What's on this Page

What is Creep?

The Causes of Creep

The Importance of Creep Data

How We Measure Creep

Measurement Errors and Site Noise

Types of Creep Behavior


What is Creep?

Fault creep is aseismic fault slip that occurs in the uppermost part of the earth's crust during the time interval between large stress-releasing earthquakes on a fault or as "afterslip" in the days to years following an earthquake. Most faults remain locked during the interval between earthquakes as elastic shear strain in the upper crust builds to a critical stress threshold when elastic strain energy is ultimately released by seismic fault slip (i.e., earthquakes) (Figure 1).

Figure 1. Generalized block diagram illustrating elastic strain buildup on a strike-slip fault that occurs between large, stress-releasing earthquakes. Elastic strain along faults in the upper crust is driven by "continuous" fault movements along aseismic, ductile shear zones in the lower crust.

The Causes of Creep

The causes of fault creep have been the subject of much study, but are most commonly attributed to factors such as low frictional strength on the fault, the low values of normal stress acting on the fault in the shallow crust, and elevated pore-fluid pressures, which act to decrease the effective normal stress on a fault. The creep rate expressed at the earth's surface depends on the rate of elastic strain in the lower crust, the fault's ability (or lack thereof) to resist against the building shear stress, and the depth at which the fault remains locked (i.e., the locking depth) where essentially no creep occurs (Savage and Lisowski, 1993) (Figure 2).

Figure 2. Generalized block diagram illustrating the relation between creep and elastic strain (Figure 1). Click on Figure for larger version.

The Importance of Creep Data

Because creep is an indicator of the shear strain on a fault, knowing how creep rates vary temporally and spatially along faults in the San Francisco Bay area has important implications for forecasting the timing, locations, and potential sizes of future earthquakes and for understanding the mechanics of fault behavior. For example, we now know that creep rates are sensitive to stress changes in the crust induced by moderate to large earthquakes on neighboring faults in the region (Galehouse, 1997; Lienkaemper et al.,1997, 2001), and such stress changes can act to either advance or delay the timing of future earthquakes on a fault (e.g., Toda and Stein, 2002). Differences between creep rates (measured locally along a fault) and elastic strain rates (measured by more regional geodetic data) are also a proxy indicators of locking depth, so creep data can be used to modify seismic hazard estimations, which rely on knowing the thickness of crust that will likely experience the greatest strain release (i.e., slip) during earthquakes. Simpson et al. (2001) used variations in creep rate along the Hayward fault to model changes in locking depth. Their models suggest that locking depths vary along fault strike from 4-12 km. Finally, creep-rate anomalies that may show up with continued monitoring might have potential for actually predicting the location and timing of future earthquakes.

How We Measure Creep

Geodetic monitoring of fault creep is usually accomplished using surveys of alinement arrays or trilateration networks or with creep meters. Creep meters can obtain micron precision (e.g., Bilham et al., 2004) and provide continuous records of the precise timing of creep. However, creep meters commonly span distances of only a few tens of meters and can significantly underestimate creep if they don't span the entire width of the creeping zone. Trilateration networks commonly overestimate creep because they span distances of hundreds of meters to kilometers, and therefore they tend to include elastic strain that occurs away from the fault.

We monitor creep with theodolite surveys of alinement arrays. Alinement arrays (Figure 3) provide the most accurate and complete measurements of creep because they are generally wide enough (typically 130 m) to span the entire creeping zone, but narrow enough to exclude significant elastic strain away from the fault. An alinement array consists of three fixed points marked on permanent survey monuments (or in some cases, nails driven into concrete or pavement). A high-precision theodolite (Wild T2002) is centered and leveled over a point (IS) on one side of the fault and a second target is centered and leveled over an (orientation) point (OS) on the same side of the fault as the theodolite. A third target is centered and leveled over an (end) point (ES) on the opposite side of the fault. The array is designed so that the IS-ES bearing is as close to perpendicular to fault strike as possible.

Figure 3. Typical alinement array. Click on Figure for larger version.

The amount of creep is determined from the change in the angle between the IS-ES and IS-OS directions (delta-theta) that occurs between successive surveys at the site (Figure 4). The amount of right-lateral creep (u) ~parallel to fault strike (i.e., perpendicular to the IS-ES direction) is calculated by the IS-ES distance times the tan(delta-theta) . The IS-ES distance does not change significantly over time, but has been precisely measured with the electronic distance measurer (EDM) compontent of the T2002 total station..

