G13C-01 13:40h
Strain Partitioning and Accumulation across Overlapping Spreading Centers: Geodetic GPS Measurements in South Iceland
Overlapping spreading centers (OSCs) and propagating ridges are important classes of mid-ocean ridges. Kinematic models of OSCs predict along strike variability in spreading rate associated with the propagation of one center and deactivation of the other. Iceland offers a unique opportunity to investigate strain accumulation and partitioning across slow, overlapping spreading centers, and the influence of a ridge centered hotspot on ridge kinematics and morphology. We present results of detailed GPS observations across the Eastern and Western Volcanic Zones, south Iceland, spanning a seven to nine year inter-rifting period, and compare our observations with two-dimensional elastic half-space models that simulate the long-term spreading process. We then compare the elastic half-space models with simple viscoelastic coupling models. We model three velocity profiles across the EVZ-WVZ system, solving for the spreading rate, locking depth and horizontal location of each spreading center. Our spreading rate estimates indicate along strike variations as expected in an OSC system and total spreading rates consistent with geodetic and geologic plate motion models. Spreading rates in the WVZ increase from northeast (3 $\pm$1 mm/yr) to southwest (7 $\pm$1 mm/yr). Spreading rates in the southwest propagating EVZ decrease from northeast (17 $\pm$1 mm/yr) to southwest (12 $\pm$1 mm/yr). These results are consistent with a model whereby the WVZ is deactivating in the direction of EVZ propagation. The morphology of the two spreading centers reflects the spreading rate differences and their location relative to the Iceland hotspot. The predicted locations of the spreading axis for each zone are consistent with mapped Holocene fissure swarms. The neovolcanic zone of the slower WVZ consists of a narrow (10-20 km wide) axial graben and has had few Holocene eruptions. The faster EVZ consists of two parallel neovolcanic zones separated by a 20 km gap of inactivity, little normal faulting, higher topography and five historical fissure eruptions, reflecting its proximity to the hotspot. The maximum velocity gradient in the EVZ is located on the Veidivotn fissure swarm, which had a small volume eruption in 1864. The last major fissure eruption in the EVZ was the 1783 Lakagigar, located 20 km to the east. This pattern of current and past strain accumulation and release suggests intra-ridge jumping of activity and crustal accretion across a 60 km wide area.
G13C-02 13:55h
Description of the Active Tectonic Deformation and Fault Friction in Iran Using Block and 3D Finite Element Modeling
The intracontinental deformation involved in the eastern part of the Arabia/Eurasia collision zone (i.e., mainly Iran) is distributed among several orogens (Zagros, Caucasus, Alborz, Kopet-Dag) surrounding rigid blocks (South Caspian, Central Iran, Lut, Helmand). In this study we investigate the more appropiated desciption of the active tectonic deformation in Iran using different modeling approaches. To do so, we have developed models of the eastern part of the Arabia/Eurasia collision ranging from eastern Turkey (~43°E) to western Afghanistan (~63°E) and from the Arabian to the Eurasian plate. The available GPS velocity fields (Vernant et al., 2004) are used to constrain (block model) or set-up the boundary conditions (finite element modeling). Two different rheologies are used for the finite element modeling: (a) one visco-elastic layer, (b) two layers: the upper one corresponds to the crust and the lower one to the lithospheric mantle. The great strike-slip faults are treated as vertical material discontinuities of the mesh. Fault slip-rates ranging from 0.3 to 0.02 are controlled by a Coulomb-type friction. A third rheology is used corresponding to the block model: (c) elastic constitutive law with fault friction = 0. The rms of the residuals (observed - modeled velocity field) are compared for the different experiments. The statistics obtained are close for the three different rheologies implying that the GPS data are too sparse to definitely conclude about the best description of the active deformation in Iran. However, elasto-plastic or elastic rheologies with law fault frictions show a slightly better fit to the GPS data. This is confirmed by the comparison between the geological fault slip rates and the modeled ones, suggesting that the active tectonic deformation in Iran can be described using rigid blocks and very low fault slip rate (0.02 to 0.0). However, more GPS measurement need to be done to conclude if the deformation is only accommodated by rigid blocks and faults or if elasto-plastic deformation zones exist somewhere in this part of the Alpine-Himalayan mountain belt.
