Geology topics

James C. Savage

Publications and source records attributed to James C. Savage.

At least 19 recordsLinked to original sources

USGS2024

Rotation of the microplates within the plate boundary in southwestern United States

I investigate the long‐term, rigid motions of the 20 microplates identified by McCaffrey (2005,https://doi.org/10.1029/2004jb003307) within the Pacific‐North America plate boundary in southwestern United States. Those motions are described by the Euler vectors ( Ω i 0 for the i th microplate) given by McCaffrey for each microplate. McCaffrey noticed that the Euler poles for those microplates were aligned along the great circle that connects the geometric center of the microplate distribution with the PACI pole, the pole of rotation of the Pacific Plate (PA) about the North American Plate (NA). To explain that alignment, Thatcher et al. (2016,https://doi.org/10.1002/2015jb0126678.0) proposed replacing each Ω i 0 by two, vertical‐axis rotations of the microplate, one Ω i R describing the trajectory (orbit) of its center of mass (CM) and the other Ω i S its rotation(spin) about that CM, where Ω i R + Ω i S = Ω i 0 . Moreover, they suggested that the orbital motion was being driven by drag from the rotating PA, which suggests that the Ω i R poles coincide with the PACI pole. Then rotation vectors Ω i R and Ω i S consistent with the given Ω i 0 can be found for 17 of the microplates; the other 3 microplates are apparently affected by Basin‐and‐Range extension as well as PA relative motion. The long‐term motion of each of the 17 microplates then can be described as an orbital rotation about the PACI pole plus spin about the CM of the microplate. The closer the microplate CM is to the PA, the more nearly its orbital rotation rate approaches the rotation rate of the PA about the PACI pole.

Arizona, California, Nevada, Utah↗

USGS2021

Comment on ‘Evidence for a large strike-slip component during the 1960 Chilean earthquake’ by H. Kanamori, L. Rivera, and S. Lambotte

Based on numerous studies of the relevant geodetic data, a low-angle thrusting mechanism has been assigned to the 1960 Chile earthquake. Kanamori, Rivera and Lambotte recently suggested that a component of dextral slip comparable to the thrusting be included in the mechanism to satisfy long-period, teleseismic observations. The absence of geodetic evidence for that huge strike-slip component is the subject of this comment. The geodetic data are largely measurements of coseismic uplift associated with the earthquake but include eight measurements of the coseismic change in shear strain. Because strike-slip produces relatively little uplift except near the end points of the rupture, identification of that strike-slip component in the geodetic data depends upon the measured, shear-strain change. I consider elastic, half-space models of oblique slip on the plate interface possibly supplemented by simultaneous dextral slip on the nearby, intra-arc Liquiñe-Ofqui Fault Zone. Slip is assumed to be uniform along strike. The best fits to the geodetic data for these models furnish little evidence for strike-slip on those structures. To satisfy the long-period, teleseismic data, Kanamori et al . proposed six examples, each of which requires a large amount of dextral slip. Because the long-period, teleseismic data do not define the slip distributions, I have used the best fits of those examples to the geodetic data to define those distributions. The large thrusting near the deformation front required by those slip distributions implies large uplift there, contrary to the uplift inferred from the inversion of tsunami data. However, an acceptable fit to the geodetic data and the tsunami data for the six examples suggested by Kanamori et al . can be obtained if the seismic moments specified by them are reduced by a factor ∼1.8, a factor within the uncertainties in estimating seismic moments of the 1960 Chile earthquake. The presence of strike-slip in those reduced-moment examples despite the lack of geodetic evidence for strike-slip is due to a remarkable coincidence that requires careful balancing of contributions from the shallower (depths < 70 km) coseismic sources against those from the deeper coseismic sources to nullify the geodetic evidence for strike-slip. Such balancing is possible, but it is remarkable that the balancing is so nearly perfect that it nullifies the geodetic evidence for strike-slip and thereby confounds the interpretation of the geodetic data.