Figure 4. Geometric relations for determining the amount of creep that occurs between successive alinement array surveys. Theta1 is measured during the previous survey at the site. Theta2 is measured during a subsequent survey. ES1 and ES2 correspond to the positions of ES at the previous and subsequent surveys. Note that the amount of creep (u) (mm's) is greatly exaggerated with respect to the IS-ES distance (~100 m). Click on Figure for larger version.

If the IS-ES direction is not perpendicular to fault strike, then u must be trigonometrically corrected so that ucorrected is resolved in a direction that is parallel to fault strike. ucorrected is always greater than u, and is determined by dividing the uncorrrected creep (u) by the cos(90 degrees minus alpha) [or the sin(alpha)], where alpha is the acute angle between the IS-ES direction and fault strike.

Figure 5. Geometric relations for correcting the measurement of creep for the case when the IS-ES direction is not perpendicular to fault strike. Note that u and ucorrected are greatly exaggerated with respect to the IS-ES distance, and so alpha and minus-alpha are essentially equal. Click on Figure for larger version.

Measurement Errors and Site Noise

We employ a high-precision Wild T2002 theodolite/total station to conduct our surveys. This instrument has a specified accuracy of ±0.5 arcsecond (±0.000139 degrees). We shade the instrument with a canopy during surveys to minimize instrument drift related to fluctuating temperatures. We make multiple sets of measurements during each survey at a site to quantify measurement uncertainties. Two sets of initial azimuth readings are initially taken alternately from the instrument (IS) to the orientation and end points (OS and ES, respectively; Figure 3). The azimuth readings to each of the the two points must must agree to within 0.00060 degrees before two angle measurements between the two azimuths are acceptable and recorded. The theodolite is then flipped vertically 180 degrees and the process is repeated producing a total of 4 angle measurements. We then rotate horizontally the tribraches (i.e., mounts) of the theodolite and both targets 180 degrees, re-level the setups, and repeat the previous proceedure, producing a total of 8 angle measurements. This proceedure is designed to account for instrumental and target setup errors as much as possible. Standard deviations (1-sigma uncertainties) are then calculated from the 8 angle measurements and these error estimates are then applied to the calculations of creep. Creep measurement error increases with increasing IS-ES distance (Figure 4). The precision of the method is such that we can confidently detect any movement greater than 1-2 mm between successive surveys. All creep rates reported on this website are least squares average rates determined by linear regression.

For simplicity we do not show measurement errors for the creep data plots on this website. However, an example of how the standard deviations commonly compare to the creep measurement signal is shown below for data collected at site H8 (HRKT on map of creep measurement sites) on the Hayward fault in Fremont between 1993 and 2004 (Figure 6).

Figure 6. Creep data plot for site H8 (HRKT on map of creep measurment sites) on the Hayward fault, in Fremont, CA showing one standard deviation error bars for creep measurements. Click on Figure for larger version.

The Fremont site (above) exhibits an average creep rate of ~6.4 mm/yr during this time interval, but also shows considerable site noise. Inspection of the 1 sigma error bars indicates that the site noise is not an artifact of measurement uncertainties and that the error estimates are very small relative to the overall creep signal. Alternatively, the site noise is almost certainly related to the response of soil (i.e., regolith) to annual rainfall and variations in water saturation (Figure 7). Similar climatic effects on the creep signal have been documented along the Parkfield segment of the San Andreas fault (Roeloffs, 2001).

Figure 7. Creep data plot (1993-2004) for site H8 (HRKT on map of creep measurment sites) on the Hayward fault, in Fremont, CA together with monthly rainfall totals recorded during the same time interval in Livermore, CA located ~24 km to the northeast. Monthly rainfall totals are from the Desert Research Institute's Western Region Climate Center website. Click on Figure for larger version.

Figure 7 illustrates how site noise and short-term records of apparent creep are influenced by variations soil moisture content. The graph shows, nearly without exception, that the onset of the rainy season marks the transitions from apparent left-lateral creep recorded during the dry Summer and early Fall months to apparent levels of right-lateral creep recorded after the ground begins to moisten with the first rains. Many of our data plots show this type of site noise, and in most cases, the noise is probably due to this same phenomenon. The graph also illustrates how long-term monitoring of creep is necessary to accurately characterize creep rates and to potentially recognize anomalous creep behavior.