G13C-03 14:10h
Master Cycles of Land-Level Change and Great-Earthquake Recurrence in Cascadia, Northeast Japan, and Southern Chile
In examples from three subduction zones, events of the past few thousand years settle apparent conflicts between rates averaged across shorter and longer periods of time. At these zones the major plate-boundary earthquakes recur in cycles too long to measure with historical and instrumental records alone. CASCADIA. The Pacific coast of Washington State rose faster in the past 100 years than it did, on average, in the past 125,000 years. This difference became apparent a quarter century ago, when coastal uplift rates were estimated geodetically by leveling and tide gauging. The short-term rates were found to exceed, by an order of magnitude, long-term rates shown by Pleistocene marine terraces. Ensuing work by many investigators revealed a Holocene history of coseismic subsidence, not only in Washington but also in British Columbia, Oregon, and California. The subsidence, at irregular intervals averaging about 500 years, mostly negates gradual uplift like that measured geodetically. The Pleistocene terraces thus follow a sawtoothed trajectory, ten steps gradually up and nine steps abruptly down. This resolution of a geodesy-vs-geology conflict laid groundwork for today's consensus that Cascadia produces great plate-boundary earthquakes. JAPAN. Short-term subsidence seemingly conflicts with long-term uplift along the southern Kuril trench. Eastern Hokkaido subsided 1 m in the past 100 years (tide gauges, bench marks) even though it rose 20-50 m in the past 125,000 years (marine terraces). Japanese geophysicists recognized the conflict over three decades ago. Some predicted that a moment-magnitude 7.8 earthquake in 1973 would resolve it. But no onshore uplift occurred during the earthquake and less than 0.1 m of uplift followed. A M 8.0 earthquake in 2003 similarly left the region's long-term uplift unexplained. However, eastern Hokkaido has a Holocene history of episodic uplift. The episodes, identified by Yuki Sawai and his coworkers, are recorded by forest peat that overlies estuarine mud. The region rose about 1 m during individual episodes. The cause of uplift is being investigated. The prime suspect is postseismic deformation induced by plate-boundary earthquakes of M ~8.5 or larger. Such events are unknown in the region's 200 years of written history.. CHILE. The M 9.5 1960 earthquake seemingly overspent its plate-tectonic budget. Its slip, 20 m on average, expended 250 years of plate motion at the NUVEL-1 convergence rate and over 500 years of plate motion at convergence rates recently estimated by campaign GPS. Yet Chile's written history contains three predecessors to the 1960 event -- in 1837, 1737, and 1575 -- for recurrence intervals averaging just 128 years. When did the 1960 slip accumulate? Marco Cisternas and his coworkers are finding clues at an estuary midway along the 1960 rupture. Sand sheets on its margins record subsidence, a tsunami, or both from 1960, 1575, and six preceding events of the past 2000 years. Consistently absent are signs of the 1737 and 1837 events; these apparently failed to break much if any of the central part of the 1960 rupture area. In that case, infrequent giants like the 1960 earthquake expend most of the plate convergence in south-central Chile.
