Creep Meters Research Articles

Surface creep on the North Anatolian Fault at Ismetpasa, Turkey, 1944–2016

AbstractWe reevaluate the 72 year history of surface slip on the North Anatolian Fault at Ismetpasa since the Mw = 7.4 1944 Bolu/Gerede earthquake. A revised analysis of published observations suggests that days after the earthquake the fault had been offset by 3.7 m and 6 years later by an additional 0.74 m. Creep was first recognized on the fault in 1969 as a 0.13 m offset of a wall constructed in 1957 that now (2016) has been offset by 0.52 m. A carbon rod creep meter operated across the fault in the past 2 years confirms results from an invar wire creep meter operated 1982–1991 that surface slip is episodic. Months of fault inactivity are interrupted by slow slip (≤10 µm/d) or multiple creep events with cumulative amplitudes of 2–10 mm, durations of several weeks, and with slip rates briefly exceeding >2.5 mm/h. Creep events accommodate 80% of the surface slip and individually release ≈ 10−6 shear strain on the flanks of the uppermost 3–7 km of the fault. GPS and interferometric synthetic aperture radar methods yield a current fault slip rate of 7.6 ± 1 mm/yr suggesting that creep meters incompletely sample the full width of the surface shear zone. The slip rate has slowed from >10 mm/yr in 1969 to 6.1 mm/yr at present, 4.65 mm/yr of which appears to be due to steady interseismic creep driven by plate boundary stressing rates. We calculate that a further 1 m of aseismic surface slip will precede the next major earthquake on the fault assuming an ≈ 260 year main shock recurrence interval on this segment.

Quantifying creep behaviour of clay-bearing rocks below the critical stress state for rapid failure: Mam Tor landslide, Derbyshire, England

Abstract: The Mam Tor landslide (Derbyshire, England) slips downhill at up to 0.5 m a −1 , and offers an excellent opportunity to study time-dependent creep in mudstones under in situ conditions. Annual surveys show that creep rates increase following heavy rainfall, but to establish detailed correlation between creep and pore water pressure required spatially and temporally higher resolution monitoring. We acquired 3 years of such data, at 3 h intervals, from wire creep meters, borehole piezometers and rainfall monitoring, showing that creep is strongly focused into the winter months and correlates well with pore water pressure. Summer grass and fern growth strongly influence rainfall infiltration, recycling much of the rainfall back to the atmosphere, explaining the seasonal variability of creep. The dependence of creep rate on pore pressure can be recast as a rheological model relating shear strain rate to shear stress using the friction angle for the creeping mudstone to link effective normal stress to shear stress. The creep is very non-linear, following a power law with a stress exponent of 48. This flow law may apply also to shear failure in poorly consolidated mudrocks under high pore pressure conditions; for example, within some landslides, subglacial sediments, tectonic fault zones, tectonic mélanges and accretionary complexes.

The effects of groundwater level and vegetation on creep of the Mam Tor landslip

The creeping Mam Tor landslide (North Derbyshire, UK) is sliding slowly downhill at up to 0.5 m per year. Annual measurements using conventional surveying techniques have shown that creep is exacerbated by periods of heavy rainfall, but to establish a detailed correlation between creep and groundwater levels requires spatially and temporally higher resolution monitoring. We have acquired three years of such data from wire creep meters, borehole piezometers and rainfall monitoring. These measurements showed that creep is strongly focused into the winter months and that it correlates well with groundwater levels. Summer grass and fern growth exerts a strong influence on rainfall infiltration, recycling much of the summer rainfall back to the atmosphere, and explaining the seasonal variability of creep motion.