Geophysical Journal International↗

USGS2018

Euler-vector clustering of GPS velocities defines microplate geometry in southwest Japan

I have used Euler-vector clustering to assign 469 GEONET stations in southwest Japan to k clusters ( k = 2, 3,..., 9) so that, for any k , the velocities of stations within each cluster are most consistent with rigid-block motion on a sphere. That is, I attempt to explain the raw (i.e., uncorrected for strain accumulation), 1996–2006 velocities of those 469 Global Positioning System stations by rigid motion of k clusters on the surface of a spherical Earth. Because block geometry is maintained as strain accumulates, Euler-vector clustering may better approximate the block geometry than the values of the associated Euler vectors. The microplate solution for each k is constructed by merging contiguous clusters that have closely similar Euler vectors. The best solution consists of three microplates arranged along the Nankaido Trough-Ryukyu Trench between the Amurian and Philippine Sea Plates. One of these microplates, the South Kyushu Microplate (an extension of the Ryukyu forearc into the southeast corner of Kyushu), had previously been identified from paleomagnetic rotations. Relative to ITRF2000 the three microplates rotate at different rates about neighboring poles located close to the northwest corner of Shikoku. The microplate model is identical to that proposed in the block model of Wallace et al. (2009, https://doi.org/10.1130/G2522A.1) except in southernmost Kyushu. On Shikoku and Honshu, but not Kyushu, the microplate model is consistent with that proposed in the block models of Nishimura and Hashimoto (2006, https://doi.org/10.1016/j.tecto.2006.04.017) and Loveless and Meade (2010, https://doi.org/10.1029/2008JB006248) without the low-slip-rate boundaries proposed in the latter.

southwest Japan↗

USGS2016

The Eastern California Shear Zone as the northward extension of the southern San Andreas Fault

Cluster analysis offers an agnostic way to organize and explore features of the current GPS velocity field without reference to geologic information or physical models using information only contained in the velocity field itself. We have used cluster analysis of the Southern California Global Positioning System (GPS) velocity field to determine the partitioning of Pacific-North America relative motion onto major regional faults. Our results indicate the large-scale kinematics of the region is best described with two boundaries of high velocity gradient, one centered on the Coachella section of the San Andreas Fault and the Eastern California Shear Zone and the other defined by the San Jacinto Fault south of Cajon Pass and the San Andreas Fault farther north. The ~120 km long strand of the San Andreas between Cajon Pass and Coachella Valley (often termed the San Bernardino and San Gorgonio sections) is thus currently of secondary importance and carries lesser amounts of slip over most or all of its length. We show these first order results are present in maps of the smoothed GPS velocity field itself. They are also generally consistent with currently available, loosely bounded geologic and geodetic fault slip rate estimates that alone do not provide useful constraints on the large-scale partitioning we show here. Our analysis does not preclude the existence of smaller blocks and more block boundaries in Southern California. However, attempts to identify smaller blocks along and adjacent to the San Gorgonio section were not successful.

Eastern California Shear Zone↗

USGS2015

Identifying block structure in the Pacific Northwest, USA

We have identified block structure in the Pacific Northwest (west of 116°W between 38°N and 49°N) by clustering GPS stations so that the same Euler vector approximates the velocity of each station in a cluster. Given the total number k of clusters desired, the clustering procedure finds the best assignment of stations to clusters. Clustering is calculated for k= 2 to 14. In geographic space, cluster boundaries that remain relatively stable as k is increased are tentatively identified as block boundaries. That identification is reinforced if the cluster boundary coincides with a geologic feature. Boundaries identified in northern California and Nevada are the Central Nevada Seismic Belt, the west side of the Northern Walker Lane Belt, and the Bartlett Springs Fault. Three blocks cover all of Oregon and Washington. The principal block boundary there extends west-northwest along the Brothers Fault Zone, then north and northwest along the eastern boundary of Siletzia, the accreted oceanic basement of the forearc. East of this boundary is the Intermountain block, its eastern boundary undefined. A cluster boundary at Cape Blanco subdivides the forearc along the faulted southern margin of Siletzia. South of Cape Blanco the Klamath Mountains-Basin and Range block extends east to the Central Nevada Seismic Belt and south to the Sierra Nevada-Great Valley block. The Siletzia block north of Cape Blanco coincides almost exactly with the accreted Siletz terrane. The cluster boundary in the eastern Olympic Peninsula may mark permanent shortening of Siletzia against the Intermountain block.