G13C-04 14:25h
Tenfold Disparity Between Decadal InSAR and Millennial Morphochronologic Slip-Rates on the Karakorum Fault
The role of the Karakorum Fault in the kinematics of present-day deformation in Tibet is debated. Although it is the main dextral strike-slip fault north of the Himalayas, recent InSAR data are interpreted to suggest that it is barely active, moving at a rate of 1 $\pm$ 3 mm/yr. Surface exposure dating of moraines and terraces south of Bangong lake tell a different story, suggesting it slips ten times as fast. At one site on the west side of the Gar graben (32$\deg$3' N-80$\deg$1' E, 4365 m-4760 m), at the foot of the Aliyari Range, we sampled quartz-rich, rooted blocks on the lateral moraine crests of a large till complex. The moraines are offset right-laterally relative to the valley of the Manikala Daer Glacier. Twenty-seven samples were dated using cosmogenic 10Be. The offsets of 2 well-defined moraine crests (M1 and M2E), obtained from retro-deformation of 1 m-resolution IKONOS images, are 220 $\pm$ 10 m and 1520 $\pm$ 50 m, respectively. The 10Be exposure ages fall in distinct clusters: 21 $\pm$ 0.1 ka and 40 $\pm$ 3 ka on M1, and 140 $\pm$ 5.5 ka and 180 $\pm$ 14 ka on M2. The ages of maximum sample abundance correlate remarkably well with the ages of the coldest periods derived from proxy paleo-temperature records (SPECMAP), suggesting that the Manikala glacier crossed the fault to emplace moraines northeast of the range front only during the LGM and penultimate glacial maximum (late stage 6), and implying little erosion. Pairing the abandonment ages of the two moraines ($\sim$ 21 and 140 ka) with their offsets yields concordant dextral slip-rates of 10.9 $\pm$ 0.6 and 10.5 $\pm$ 0.5 mm/yr. The resulting, average slip rate (10.7 $\pm$ 0.7 mm/yr) we derive for the last 150 ka is comparable to the geological rate obtained by Lacassin et al., [2004] (10 $\pm$ 3 mm/yr in the last 34 Ma) and to the GPS geodetic rate determined by Banerjee and Burgmann, [2002] (11 $\pm$ 4 mm/yr). It is almost 3 times greater than that measured in India by Brown et al., [2002] (4 $\pm$ 1 mm/yr) and three times smaller than that predicted by Liu Qing [1993] (30 mm/yr). The $\sim$ 11 mm/yr slip-rate along the Ayilari range should be considered a minimum because the site is situated on one side of the Gar graben, and at least two branches of the Karakorum fault system splay off into that graben north of Gar. Similar morphochronologic evidence at other sites north and south of Gar yields results in keeping with a millennial slip-rate of at least 1 cm/yr on the fault, implying that extrusion of Tibet between the converging Indian and Tarim plates at a comparable rate is not in doubt. That the GPS and InSAR values differ by a factor of ten suggests that strong monsoonal tropospheric effects might bias the InSAR data. If the tenfold disparity between decadal and millennial rates is real, then it might reflect "go/stop/go" slip-rate fluctuations, over periods comparable to or longer than the seismic cycle, related to the mechanical behavior of the fault zone integrated between the base of the crust and the surface.
G13C-05 14:40h
GPS-Constrained Microplate Kinematics and Plio-Pleistocene Tectonic Evolution of the North Anatolian Fault and North Aegean Sea
Emerging evidence from Global Position System (GPS) survey measurements in the Aegean and elsewhere suggests that present-day active continental deformation occurs largely due to the relative motions of a small number of rigid blocks or microplates. However, it is not universally agreed whether the continental microplate description of the GPS data is superior to other proposed models, nor is it clear whether present-day movement patterns can be usefully extrapolated into the geologic past. Here I examine the known deformation history of the North Aegean over the past ~10 Ma and compare it with predictions based on the present-day microplate model. Agreement provides independent support for the GPS-based model and demonstrates its value in bettering our understanding of Aegean tectonics. If we knew nothing about late Cenozoic North Aegean tectonics and provisionally assumed the correctness of the Aegean microplate model of Nyst & Thatcher [2004 JGR], we would predict several features of the tectonic evolution that accord with geologic evidence. First, the North Aegean Sea would be created by extension due to SSW motion of the South Aegean and concomitant CW rotation of central Greece during the past 10 Ma. The same kinematic process would cause extension to be succeeded by strike-slip motion as the `ridge-transform-ridge' triple junction migrates WSW 24 km/Ma and the North Anatolian fault propagates