G52A-01 INVITED
A More Accurate Vertical Velocity Field for Coastal Oregon Reveals Variations in Extent of Locking on the Cascadia Subduction Zone
The pattern of uplift observed at the Earth's surface provides a strong constraint on the extent of locking at depth on subduction thrust faults. We have utilized the rich history of high quality leveling and sea level observations made in coastal Oregon and northernmost California over the last century to calculate more accurate, spatially- dense observations of uplift rate with realistic uncertainties. We have extended the water level time series of the six primary NOAA tide gauges from Crescent City, CA to Astoria, OR to include all observations 1925-2006, placed in a common reference frame at each gauge by local leveling. Analysis of the tidal leveling loops shows that tidal records are contaminated by instability of individual primary benchmarks, which have been used to define the local vertical datum. Some of the primary benchmarks changed elevation relative to nearby monuments up to 1.6 mm/yr at South Beach and Astoria. We corrected the tidal records and tied them to a stable reference monument at each gauge that has been releveled in regional lines. Sea level trends are refined with more precise inter-site differences, weighted by uncertainties that account for autocorrelation. The estimated standard errors are ~0.1 mm/yr for the six sites after the adjustment. We use a regional rate of sea level rise specific for the data interval of 2.3±0.2 mm/yr. This rate is consistent with the long record at Seattle, which lies outside the region of interseismic subduction uplift, and global geocentric reconstructions of regional sea level change. Regional releveling conducted by the NGS provides estimates of relative uplift rates along the north-south coastal route, and three lines extending east. At least three leveling epochs along most of these routes allow us to identify systematic errors in the data. We double the number of previously published high precision differences by making secondary ties between nearby monuments using tidal leveling, reset benchmarks, and 1941 elevation differences to span between 1930s and 1980s and minimize the impact of systematic errors in long 1941 lines. We attach the relative uplift rates from leveling to the rates at the tide gauges through a weighted adjustment that accounts for uncertainties in both the relative tidal and leveling uplift rates, as well as 0.1-0.3 mm/yr errors in tying the leveling to the tidal rates. Tidal and leveling uplift rates agree within error for all but one of the coastal segments, where previously unrecognized systematic leveling error affects the 1988 line. Down-dip profiles centered on the east-west leveling lines show that vertical velocities decrease from maxima near the coast to near zero in a smooth, concave-up pattern. However, along-strike, the pattern shifts east-west, with northern coastal Oregon uplifting up to 3.5 mm/yr faster than central Oregon, for the same distance from the trench. There is a ~1 mm/yr uplift rate increase near Cape Blanco, OR, with the southern area uplifting more slowly. We model the area with an elastic dislocation model of the subduction zone with smoothly varying locking depth along strike, which matches the observed data well. The vertical velocity field is a more direct probe of subduction locking than the horizontal field, which reflects both forearc motion and subduction strain accumulation. We will compare predictions of horizontal velocities from our model to GPS observations. Predicted uplift rate patterns for known active crustal structures and post-glacial isostatic adjustment do not appear consistent with the uplift rate data, suggesting that subduction zone locking is the only resolvable vertical signal.
G52A-02
A Microfossil-Based Approach to Constraining Megathrust-Induced Coseismic Land Displacement of the 1700 Event in the Pacific Northwest
Continuing subduction of the Juan de Fuca plate beneath the North America plate in central western North America constitutes a major seismic hazard but the history of great earthquakes in the region remains unclear. The tsunami accompanying the last great earthquake along the Cascadia subduction zone was widely recorded in Japanese records as an "orphan" tsunami (no source earthquake was felt in Japan) that would have been on the evening of January 26th AD 1700. Models of the inferred tsunami indicate an earthquake magnitude between M8.7 and M9.0, suggesting a plate-boundary rupture of over 1000km causing up to 20m of slip. Relative sea-level changes along much of Cascadia's coast are thought to be dominated by an earthquake cycle of rapid coseismic subsidence during plate-boundary rupture followed by gradual interseismic relative sea-level rise due to upper plate strain accumulation lasting hundreds of years. To learn more about the deformation cycle of the AD 1700 and earlier great earthquakes, we apply transfer functions to modern foraminiferal datasets from coastal Oregon to interpret the fossil foraminiferal record of sea-level change in cores. We collected seven modern salt-marsh transects (162 samples) extending from mudflat to upland. Nine of our tidal marsh cores sample sediment spanning the AD 1700. Furthermore, sediment in 4m-long vibracores collected at each of Nehalem River and Siuslaw River estuaries span the times of three to five earlier great earthquakes. The AD 1700 earthquake is marked in all cores by a distinct lithological and biostratigraphical changes. Preliminary foraminiferal data show a sudden subsidence of 0.44m ± 0.07m during the AD 1700 earthquake in the Nehalem River core and between 0.16m and 0.36m ± 0.08cm at Siuslaw River core. The microfossil-based transfer function approach produces high-precision geological reconstructions of relative sea-level of sufficient resolution to better estimate vertical ground displacements associated with the earthquake deformation cycle.