California, Idaho, Nevada, Oregon, Washington↗

USGS2015

Strain accumulation across the Prince William Sound asperity, Southcentral Alaska

The surface velocities predicted by the conventional subduction model are compared to velocities measured in a GPS array (surveyed in 1993, 1995, 1997, 2000, and 2004) spanning the Prince William Sound asperity. The observed velocities in the comparison have been corrected to remove the contributions from postseismic (1964 Alaska earthquake) mantle relaxation. Except at the most seaward monument (located on Middleton Island at the seaward edge of the continental shelf, just 50 km landward of the deformation front in the Aleutian Trench), the corrected velocities qualitatively agree with those predicted by an improved, two-dimensional, back slip, subduction model in which the locked megathrust coincides with the plate interface identified by seismic refraction surveys, and the back slip rate is equal to the plate convergence rate. A better fit to the corrected velocities is furnished by either a back slip rate 20% greater than the plate convergence rate or a 30% shallower megathrust. The shallow megathrust in the latter fit may be an artifact of the uniform half-space Earth model used in the inversion. Backslip at the plate convergence rate on the megathrust mapped by refraction surveys would fit the data as well if the rigidity of the underthrust plate was twice that of the overlying plate, a rigidity contrast higher than expected. The anomalous motion at Middleton Island is attributed to continuous slip at near the plate convergence rate on a postulated, listric fault that splays off the megathrust at depth of about 12 km and outcrops on the continental slope south-southeast of Middleton Island.

Journal of Geophysical Research B: Solid Earth↗

USGS2014

Continuous uplift near the seaward edge of the Prince William Sound megathrust: Middleton Island, Alaska

Middleton Island, located at the seaward edge of the continental shelf 50 km from the base of the inner wall of the Aleutian Trench, affords an opportunity to make land-based measurements of uplift near the toe of the Prince William Sound megathrust, site of the 1964, M  = 9.2, Alaska earthquake. Leveling surveys (1973–1993) on Middleton Island indicate roughly uniform tilting (~1 µrad/a down to the northwest) of the island, and GPS surveys (1993–2012) show an uplift rate of 14 mm/a of the island relative to fixed North America. The data are consistent with a combined (coseismic and postseismic) uplift (in meters) due to the 1964 earthquake as a function of time τ (years after the earthquake) u ( τ ) = (3.5 + 1.21 log 10  [1 + 1.67  τ ]) H ( τ ) where 3.5 is the coseismic uplift and H ( τ ) is 0 for τ  < 0 and 1 otherwise. The current uplift on Middleton Island is attributed to continuous slip on a fault splaying off from the megathrust, and the long-term uplift is the superposition of the effects of past earthquakes, each earthquake being similar to the 1964 event. Then, the predicted uplift at time t due to a sequence of earthquakes at times t i would be . From studies of strandlines associated with the uplifted terraces on Middleton Island, Plafker et al. (1992) estimated the occurrence times of the last six earthquakes and measured the present-day elevations of those strandlines. The predicted uplift is in rough agreement with those measurements. About half of the predicted uplift is due to postseismic relaxation from previous earthquakes.

Alaska↗

USGS2013

Clustering of velocities in a GPS network spanning the Sierra Nevada Block, the northern Walker Lane Belt, and the Central Nevada Seismic Belt, California-Nevada

The deformation across the Sierra Nevada Block, the Walker Lane Belt, and the Central Nevada Seismic Belt (CNSB) between 38.5°N and 40.5°N has been analyzed by clustering GPS velocities to identify coherent blocks. Cluster analysis determines the number of clusters required and assigns the GPS stations to the proper clusters. The clusters are shown on a fault map by symbols located at the positions of the GPS stations, each symbol representing the cluster to which the velocity of that GPS station belongs. Fault systems that separate the clusters are readily identified on such a map. Four significant clusters are identified. Those clusters are strips separated by (from west to east) the Mohawk Valley-Genoa fault system, the Pyramid Lake-Wassuk fault system, and the Central Nevada Seismic Belt. The strain rates within the westernmost three clusters approximate simple right-lateral shear (~13 nstrain/a) across vertical planes roughly parallel to the cluster boundaries. Clustering does not recognize the longitudinal segmentation of the Walker Lane Belt into domains dominated by either northwesterly trending, right-lateral faults or northeasterly trending, left-lateral faults.

