adjacent material slips aseismically. These repeating earthquakes act as buried indicators of changing slip rates beyond the reach of surface instruments.
The magnitude-6 sequence behaves differently. Earthquakes occurred near Parkfield in 1857, 1881, 1901, 1922, 1934, 1966, and 2004, although their intervals and rupture details were not identical. The apparent regularity led to a formal prediction that another event was likely before 1993. It did not arrive until September 28, 2004, demonstrating that a short historical sequence cannot turn a variable fault into a reliable clock.
The 2004 earthquake ruptured a deeper locked area and was followed by substantial afterslip. Some of the displacement released after the mainshock spread through portions of the fault already prone to stable sliding. This response illustrated how rapid failure of one patch can redistribute stress into surrounding creeping material.
Drilling Into the Moving Fault
The San Andreas Fault Observatory at Depth was constructed northwest of Parkfield to investigate the fault from within. Rather than drilling vertically through it, engineers directed a borehole at an angle until it crossed active strands about 2.7 kilometers beneath the surface. Instruments recorded earthquakes close to their sources, and recovered core provided actual fault-zone material for laboratory testing.
The active zones contain finely ground, foliated fault gouge that is rich in clay and mechanically weak. Serpentinite-bearing rock contributed magnesium-rich minerals, including saponite, that can slide with relatively low friction under the measured conditions. The gouge also has extremely low permeability in tested samples, allowing it to restrict fluid movement across the fault.
These findings support a chemical and mineralogical explanation for creep, but they do not reduce it to one ingredient. Rock composition, temperature, pore-fluid pressure, stress orientation, fault-zone structure, and the rate at which surfaces weaken or strengthen during sliding all influence the outcome. Material that creeps at one depth may behave differently where pressure and temperature change.
Measuring Several Kinds of Slow Slip
No single instrument captures the full process. Creepmeters span the fault at the surface and continuously record changes in distance. Alignment arrays track offset markers across a wider deformation zone. Satellite radar, survey measurements, and satellite positioning reveal displacement across the surrounding landscape, while borehole strainmeters detect very small changes that may not leave an obvious surface offset.
Seismic records add another dimension. Repeating microearthquakes reveal slip changes at depth, and low-frequency earthquakes and tremor illuminate motion below the ordinary seismogenic zone. Recent analyses of continuous strain records have identified numerous short-duration slow-slip candidates near Parkfield, with supporting signals on nearby creepmeters. Such observations indicate that aseismic motion spans steady creep, discrete surface steps, and brief underground episodes.
What Parkfield Reveals About Fault Creep
Creep releases tectonic strain, but it does not guarantee that every nearby patch is safe from fast rupture. A slowly moving shallow section can overlie locked rock, and a creeping region may transmit or redirect stresses toward an asperity—a comparatively strong area that resists motion until it fails. Conversely, stable sliding can sometimes slow or stop an advancing rupture, depending on fault geometry and friction.
Parkfield’s importance lies in this close coexistence. The site provides surface measurements, borehole observations, recovered fault rock, and a recorded magnitude-6 earthquake within the same natural laboratory. Together they show a fault as a layered mechanical system, not a line that is simply locked or unlocked. Its continuing creep documents plate motion in real time while exposing the boundaries where quiet displacement gives way to earthquake-producing failure.