Types of Creep Behavior

Details of creep behavior vary broadly from site to site, but most sites show patterns of creep that resemble one of three different types of creep; steady-state (Figure 8), episodic (Figure 9), and episodic with steady-state background creep (Figure 10). Some sites also show different types of creep behavior during differrent time intervals. In addition, triggerred slip and longer-term changes in creep rates have been recognized and shown to coincide with moderate to large earthquakes in the region, such as the 1984 Morgan Hill and 1989 Loma Prieta earthquakes (Galehouse, 1990, 1997; Galehouse and Lienkaemper, 2003; Lienkaemper et al.,1997, 2001). Triggered slip can occur at the time of the distant earthquake due to transient, dynamic stress changes associated with the passing of seismic waves (e.g., Bodin et al., 1994), whereas longer-term creep-rate changes are related to more permanent, static stress changes. Creep behavior at site CV7S along the Calaveras fault, near Holister (see map of creep measurement sites), appears to show evidence of the effects of both dynamic and static stress changes associated with the Loma Prieta earthquake. (To access up-to-date creep data and data plots for all of our sites, go to http://pubs.usgs.gov/of/2009/1119/.
Figure 8. Model for steady-state creep (average creep rate of 5.0 mm/yr). Steady-state creep is charaterized by a linear relation of creep through time. Sites that appear to show this type of behavior or intervals of this behavior include most sites along the Hayward fault, several sites on the Calaveras fault, and SAMV on the San Andreas fault (see map of creep measurement sites). However, the steady-state signals are often obscurred by site noise. Click on Figure for larger version.
Figure 9. Model for episodic creep (average creep rate of ~5.9 mm/yr). This type of creep is characterized by distinct creep events that are often separated by similar time intervals. This pattern of creep resembles frictional, stick-slip fault behavior. Sites that show this type of behavior include our Concord fault sites and CV7S along the Calaveras fault in Holister, CA (see map of creep measurement sites). Click on Figure for larger version.
Figure 10. Model for episodic creep with a background (steady-state) creep rate of ~2.2. mm/yr (total average creep rate of ~6.0 mm/yr). Sites that show intervals of this type of behavior include sites along the Concord fault. Click on Figure for larger version.


Bilham, R, N. Suszek, and S. Pinkney (2004). California creep meters, Seism. Res. Lett. 75 (4), 481-492.

Bodin, P., R. Bilham, J. Behr, J. Gomberg, and K. Hednut (1994). Slip triggered on the southern California faults by the Landers earthquake sequence, Bull. Seism. Soc. Am. 84, 806-816.

Galehouse, J.S. (1990). Effect of Loma Prieta earthquake on surface slip along the Calaveras fault in the Hollister area, Geophys. Res. Lett. 17, 1219-1222.

Galehouse, J.S. (1997). Effect of Loma Prieta earthquake on fault creep rates in the San Francisco Bay region, U.S. Geol. Surv. Prof. Pap. 1550-D, D193-D207.

Galehouse, J.S., and J.J. Lienkaemper (2003). Inferences drawn from two decades of alinement array measurements of creep on faults in the San Francisco Bay region, Bull. Seis. Soc. Am. 93 (6), 2415-243.

Lienkaemper, J.J., J.S. Galehouse, and R.W. Simpson (1997). Creep response of the Hayward fault to stress changes caused by the Loma Prieta earthquake, Science 276, 2014-2016.

Lienkaemper, J.J., J.S. Galehouse, and R.W. Simpson (2001). Lon-term monitoring of creep rate along the Hayward fault and evidence for a lasting creep response to the 1989 Loma Prieta earthquake, Geophys. Res. Lett. 28, 2265-2268.

McFarland, F.S., Lienkaemper, J.J., and Caskey, S.J. (2009), Data from theodolite measurements of creep rates on San Francisco Bay region faults, California, 1979-2009; USGS Open-file Report 2009-1119. (http://pubs.usgs.gov/of/2009/1119/)

Roeloffs, E.A. (2001). Creep rate changes at Parkfield, California, 1966-1999: seasonal, precipitation induced, and tectonic, J. Geophys. Res. 106, 16,525-16,547.

Savage, J.C., and M. Lisowski (1993). Inferred depth of creep on the Hayward fault, central California, J. Geophys. Res. 98, 787-793.

Simpson, R.W., J.J. Liekaemper, and J.S. Galehouse (2001). Variations in creep rate along the Hayward fault, California, interpreted as changes in depth of creep, Geophys. Res. Lett. 28, 2269-2272.

Toda, S., and R.S. Stein (2002). Response of the San Andreas fault to the 1983 Coalinga-Nunez earthquakes: an application of interaction-based probabilities for Parkfield, J. Geophys. Res. 107.

Back to Creep Homepage