Active fault creep variations at Chihshang, Taiwan, revealed by creep meter monitoring, 1998–2001

The daily creep meter data recorded at Chihshang in 1998–2001 are presented. The Chihshang creep meter experiment was set up across the Chihshang thrust fault, the most active segment of the Longitudinal Valley Fault, which is the present‐day plate suture between the Eurasian and the Philippine Sea plates in eastern Taiwan. Near‐continuous data recording at two sites revealed different surface fault motions yet similar annual shortening rates: 16.2 mm at the Tapo site (comprising two connected creep meters) and 15.0 mm at the Chinyuan site (three creep meters straddling parallel fault branches). Four of the five creep meters showed a seasonal variation, with the fault moving steadily during the rainy season from April to October, and remaining quiescent during the rest of the year. The only exception was recorded by the creep meter located on a mélange‐composed hillslope, where local gravitational landsliding played an additional role other than tectonic faulting. Through comparison with daily precipitation data, we inferred that moderate rainfall suffices to trigger or facilitate slippage on the surface fault, during the transition period of the dry/wet season. During the observation period from 1998 to 2001, the subsurface seismicity exhibited clusters of microearthquakes on the Chihshang Fault at depths of 10–25 km. Recurrent earthquakes occurred regardless of whether the season was wet or dry, indicating that the stress relaxation associated with seismicity in the seismogenic zone did not transfer immediately up to the surface. The accumulated strain on the Chihshang Fault at shallow surface levels was released through creep during the wet season. In addition to these short‐term seasonal variations, an apparent decrease in the annual slipping rate on the Chihshang Fault during the last few years deserves further investigation in order to mitigate against seismic hazard.

Creep rate changes at Parkfield, California 1966–1999: Seasonal, precipitation induced, and tectonic

Measured fault creep at Parkfield contains information about the slip rate on the San Andreas fault but also is affected by rainfall and other seasonal effects. Seven creep meters record a repeatable seasonal variation that can be removed from the data. Four of the creep meters exhibit surges in creep rate during the rainy season. For these instruments a function of rainfall is developed that crosses a threshold within a few days of the onset of accelerated creep. In addition to the seasonal variations and surges, most Parkfield creep meters record higher creep rates more frequently during the wet season than during the dry season. Characterizing this behavior makes it possible to distinguish between rainfall‐influenced creep and creep rate changes of tectonic origin. Accelerated creep beginning in January 1983 was probably induced by rainfall, rather than by tectonic sources related to the Coalinga earthquake in May 1983. Four creep meters in the central part of the Parkfield fault segment are recording higher creep rates since 1993. A strong case can be made that the increased creep rate at CRR1 is tectonic, but for the other three creep meters an influence of rainfall is difficult to rule out. Strain rate changes recorded on three‐component borehole strain meters suggest that the rate of subsurface creep to the northwest of these instruments is even greater. The 1993 creep rate increase is more abrupt and farther south than the expected signal from accelerating slip near the hypocenter of the 1966 Parkfield earthquake but is in the same location as rapid fault slip reported prior to that event.

Behavior of the southernmost San Andreas Fault during the past 300 years

Surficial creep occurs at low rates along the Coachella Valley segment of the San Andreas fault, which has not produced a large earthquake during the period of historical record. Geodetic data indicate, however, that the crust adjacent to this segment of the San Andreas fault is accumulating strain at a high rate. Furthermore, neotectonic and paleoseismic data indicate that the fault does produce very large earthquakes every two to three centuries. In view of its long‐term behavior, the occurrence of creep along the surficial trace of the fault in the Coachella Valley is of particular interest. Along two short reaches of the San Andreas fault in the Coachella Valley, measurements of offset geological deposits and man‐made structures and from alignment arrays and creep meters show that slip rates of 2–4 mm/yr near Indio and near the Salton Sea have persisted for the past three centuries. This slow aseismic surficial creep is not a transient precursor to seismic failure of this segment of the fault. We suggest that the Coachella Valley segment of the San Andreas fault creeps in its upper few kilometers. This behavior may be due to tectonically induced high pore pressures in the coarse sediments that abut the fault.

