G32A-01 10:20h
Delineating Block Boundaries of the Earth's Crust in the Pacific Northwest from GPS and Seismicity
GPS measurements indicate that much of the deformation in the Pacific Northwest occurs from the clockwise rotation of the "Oregon block" about a rotation pole located near the eastern end of the Oregon-Washington boundary. Strain rates of less than 1 nanostrain per year in Oregon (corrected for strain due to subduction along the Pacific coast)establish that the Oregon block is relatively rigid as it rotates at 0.8 degrees per million years. The Oregon block may also include part of southwest Washington but in the rest of Washington the rigid block model breaks down and there appears to be significant strain and north-directed compression in the Puget lowland. The details are likely important to understand the nature of earthquake hazards in the Pacific Northwest. In 2004 we reoccupied 60 GPS sites in northeast Oregon, eastern Washington and northern Idaho in order to better define the interaction of the rotating Oregon block and eastern Washington. We focused on sites first surveyed using high precision GPS in 1998 and reoccupied in 2001. Computed velocities at these sites have uncertainties of order 0.8mm/yr or less and indicate a marked change in velocity across the northwest trending Olympic Wallowa lineament, a major tectonic feature cutting through Washington and northeast Oregon. The recomputed velocities east of the Cascade mountains indicate that northeast Washington and northern Idaho move with the rest of stable North America and the observed strain pattern may also explain the increased earthquake activity in central Washington compared to central Oregon.
G32A-02 10:35h
Major subsidence of the south-central United States of America and future inundation of coastal areas
The northern shore of the Gulf of Mexico is the site of America's greatest wetland, the gateway to vast energy resources, and home to over 10 million people. This critical area is being increasingly threatened by progressive inundation by the relative rise of the Gulf of Mexico. This slow inundation was detected several decades ago and has been generally attributed to eustatic sea level rise, sediment starvation of the delta due to construction of flood control levees along the Mississippi River, and subsidence of the land relative to sea level. Although the former two effects are reasonably well understood, the lack of precise quantitative spatial data on the later related to a well defined, common datum has prevented the development of a satisfactory theory to explain modern surface motions. Analysis of National Geodetic Survey (NGS) 1st order leveling data produced vertical velocities for over 2700 benchmarks in Louisiana, Mississippi, Alabama, Texas, Arkansas, Florida, and Tennessee. All motions were related to NAVD88 and show that subsidence is not limited to coastal wetland areas, but rather includes the entire coastal zone as well as inland areas several hundred km from the shore. Subsidence can also be tracked to the north and follows the trend of the alluvial valley of the Mississippi River. Regionally, vertical velocities range from less than -30 mm/yr along the coast to over +5 mm/yr in peripheral areas of eastern Mississippi-Alabama. The mean rate is ~11 mm/yr in most coastal parishes of Louisiana. In the Mississippi River deltaic plain, subsidence was significantly higher than previous estimates based on long-term geologic measurements. The data also indicate that adjacent alluvial ridges where the population is concentrated have been similarly affected. In the Chenier plain of southwest Louisiana, a region previously thought to be subsiding at slowly, rates of sinking are similar to those of the deltaic plain. Demonstration that all areas of the coastal landscape as well as inland areas are affected implies that subsidence recorded by benchmarks is not solely due to local sedimentary processes and/or the activities of humans. Instead, geodetic data when integrated with subsurface geologic information suggest that subsidence includes a strong regional component that is the product of lithospheric flexure and normal faulting. This component is mainly due to the derivative effects of late Quaternary sediment loads such as the modern Mississippi River delta and Pleistocene deposits offshore. Models of simple flexure are inadequate, however, to explain the regional component of subsidence. Instead, it is proposed that active faulting plays a key role in regional subsidence throughout the coast by episodically weakening the lithosphere, which in turn changes the way that the lithosphere bears the load of sediments over time. Salt intrusion/evacuation induced by loading is a major cause of subsidence in southwest Louisiana. If subsidence continues at similar rates and construction efforts fail to build protection levees to appropriate heights, substantial portions of the gulf coast (primarily Louisiana) will lie below sea level and be inundated by end of this century. In Louisiana, this will result in a loss of ~$140B of land and property, as well as the land, livelihoods, and cultural heritage of over 2 million people.
G32A-03 10:50h
Testing intraplate deformation in the North American plate interior from a combined geodetic solution: implication for strain accumulation on potentially seismogenic faults in the central and eastern U.S.
Large earthquakes within stable plate interiors are direct evidence that significant levels of strain can accumulate along geologic structures far from plate boundary faults. The 1811-1812 M7-8 New Madrid events in the Mississippi valley (central U.S.) are classical examples of such large intraplate earthquakes. Quantifying the associated long-term strain geodetically remains a challenge because the expected rates are close to the magnitude of GPS errors. In order to improve the accuracy and signal-to-noise ratio of GPS estimates of crustal strain in stable North America, we have combined five independent geodetic solutions covering some or all (324 sites total) of the area. Residual velocities of the combined solution with respect to stable North America show an average weighted misfit of +-1mm/yr (1 standard deviation), with preliminary evidence that the misfits are relatively insensitive to monumentation quality. We find no evidence for regions of significant strain rates (at the accuracy level of our measurements). Our velocity field instead defines a coherent residual velocity pattern in the upper Midwest and New England that may reflect post-glacial rebound effects. We discuss the implications of these results for earthquake hazard, glacial isostatic adjustment models, and for defining a Stable North America reference frame for geodetic studies in western North America.