California;Nevada↗

USGS2013

Clustering of GPS velocities in the Mojave Block, southeastern California

We find subdivisions within the Mojave Block using cluster analysis to identify groupings in the velocities observed at GPS stations there. The clusters are represented on a fault map by symbols located at the positions of the GPS stations, each symbol representing the cluster to which the velocity of that GPS station belongs. Fault systems that separate the clusters are readily identified on such a map. The most significant representation as judged by the gap test involves 4 clusters within the Mojave Block. The fault systems bounding the clusters from east to west are 1) the faults defining the eastern boundary of the Northeast Mojave Domain extended southward to connect to the Hector Mine rupture, 2) the Calico-Paradise fault system, 3) the Landers-Blackwater fault system, and 4) the Helendale-Lockhart fault system. This division of the Mojave Block is very similar to that proposed by Meade and Hager. However, no cluster boundary coincides with the Garlock Fault, the northern boundary of the Mojave Block. Rather, the clusters appear to continue without interruption from the Mojave Block north into the southern Walker Lane Belt, similar to the continuity across the Garlock Fault of the shear zone along the Blackwater-Little Lake fault system observed by Peltzer et al. Mapped traces of individual faults in the Mojave Block terminate within the block and do not continue across the Garlock Fault [Dokka and Travis, ].

California↗

USGS2013

Global earthquake fatalities and population

Modern global earthquake fatalities can be separated into two components: (1) fatalities from an approximately constant annual background rate that is independent of world population growth and (2) fatalities caused by earthquakes with large human death tolls, the frequency of which is dependent on world population. Earthquakes with death tolls greater than 100,000 (and 50,000) have increased with world population and obey a nonstationary Poisson distribution with rate proportional to population. We predict that the number of earthquakes with death tolls greater than 100,000 (50,000) will increase in the 21st century to 8.7±3.3 (20.5±4.3) from 4 (7) observed in the 20th century if world population reaches 10.1 billion in 2100. Combining fatalities caused by the background rate with fatalities caused by catastrophic earthquakes (>100,000 fatalities) indicates global fatalities in the 21st century will be 2.57±0.64 million if the average post-1900 death toll for catastrophic earthquakes (193,000) is assumed.

Earthquake Spectra↗

USGS2012

Coulomb plasticity within the fault zone

We represent a well‐developed fault by a layer of granular material (fault gouge) confined between two competent fault blocks. Slip on such a fault involves plastic shearing of the fault gouge. That is, the fault gouge behaves as a Coulomb material, and the plastic flow is accomplished by slip on the two sets of Coulomb shears appropriate to the stress state and the frictional properties of the gouge. Neither set of Coulomb shears is coplanar with the fault zone. Generally, the initial plastic flow is self arresting, and increased shear stress is required to drive further shear. But as the shear stress is increased a stress state is reached within the gouge in which the plane of maximum shear stress is parallel to the plane of the fault zone. Then the fault gouge shears at whatever rate is required to keep the shear stress across the fault from increasing further as was demonstrated experimentally by Mandl et al. [1977]. In a mature fault previous slip cycles have presumably left the fault gouge in a stress state such that the plane of maximum shear is parallel to the fault plane. Then, with additional shear loading, unlimited shear deformation is possible at the shear stress corresponding to the onset of Coulomb failure, and the fault appears to slip as if friction on the plane of the fault were less than on planes of other orientations. Slip on the fault does not imply that the fault plane coincides with a Coulomb shear as is sometimes assumed.

Geophysical Research Letters↗

USGS2012

Probability of one or more M ≥7 earthquakes in southern California in 30 years

Eight earthquakes of magnitude greater than or equal to seven have occurred in southern California in the past 200 years. If one assumes that such events are the product of a Poisson process, the probability of one or more earthquakes of magnitude seven or larger in southern California within any 30 year interval is 67% ± 23% (95% confidence interval). Because five of the eight M ≥ 7 earthquakes in southern California in the last 200 years occurred off of the San Andreas fault system, the probability of one or more M ≥ 7 earthquakes in southern California but not on the San Andreas fault system occurring within 30 years is 52% ± 27% (95% confidence interval).

Southern California↗

USGS2012

Strain accumulation north of Los Angeles, California, as a function of time, 1977–1992

No significant change in the rate of strain accumulation in a 40×120 km trilateration network spanning the San Gabriel mountains was observed from 1977.5 to 1991.8 despite an apparent increase in seismicity ( M L > 4.5) beginning in late 1987 in the northern Los Angeles basin immediately to the south. The observed deformation (0.13±0.01 µstrain/yr right‐lateral shear across a vertical plane striking N63°W±1°) can be attributed to strain accumulation on the San Andreas fault (local strike N65°W). No significant accumulation of compression normal to the strike of the San Andreas fault was observed in the network as a whole.