Variations in fault slip and strain accumulation at Parkfield, California: Initial results using two‐color geodimeter measurements, 1984–1988

Repeated length measurements of several geodetic baselines near Parkfield, California, have revealed significant variations in the local rates of shallow fault slip and strain. This network of baselines, surveyed several times each week, straddles the San Andreas fault in the transition zone between the creeping section to the northwest and the locked section to the southeast. The length measurements, characterized by a precision approaching 0.1 ppm, reveal large fluctuations in the rates of baseline extension. Principal mode analysis of the length change data indicates that the two largest components of the signal are (1) secular extensions and contractions consistent with surface slip on the main strand of the San Andreas fault, and (2) a large seasonal oscillation with no obvious spatial coherence. On most of these baselines, the second component appears to be in phase with seasonal rainfall. When data from the baselines with the largest amplitude of the seasonal signal are excluded, the remaining data can be modeled in terms of both spatial and temporal variations in surface slip, variations in the components of the changes in uniform strain, and the possible displacement of the central monument in this radial network. In parameterizing this model, the spatial variation of slip beneath the near surface is reflected by changes in shear strain. Although the computed secular shear is highly dependent upon the specified parameterization of surface slip, the data are consistent with the hypothesis that slip at intermediate depths lags behind the surface slip rate. However, the range in models that fit the data does not necessarily imply that there is a deficit in slip at depth relative to the surface. Comparison of the inferred values of surface slip from the model with the observed fault slip measured by very short baseline creep meters indicates close agreement in secular rates, but the short‐term variations observed with the creep meters are either highly attenuated or nonexistent in the modeled slip since the modeled slip is a spatial average which smooths out possible short‐wavelength variations in the surface slip for which the creep instruments are most sensitive. An interesting conclusion from the two‐color data is that surface slip on the San Andreas fault appears to be spread over a 2‐km‐wide zone on the south flank of Middle Mountain but is confined to a very narrow zone to the south as the fault passes through the center of the network. This conclusion is dependent upon the assumption that a few critical monuments are stable and track tectonic displacements in the long term. Finally, the largest observed strain change is an extensional strain coincident with the Kettleman Hills earthquake M5.5 in August 1985.

Historic surface slip along the San Andreas Fault near Parkfield, California

The Parkfield Earthquake Prediction Experiment is focusing close attention on the 44‐km‐long section of the San Andreas fault that last ruptured seismically in 1966 (Ms 6.0). The 20‐km‐long central segment of the 1966 Parkfield rupture, extending from the mainshock epicenter at Middle Mountain southeastward to Gold Hill, forms a 1‐ to 2‐km salient northeastward away from the dominant N40°W strike. Following the 1966 earthquake afterslip, aseismic slip has been nearly constant. Moderate Parkfield earthquakes have recurred on average every 21 years since 1857, when a great earthquake (M≈8) ruptured at least as far north as the southern Parkfield segment. Many measurements of slip have been made near Parkfield since 1966. Nevertheless, much of the history of surface slip remained uncertain, especially the total amount associated with the 1966 event. In 1985 we measured accumulated slip on the four oldest cultural features offset by the fault along the 1966 Parkfield rupture segment. We interpret net slip on each feature as a sum of event slip (sum of coseismic and rapid preseismic and postseismic slip) from Parkfield earthquakes and steady interseismic slip as measured over the last 20 years on nearby alinement arrays, creep meters, and trilateration lines. We assumed for each site that event slip was identical for the 1922, 1934, and 1966 Parkfield events and that long‐term average rate of interseismic slip was constant between all events. Two fences on the southern segment, southeast of Gold Hill, indicate event slip of 13 and 15 cm and interseismic slip rate of 0.36 and 0.30 cm/yr since 1959 and 1908, respectively. At these sites, redundant independent data support our assumption that both event and interseismic slip occur uniformly. On the central segment, near Parkfield, both the 1934 and the 1966 ruptures offset a bridge built in 1932. Interseismic slip rate near the bridge has been about 1.1 cm/yr since 1966; thus we deduce an average event slip of 31 cm for the 1934 and 1966 earthquakes. On a parallel fault trace, 1 km to the southwest, slip was about 8 cm in 1966; thus total event slip summed across the entire fault zone near Parkfield was nearly 40 cm. On Middle Mountain, 4 km north of the 1966 mainshock epicenter, an offset fence indicates 17 cm of slip in 1966 and a 2.26‐cm/yr interseismic slip rate since circa 1946. Thus the central segment of the 1966 rupture is characterized by much larger event slip (∼40 cm) than both distal segments (∼15 cm). This amount of surface slip per event is about twice what had been previously assumed. Larger 1966 surface slip in the central part of the rupture is geodetically compatible with a coseismic slip of 65±10 cm slip on a narrow, buried asperity between Middle Mountain and Gold Hill that has been inferred from the depth distribution of early aftershocks. Assuming our characteristic surface slip model, one can further deduce a deficit in slip since the great 1857 earthquake. Taking the long‐term slip rate as 3.3 cm/yr, the surface slip deficit is 3±0.2 m south of Gold Hill but only 0.3±0.3 m northward from Parkfield.