G32A-04 11:05h
Satellite radar interferometry time series analysis of surface deformation for the metropolitan Los Angeles and San Francisco, California, areas
The Los Angeles and San Francisco, California, metropolitan areas are tectonically active regions with surface deformations that are a combination of fault related tectonics plus a variety of natural and anthropogenic signals, most notably aquifer changes and oil extraction. We produce interferometric SAR (InSAR) time series analyses for these areas for 1993 into 2002 and determine space-time maps of surface deformation at each ERS epoch for which we have data. The results are space-time deformation products that can be exploited to view not only the mean or smoothly varying long-term surface motion, but also its time varying patterns. Large seasonal oscillations of the Santa Ana aquifer observed in Southern California Integrated GPS Network (SCIGN) data are accurately matched in the InSAR time series, moreover, correlations of the InSAR time series with an annual sinusoid reveals amplitude and delay-time patterns that reflect the dynamics of the hydrologic system. Similar patterns can be found for the San Francisco area. We will explore these features and their implications for detecting earthquake related transient deformation.
G32A-05 11:20h
Variation in aseismic slip and fault normal strain along the creeping section of the San Andreas fault from GPS, InSAR and trilateration data
In central California most of the relative motion between the Pacific and North American plates is accommodated by strike slip along the San Andreas fault system. However, a small amount of convergence is accommodated by compressional structures in the California Coast Ranges on both sides of the fault. Recent examples of such activity are the Coalinga and the 2003 San Simeon earthquakes. Along the central San Andreas fault (CSAF), from San Juan Bautista to Parkfield, almost all the slip along the CSAF in the brittle upper crust is accommodated aseismically. We use GPS, InSAR and trilateration data to resolve both the distribution of aseismic slip along the CSAF, and the deformation across adjacent, secondary fault structures. In 2003 and 2004, we conducted several GPS surveys along the CSAF. We resurveyed 15 stations of the San Benito triangulation and trilateration network, which extends 40 km to the northeast of the creeping segment. We combine these measurements with old EDM measurements and data from a GPS campaign in 1998. We also occupied 13 sites along the creeping segment, for which previous data exist in the SCEC archive. These dense GPS measurements, along with data from permanent GPS stations in the area, allow us to constrain the regional strain distribution and contributions from adjacent faults. With the addition of InSAR data, we can also better resolve active strain accumulation and aseismic slip along the CSAF. We use a stack of about 10 interferograms from ERS-1 and ERS-2 satellites spanning 8 years. InSAR is well suited to monitoring details of the shallow slip along the CSAF and, in concert with the broadly spaced GPS velocities, to resolving the distribution of deformation along and across the plate boundary. The results are the basis for determining the kinematics of spatially variable fault slip on the CSAF, and help to better constrain the fault's constitutive properties, and fault interaction processes.
G32A-06 11:35h
Creep on the San Andreas fault near San Juan Bautista and its relationship to large historic earthquakes
The San Juan Bautista segment of the San Andreas fault is an area of moderate seismicity that forms the transition zone between the creeping section and the locked Santa Cruz segment. Here the San Andreas fault experiences active surface creep and transient slip events (slow earthquakes), which have been measured with creepmeters, strainmeters and short baseline trilateration arrays since the 1960s. The fault frictional properties that allow creep are thought to also prevent the nucleation and inhibit the rupture propagation of large earthquakes. In fact, the largest instrumentally recorded earthquakes on the San Juan Bautista segment have been between M5.3-5.5. However, between 1840 and 1899, historic records suggest six M$\geq$6 earthquakes ruptured the San Juan Bautista segment. These earthquakes occurred during a time of intense seismic activity in the Bay Area before the 1906 San Francisco earthquake, and the absence of more recent similar events may be attributable simply to the stress shadow produced by the 1906 quake. On the other hand, the San Juan Bautista segment experiences slip transients on time scales from days to years and it is possible that its rate and distribution of slip is highly time variable; for example, in relation to the regional stress state. We examine the current creep conditions on the San Juan Bautista segment using space geodetic data to determine whether they would allow a similar size and rate of earthquake production as seen in the historic record. We perform a joint inversion of GPS and InSAR data on distributed fault patches using an elastic half space model to determine the distribution of interseismic creep. We use GPS data from the B\={A}V\={U} dataset spanning 1994-2003. B\={A}V\={U} contains a combination of campaign and continuous data collected by several agencies and processed in a uniform manner using GAMIT/GLOBK. The InSAR data is taken from a stack of nine interferogram pairs made using ERS1 and ERS2 data collected between 1995-2001. We use a stack of small time span (less than 1.25 years) pairs because interferograms in this area become completely decorrelated when they span much more than two years. We choose the stacked interferograms such that the slave in one pair is the master in the next to remove the atmospheric errors from all but the very first and very last scenes. Our stack is then, in effect, a six year interferogram. Our model includes deep slip on the regional fault network, shallow creep on the nearby Calaveras fault and distributed creep on 3 $\times$ 5 km patches on the San Juan Bautista segment. We enforce a positivity constraint on all faults and Laplacian smoothing on the distributed segments. We also enforce a slip rate of 35 mm/yr on the creeping section to compensate for poor data coverage south of our study area. We find that below the shallow-most row of fault segments (3 km depth), the creep rate falls off considerably, causing a low slip zone at mid-seismogenic depths in the northern part of the San Juan Bautista segment that may represent the source region for some of the 19th century earthquakes. We consider further the earthquake potential of the low-slip zone and estimate a slip budget for the San Juan Bautista segment.