California↗

USGS2012

Changes in long‐term extension rates associated with the Morgan Hill and Loma Prieta earthquakes in California

Frequent measurements since mid‐1981 of the distances from a geodetic monument located about 100 km south‐southeast of San Francisco to three monuments 30 to 40 km distant provide an unusually complete record of the deformation before and after two nearby earthquakes, the 1984 Morgan Hill ( M L = 6.2) and 1989 Loma Prieta ( M s = 7.1) earthquakes. Except possibly for the first few months postseismic, the extension rates indicated by these measurements appear to be steady over the four or five years both preceding and following those earthquakes. However, the preseismic and postseismic rates differ significantly for at least one of the baselines measured for each earthquake. The data over the four to five year postseismic records available are not adequate to demonstrate whether the postseismic rates are relaxing back to the preseismic rates.

California↗

USGS2012

Strain‐rate profile across the Elsinore, San Jacinto, and San Andreas Faults near Palm Springs, California, 1973‐81

A profile of the strain accumulation rate along a line trending N50°E across the subparallel Elsinore, San Jacinto, and San Andreas faults near Palm Springs, California, has been constructed from trilateration surveys in the 1973‐81 interval. The strain accumulation is principally right‐lateral shear across a vertical plane parallel to fault strike (N40°W). The strain rate profile for that component exhibits two clearly resolved maxima, one centered on the San Jacinto fault (γ max = 0.35 ± 0.02 µrad/a) and the other on the San Andreas fault (γ max = 0.40 ± 0.02 µrad/a); no maximum is associated with the Elsinore fault. This result clearly implies that slip at depth on both the San Andreas and San Jacinto faults is loading the shallower sections of these faults, and that eventually rupture can be expected on both.

California↗

USGS2012

A possible geodetic anomaly observed prior to the Loma Prieta, California, Earthquake

Monthly measurements since mid‐1981 of distance from a geodetic station located 11 km from the epicenter of the Loma Prieta earthquake ( M s = 7.1; October 17, 1989) to three stations 30 to 40 km distant provides an unusually complete record of deformation in the epicentral region in the years prior to an earthquake. Roughly 1.3 years before the earthquake, at about the time of the first magnitude‐5 foreshock, the rate of change in line length for two of the lines appears to change; the rate for the third line does not change. Other similar, though smaller, changes in rate are apparent in the eight‐year record. Thus, there is marginal evidence for a change in deformation rate about one year before the Loma Prieta earthquake, but that change need not be a precursor.

California↗

USGS2012

Geodetic estimate of coseismic slip during the 1989 Loma Prieta, California, Earthquake

Offsets in the relative positions of geodetic stations resulting from the Loma Prieta earthquake can be explained with a dislocation model that includes buried oblique slip on a rupture surface extending 37 km along the strike of the San Andreas fault, dipping 70° to the SW, and extending from a depth of about 5 to 17.5 km. Assuming uniform slip on a rectangular surface, the mean values for a range of reasonable fault geometries are 1.6 ± 0.3 m right‐lateral strike slip and 1.2 ±0.3 m reverse slip. Slip on an adjacent extension of the rupture to the southeast, recorded in the aftershock sequence, is not well constrained by the geodetic data. The geodetic data clearly preclude rupture extending near the surface.

California↗

USGS2012

An apparent shear zone trending north‐northwest across the Mojave Desert into Owens Valley, eastern California

Strain rates measured at four geodetic networks in eastern California situated between northern Owens Valley and the Transverse Ranges along a small circle drawn about the Pacific‐North America pole of rotation are remarkably consistent. Each exhibits 0.14 μrad/yr simple right‐lateral engineering‐shear‐strain accumulation across the local vertical plane tangent to the small circle. Local faults ( e.g. , Owens Valley, Garlock, Helendale) traversing these networks are not as closely aligned with the vertical planes of maximum shear‐strain accumulation as is the local tangent to the small circle. A fifth network slightly east of the small circle shows no significant strain accumulation. Thus, a shear zone trending N35° W from near the eastern end of the big bend of the San Andreas fault to northern Owens Valley is indicated by these data. This corresponds to the Eastern California shear zone proposed on geological evidence by Dokka and Travis. The shear zone carries ∼8 mm/yr of the Pacific‐North America relative plate motion from the San Andreas fault north‐northwest across the Mojave Desert into Owens Valley and the northern Basin and Range province. The shear zone observed at the surface may be a manifestation of a through‐going subcrustal fault.

California↗