Assessment of global positioning system measurements for studies of crustal deformation

The U.S. Geological Survey has been using the Global Positioning System (GPS) to perform repeated surveys of networks of different sizes since January 1986. The intersite distances from these various networks are 223 km for a single line connecting the Palos Verdes and Vandenberg sites in southern California; 30–50 km for a network of three sites south of the San Francisco Bay near Loma Prieta; 9–30 km for a network located near Hebgen Lake, Montana; 7–12 km for a network of sites in the vicinity of Parkfield, California; and 240 m for a line located at the Goldstone antenna complex in the Mojave desert. The lengths quoted do not include the intersite distances for the “fiducial” sites included in the solutions for orbit determination, which can be up to 8000 km. We have attempted to assess the “long‐term” precision and accuracy of determinations of relative site position for a range of line lengths. This assessment has consisted of several different studies. In assessing precision, we have generally investigated repeatability, usually about a best fit linear temporal trend. The smallest fractional repeatability was yielded by the measurements on the Palos Verdes to Vandenberg line. The root‐mean‐square (rms) residual about the best fit line for these determinations was 6 mm (0.03 parts per million, or ppm) for the north component, 11 mm (0.05 ppm) for the east component, and 40 mm (0.18 ppm) for the vertical component. For these determinations, GPS data from 3–4 consecutive days were combined to produce a single estimate of the vector separation. We also used the day‐to‐day scatter from these measurements to perform an analysis of variance and found that these multiday determinations can be significantly more precise than estimates determined from a single observing session. The worst results were obtained for the Loma Prieta network, where we obtained rms residuals of 11–12 mm (0.3–0.4 ppm) for north, 14–18 mm (0.3–0.6 ppm) for east, and 18–21 mm (0.4–0.7 ppm) for vertical. We cannot definitely explain these poor results, but we have evidence to indicate that this poor scatter is due to the short observing session (∼4.5 hours, compared to ∼6.5 hours for the Palos Verdes to Vandenberg results). In assessing accuracy, we compared the GPS results to those from independent techniques. At Parkfield we compared the GPS estimates of fault‐parallel motion to those from creep meters and alignment arrays and found excellent agreement (within 1–2 mm yr−1) not only for the slip but also for the variation of slip along the San Andreas fault. We compared the Loma Prieta and Hebgen Lake line length estimates to those obtained from a Geodolite, a high‐precision laser distance‐measurement device that the U.S. Geological Survey has used for 20 years for studies of crustal deformation. If we model the Geodolite minus GPS line length differences ΔL as ΔL = α + βL, where L is the line length, then we find α = 0.6 ± 0.5 mm and β = 0.1 ± 0.1 mm km−1. The uncertainties quoted are one standard deviation. From this comparison, we conclude that at this level of significance there is no detectable difference between lengths determined from Geodolite and GPS data. Finally, we compared the GPS estimates of the Palos Verdes to Vandenberg line length to those obtained from very long baseline interferometry (VLBI). The mean difference between the six GPS estimates of line length and a best fit line through the VLBI estimates was −1 mm, and the rms difference was 12 mm (0.05 ppm), close to the standard deviation of the GPS measurements for this predominantly east‐west line.