G32A-07 11:50h
Coseismic Slip Distribution of the 2002 Mw7.9 Denali Fault Earthquake
We have estimated coseismic displacements for the 2002 Mw7.9 Denali Fault Earthquake at 232 GPS sites in Alaska and Canada, 180 of them within one rupture length of the earthquake. The displacement field shows a typical right-lateral deformation pattern for an E-W trending fault that cuts the surface. In general sites SW and NE of the fault move away from the fault and subside, but sites NW and SE of the fault move toward the fault and rise. Displacements along a N-S profile, crossing the fault along the Trans-Alaska Pipeline, indicate right-lateral slip on a near vertical fault with a significant component of vertical motion, north side up. These observations are in agreement with typical geological surface offset measurements along the Denali Fault (Haeussler et al., in press 2004). We estimated slip on a 3D fault model using a bounded variable least squares inversion, allowing only right-lateral strike slip and north-side-up dip slip. Allowing for oblique slip along the Denali and Totschunda faults improves the model fit to the GPS data by about 30%. We see mostly right lateral strike-slip motion on the Denali and Totschunda fault but in few areas we see a significant component of dip slip. The slip model shows increasing slip from west to east along the Denali Fault, with a few higher-slip patches. We infer maximum slip (9-12 m) about 40 km west of the Denali-Totschunda junction. This coincides with where geological observations estimated maximum slip of 8.8 m. Slip of 1-3 m was estimated along the Totschunda fault with majority of the slip being at shallower than 9 km depth. We have limited resolution on the Susitna Glacier fault but the estimated slip along the fault is consistent with a Mw7.2 sub-thrust-event. Total moment release in the Denali earthquake is Mw7.90, assuming a rigidity of 30 GPa. The estimated slip distribution along the surface is in very good agreement with geological surface offset measurements not measured on glaciers, but we find that surface offsets measured on glaciers are biased to lower values. We use the geological surface data, ignoring data from glaciers, to further constrain slip at the surface. This additional data set gives us important information on slip on shallower portions of the fault where we lack near field GPS data. The resulting slip model gives a similar but more refined slip distribution, with Mw7.89.
G32A-08 12:05h
The Stress/Strain/Fault Slip Cycle in Guerrero, Southern Mexico
GPS data from the Cocos-North American plate boundary in Guerrero, southern Mexico indicate four large transient slip events or ``silent earthquakes'' since 1992. We have developed an inverse model of the data that characterizes both the slip during events and the subduction megathrust coupling during inter-event times. Among the most robust features of this model are timing of the events and equivalent moment magnitudes of slip. Unlike similar (but smaller) silent earthquakes observed in Cascadia and Japan, interevent times vary significantly, ranging from 0.68 +0.08/-0.10 years to 4.01 +0.11/-0.13 years. Event magnitudes are 7.18 +0.19/-0.08 in 1998, 7.48 +0.01/-0.01 in 2002 and 7.22 +0.02/-0.15 in 2003. Within uncertainties, the pattern of larger events following longer periods of quiescence is similar to expectations for the strain accumulation and release in radiative earthquakes, with the primary difference that frictional coupling and subsequent slip occurs in the ``stable'' slip regime downdip of the seismogenic megathrust. Estimates of the steady-state frictional coupling on the fault, when compared with location and magnitude of the slip events, suggests that these events do not relieve seismogenic strain. If this pattern of behavior is consistent throughout the seismic cycle, the Guerrero segment of the Cocos/North America plate boundary currently has sufficient strain accumulated to generate a magnitude $\sim$7.9 earthquake. In this presentation, we will examine the cycle of stress/strain accumulation and release via frictional coupling and stable slip. We also will present InSAR deformation snapshots of the 1995 aseismic slip event, a slip event apparently initiated by the 1995 $M_w$=7.3 Copala earthquake.
http://anquetil.colorado.edu/~arlowry/Guerrero/models.html