into the region. The Plio-Pleistocene history of the North Aegean shows these same general features. Drilling and seismic imaging document the existence of young (< 10 Ma) and thick (up to 6 km) sedimentary sequences attributed to crustal extension by a factor of 3-4. Seismic profiling and bathymetric mapping show a mesh of roughly orthogonal faults with dip-slip offsets. Structural studies of sub-aerial exposures of these faults suggest an earlier episode of extension was followed by predominantly strike-slip motions. The plate kinematic reconstruction of late Cenozoic tectonics thus has several strengths but some clear limitations. Using only the GPS constrained microplate model and no adjusted parameters, backwards extrapolation over the past 10 Ma explains the observed large scale Plio-Pleistocene extension of the North Aegean and its evolution towards present-day strike-slip tectonics via the much-discussed propagation of the North Anatolian fault into the North Aegean. The spatial distribution of Plio-Pleistocene sediments and large-offset normal faults is considerably more complex than the simple extensional faulting pattern expected from the plate kinematic extrapolation. The detailed pattern of Plio-Pleistocene extension is compatible with several abrupt jumps in the principal zone of extension from NE to SW with time, which would account for the localized sediment accumulations in the Kavala and Thermaikos basins and the smaller than expected offset of the North Anatolian fault in the Dardanelles. The relative youth of extension in the Gulf of Corinth indicates the current geometry and motion of the Central Greece microplate is correspondingly recent, so the extrapolated rotation of this block during the past 10 Ma cannot be correct. Extension prior to about 1 Ma BP must have occurred elsewhere, perhaps in the central and southern Aegean.
G13C-06 14:55h
Strain Partitioning During Oblique Divergence: a Key Concept in Matching Geodetic, Seismological and Geological Datasets.
A quantitative analysis of geodetic datasets shows that zones of oblique deformation are broadly diffuse either at regional or local scale. Oblique relative plate motion (i.e. transtension or transpression) inevitably leads to 3-D rotational strains, even in the simplest case of homogeneous deformation. However, many existing models of plate boundary deformation are based either on the assumption of 2-D plane strain and/or use systems of rigid rotating blocks overlying a viscous continuum at depth. In this presentation, we demonstrate that the deformation of obliquely divergent plate margins is typically heterogeneous and partitioned into fault-bounded domains. As a result, a more complex relationship exists between geodetic, seismological and geological datasets compared to orthogonal or transform margins. For example, seismicity associated with the N-S trending Dead Sea transform fault is spatially partitioned, with sinistral transcurrent events - concentrated in the centre of the basin - and oblique extensional events within the adjacent basin margins. Outcrop studies of active faults reveal a similar pattern and when integrated with the seismological data yield a NNE-oriented divergent plate motion vector oriented 10° clockwise of the plate margin. Recent GPS data confirm that a minor component of extension occurs oblique to the margin, with velocity vectors rotating clockwise away from the plate boundary. Similarly, in the Basin and Range Province abrupt changes in the geodetically-derived velocity fields (Oldow, 2003, Geology) are consistent with the changes in the distribution and character of seismicity and patterns of exposed active faults. Deformation is partitioned with different components of displacement being accommodated in a series of fault-bounded crustal blocks that can be collectively integrated to derive the far-field plate tectonic velocity field. We conclude that obliquely divergent margins are characterized by heterogeneous patterns of bulk 3-D strain in which adjacent fault-bounded domains deform internally, accommodating differing proportions of margin-parallel and margin-normal displacement. Key controlling variables are the angular relationships between plate motion vectors, plate boundaries and pre-existing intraplate margin faults. Our findings also illustrate that the matching of geodetic, earthquake and geological data is not always a straightforward procedure, since the results may yield velocity fields that reflect local, domainal strains that are not necessarily of regional, plate-scale significance.
G13C-07 15:10h
Kinematics of the southern Alaska constrained by westward-decreasing post-glacial slip-rates on the Denali fault, Alaska.