Geodetic Predictions of a strike‐slip fault model: Implications for intermediate‐ and short‐term earthquake prediction

Mechanical models of faulting during an earthquake cycle now exist that are based on laboratory‐derived constitutive laws for rock friction. Such models allow calculation of geodetic changes as a function of depth, distance from the fault, and time and are useful for gaining insight on the magnitude and timing of geodetic changes that might occur prior to an earthquake. We have used the Tse and Rice (1986) model of a crustal strike‐slip fault to study the sensitivity of surface deformation to changes in the values of the friction constitutive parameters and the amount of dynamic overshoot during an earthquake. Models were examined with values of the characteristic decay distance Dc in the constitutive law ranging from 5 to 80 mm, with near‐surface values of the steady state velocity dependence a−b of −0.0015 and −0.003, and with either zero or 30% dynamic overshoot. The time variations of the predicted displacement, velocity, strain, and strain rate distributions are strongly affected by changes in Dc and a−b, and only slightly affected by the amount of dynamic overshoot. The larger values of Dc and the less negative value of a−b (i.e., large ratios of Dc to b−a) produce larger and earlier precursory signals. The most useful signals for predicting the model earthquakes are strain changes near the fault, because for nearly all model parameters studied, the predicted strain changes are considerably larger than the detectability limits of existing borehole strain meters in the presence of earth noise. Detectable changes occur a few minutes to a month prior to the earthquake, depending on parameter values. Premonitory increases in velocity also occur; a pattern of larger velocities at distances a few kilometers away from the fault trace is a diagnostic indication of accelerating slip at depth that would be useful for predictive purposes. Unfortunately, these velocity changes are detectable with creep meters and two‐color laser electronic distance measuring devices only for the largest ratios of Dc to b−a. In agreement with recent measurements for several earthquakes, the predicted amount of preseismic moment release is <0.5% of the earthquake moment for most simulations, and the postseismic moment release is a significant fraction of the earthquake moment. This two‐dimensional model suggests that earthquakes of magnitude 6 or greater on strike‐slip faults that cut the Earth's surface could be predicted using existing instrumentation, unless the ratio of Dc to b−a for natural faults is equal to or smaller than the smallest values investigated by Tse and Rice (1986). Study of three‐dimensional models is needed, since premonitory signals will be smaller if along‐strike variation is considered.

Modelling aftershock migration and afterslip of the San Juan Bautista, California, earthquake of October 3, 1972

The San Juan Bautista earthquake of October 3, 1972 ( M L = 4.8), located along the San Andreas fault in central California, initiated an aftershock sequence characterized by a subtle, but perceptible, tendency for aftershocks to spread to the northwest and southeast along the fault zone. The apparent dimension of the aftershock zone along strike increased from about 7–10 km within a few days of the earthquake, to about 20 km eight months later. In addition, the mainshock initiated a period of accelerated fault creep, which was observed at 2 creep meters situated astride the trace of the San Andreas fault within about 15 km of the epicenter of the mainshock. The creep rate gradually returned to the preearthquake rate after about 3 yrs. Both the spreading of the aftershocks and the rapid surface creep are interpreted as reflecting a period of rapid creep in the fault zone representing the readjustment of stress and displacement following the failure of a “stuck” patch or asperity during the San Juan Bautista earthquake. Numerical calculations suggest that the behavior of the fault zone is consistent with that of a material characterized by a viscosity of about 3.6×10 14 P, although the real rheology is likely to be more complicated. In this model, the mainshock represents the failure of an asperity that slips only during earthquakes. Aftershocks represent the failure of second-order asperities which are dragged along by the creeping fault zone.

Geodetic measurements near Parkfield, California, 1959–1984

Trilateration measurements have been made by various agencies since 1959 near Parkfield, California, in the transition zone between the central creeping section and the southern locked section of the San Andreas fault. Within the transition zone there are several creep meters and alignment arrays, four small‐aperture trilateration networks, and a broad trilateration network extending 70 km southwest from the San Andreas fault to the Pacific coast. Line length changes observed after the 1966 ML = 5.6 Parkfield earthquake indicate a geodetic moment of 4.5 ± 1.0 × 1025 dyn cm. The data are consistent with either 91 ± 8 cm slip on a rupture zone which is 30 km long and extends from 3 to 8 km below the surface or 59 ± 5 cm on a 30‐km‐long zone 2–10 km deep. This moment estimate includes some postseismic slip and is a factor of 2–4 greater than the moment determined from seismic surface waves. The post‐1966 data set is well suited to discriminate between the effects of surface and deep slip. Surface slip rates determined by creep meters, alinement arrays, and small‐aperture trilateration networks decrease from 29 mm/yr at Slack Canyon to zero south of Highway 46; a gradient of roughly −0.48 ± 0.01 mm/yr/km. Deformation of a subset of the trilateration network between Cholame and San Luis Obispo is consistent with shallow slip of 2.9 ± 1.2 mm/yr from the surface to a depth of 16 km and 33.4 ± 5.5 mm/yr below 16 km. The trilateration data alone do not uniquely constrain both the transition depth from shallow to deep slip and the deep slip rate. A transition depth near 16 km is preferred because it provides the best fit to the data and because the predicted slip rate is consistent with independent estimates of the total slip rate across the San Andreas fault. This simple two‐dimensional model neglects along‐strike variations in the deformation. The data show no evidence for significant strike‐slip deformation across faults between the San Andreas and the Pacific coast.

Coseismic steps recorded on creep meters along the San Andreas Fault

Creep meters spanning an active trace of the San Andreas fault in central California recorded many apparent fault-slip steps at times of moderate local earthquakes of magnitudes 4–5. These steps are small, (less than 1 mm) but are recorded over long fault segments (tens of kilometers). The corresponding strain steps, ranging from 10−5 to 10−4, are about 2 orders of magnitude larger than those calculated from the seismically determined source parameters on the assumption of uniform crustal materials. These anomalously large coseismic steps may be due to low effective rigidity of the fault gouge zone or to seismically triggered secondary surface breaks that occurred along the fault.

Short-period nonseismic tilt perturbations and their relation to episodic slip on the San Andreas Fault in central California

Data from four tilt meters and several creep meters along the San Andreas fault in central California 30 km south of Hollister were selected to investigate nonseismic short-period tilt perturbations of minutes to hours duration. The group of short-period tilt amplitude changes with residual deflections was chosen because many nonseismic creep events have similar durations. Although each of the four tilt meters recorded many short-period events, the distribution in space and time suggests that most of these events are due to causes other than slip on the San Andreas fault, such as local strain release. Only at one station used in this investigation (MEL) were creep-related tilt changes observed. A quasi-static dislocation model was constructed to predict the tilt changes associated with propagating slip. Comparing predicted to observed tilt wave forms and amplitudes and comparing the tilt event data to creep event data suggest that (I) a tilt amplitude change of 10 -8-10 -8 rad associated with nonseismic creep events will be observed only if the tilt meter is less than 0.5-1.0 km from the fault, (2) the vertical extent of the slip zone associated with these creep events is greater than about 0.5 km but perhaps not more than a couple of kilometers, (3) nonseismic episodic slip on this part of the San Andreas fault does not appear to propagate uniformly over more than a couple of kilometers, and (4) the model presented in this paper requires 0.5-1.0 mm of dip-slip displacement to reproduce the residual deflection in the creep-related tilt wave forms observed.