Cosmogenic exposure age analyses of offset glacial moraines show that the Holocene slip-rate of the arcuate right-lateral Denali fault decreases westward from about 13 mm/yr at $144^{\circ}$W to about 7 mm/yr at $149^{\circ}$W. This pattern is consistent with a tectonic model in which the southern part of Alaska (the Wrangell Block) is translating northward relative to northern Alaska (North American plate) without rotation. It also implies that convergence rates across the Alaska Range's northern frontal thrust fault system increase significantly from east to west from east to west from about 5 to 14 mm/yr. Slip distribution along the Denali fault during the great 2002 earthquake was decidedly asymmetric. This distribution of dextral slip along the 220-km rupture - decreasing from about 10 m in the east to about 3 m in the west - suggested that the long-term slip rate on the Denali fault might also decrease westward. If true, this would have major implications for the kinematics of southern Alaska, in general, and the Alaska range, in particular. If the Wrangell block is rotating counter-clockwise, the slip-rate on the arcuate Denali fault would be roughly uniform along strike. If however, southern Alaska is translating northwestward without rotation, the slip-rate magnitude should diminish westward. To constrain these models, we sampled offset moraines for $^{10}{Be}$ cosmogenic exposure dating at two sites along the Denali fault. At Slate Creek (\approx$144^{\circ}$W), we sampled boulders, both upstream and downstream from the fault, on a moraine offset 156$\pm$20 m. The minimum exposure ages range from 7.8$\pm$0.5 to 12.3$\pm$0.8 ka and average 10.7$\pm$1.3 ka. About 235 km to the west, at Bull Creek (\approx$149^{\circ}$W), we collected large cobbles downstream from the fault, on a moraine offset 82$\pm$5 m. There, the minimum exposure ages range from 9.8$\pm$0.8 to 12.3$\pm$0.9 ka and average 11.0$\pm$0.8 ka. We corrected these minimum exposure ages for snow shielding, using historical snow cover data averaged over 30 years near our sites. Furthermore, to estimate the variation of snow cover over time, we included the changes in effective wetness over the last 11 ka, calculated from lake-level records and $\delta^{18}O$ variations from lakes north of the Alaska range. The time-averaged snow corrections increase the ages of the moraines by 9$\pm$0.3% at Slate creek and 12$\pm$0.3% at Bull creek. Division of the corrected ages into the offsets yields slip-rates of 13.2$\pm$4 mm/yr at Slate creek and 6.6$\pm$1.7 mm/yr at Bull Creek. Thus it is clear that the long-term rate of slip on the Denali fault decreases appreciably from east to west. These values are consistent with a kinematic model in which southern Alaska is translating northwestward without rotation at a rate of about 15 mm/yr. The rates also suggest progressive slip partitioning between the Denali fault and the active fold and thrust belt at the northern front of the Alaska range, with rates of convergence increasing westward from ~5 to 14 mm/yr between ~149 and $144^{\circ}W.
G13C-08 15:25h
Denali Fault Earthquake Cycle Models
Although fault slip rates are often estimated from geodetic data such as GPS velocities, "geodetic fault slip rates" depend on the assumptions in the earthquake cycle model used. The most commonly-used model is a simple elastic half-space model with a buried dislocation, but in this model there is no accounting for the time-varying nature of deformation through the earthquake cycle, including postseismic deformation. Clearly the earth is more complicated. Pre-earthquake (1993-2002) GPS velocities are best fit by a fault slip rate of 6-8 mm/yr with a poorly constrained locking depth (Fletcher, 2002, Ph.D. thesis, University of Alaska Fairbanks). Some paleoseismic data appear consistent with the same rate but others suggest a rate as much as twice as high. Furthermore, new paleoseismic data are becoming available that constrain the earthquake history and further refine the long-term slip rate of the fault. I investigate alternative earthquake cycle models that account for postseismic relaxation following large events in an attempt to reconcile the paleoseismic and geodetic slip-rate estimates, and to determine a robust long-term slip rate for the central Denali fault. The 2002 Denali fault earthquake provides an excellent opportunity to investigate the validity of such alternative earthquake cycle models, and to better constrain model parameters. Postseismic displacements at a number of sites have exceeded 15 cm within the first two years, and the postseismic deformation is long wavelength and appears to be dominated by a deep source such as viscoelastic relaxation in the upper mantle. The postseismic response to the earlier 1964 Alaska earthquake provides further constraints on regional rheological parameters.