G52A-03 INVITED
Vertical Ground Motion Measurements with GPS
In this talk we discuss the measurement of vertical ground motion with the Global Positioning System (GPS). The measurement of heights with space based geodetic systems are intrinsically less accurate than measurements of horizontal positions because of the distribution of signals from only the upper hemisphere of the sky and because of the correlation between atmospheric propagation delays and the estimates of height. In addition, for GPS, the effects of phase changes through and around the antennas and radomes are strongly elevation angle dependent and effect the determinations of heights. Over relatively short distances, less than a few hundred kilometers, the atmospheric delays and geometry limit the precision of relative height measurements, averaged over 24-hours, to 2-3 mm. Over longer distances and for height measurements in a global reference frame effects for phase patterns on the GPS satellites and the effects of orbital modeling effect the accuracy of height measurements. These latter effects limit the height precision to 5-10 mm. Temporal averaging reduces the noise in height measurements but systematic effects from satellite antenna patterns and possibly long term orbital effects limit the accuracy of height rate measurements. We show GPS results from a variety of spatial and temporal scales and assess the accuracy characteristics of GPS determination of height.
G52A-04 INVITED
Postseismic Relaxation at the Central Nevada Seismic Belt Observed in Vertical GPS Time Series
Between 1872 and 1954 the Basin and Range province of the western United States experienced six major earthquakes that occurred in a quasi-linear belt known as the Central Nevada Seismic Belt (CNSB). These strike-slip to normal events account for most of the seismic moment release that has occurred in the Basin and Range in historic time. Several studies have noted the possible presence of a contemporary geodetic signal owing to postseismic relaxation from these earthquakes, implying that this signal has persisted for decades after the events. Observations that support the existence of this postseismic signal are: 1) GPS-derived horizontal strain rates that are relatively high compared to the surrounding regions, 2) 3-4 mm/yr vertical upward doming observed with Interferometric Synthetic Aperture Radar (InSAR) that is consistent with models of viscoelastic relaxation of the mantle, 3) possible rapid vertical postseismic motion observed using leveling measurements immediately following the 1954 events, 4) disagreement between geodetically and geologically inferred strain rates that is consistent with elevated contemporary transient strain. Since 2004 the University of Nevada, Reno has operated a semi-continuous GPS network with ~20 km spacing (the Mobile Array of GPS for Nevada Transtension: MAGNET) that complements the spatially less dense Basin and Range Geodetic Network (BARGEN) and Plate Boundary Observatory (PBO) continuously recording networks. MAGNET is ideally deployed to observe the ongoing transient deformation associated with the central Nevada earthquake cycle because it spans the CNSB faults where the predicted postseismic signal is greatest. We analyze the GPS data with the GIPSY/OASIS II software package in precise point positioning and apply regional filtering to remove common mode effects (uniform displacements and/or rotations that are present in all the time series). These common mode signals have a much larger impact on the vertical than on horizontal rates, especially for shorter time series, so accounting for them is essential for studying vertical motions with GPS. The filtered time series enhance our resolution of the relative vertical motion between sites. Our preliminary results based on 2.5-3.3 years of GPS data from MAGNET indeed detect a horizontal gradients in the vertical rates of 3 - 4 mm/yr that is consistent with earlier studies based on InSAR and campaign GPS. We find that the MAGNET sites directly east of Dixie Valley (in the Clan Alpine Range) have the greatest vertical uplift rate, suggesting that the 1954 Dixie Valley earthquake makes a larger contribution to the contemporary uplift than the other earthquakes associated with the CNSB. The largest amplitudes in the vertical signal appear to be limited to the MAGNET sites, highlighting the need for arrays denser than provided by the continuous networks (i.e. PBO and BARGEN) in this region. http://geodesy.unr.edu/networks
G52A-05 INVITED
Vertical Displacements During the Earthquake Cycle: Observations with InSAR.
InSAR is a satellite based radar technique capable of measuring small ground displacements at high spatial resolution and, due to the look angle of the satellites used, is especially sensitive to vertical displacements. By combining images taken from different directions, and by using independent measurements of horizontal displacements (e.g. GPS, image matching), it is possible to measure the vertical component directly. We investigate the associated advantages and disadvantages in relation to studies of tectonic deformation through the earthquake cycle using examples from the Denali Fault, Alaska. InSAR has been used to measure the slip rate on several strike-slip faults using a single look direction under the assumption that displacements are horizontal only. However, many strike-slip faults are associated with mountain ranges and in reality, motion is likely to be transpressive. In addition, vertical motions not related to fault slip may also be present, caused by loading-related deformation or local effects such as differential subsidence. The tradeoff between horizontal and vertical motions on observations from a single look direction means the convergent motion or unrelated vertical movements can have a serious impact on rate estimates. For example, the observed displacement on the Denali Fault, Alaska could be explained by either a horizontal slip rate of 12 +/- 5 mm/yr or a vertical slip rate of 1.9 +/- 1.5 mm/yr. Using Radarsat data collected following the 2002 M7.9 Denali Fault Earthquake, the postseismic response occurred is found to occur at depths greater than 40 km. Vertical measurements from InSAR are seen to show a clearer spatial pattern than vertical GPS, but cover a more limited region. The spatial pattern can be modelled using either afterslip or viscoelastic relaxation, or a combination of mechanisms.
G52A-06
Observations and modelling of long-term inflation of a giant magma body in Socorro, New Mexico
The Socorro magma body (SMB) in central New Mexico is one of the largest known active intrusions in the Earth's continental crust. Previous geodetic studies spanning various time intervals between 1911 and 1999 indicated an uplift occurring at an average rate of a few millimeters per year. The data can be reasonably well explained in terms of elastic inflation of a sill-like magma chamber, but the steady-state elastic inflation over the lifetime of the SMB is difficult to reconcile with thermodynamics of intrusion (Fialko and Simons, 2001). We present new InSAR data covering the period between 2000-2007 that reveal that the uplift persists at an essentially constant rate. We use 3-D finite element models to constrain the age of the emplacement, the inflation history, and the rheology of the lower crust. The magma chamber is modelled as a sill with a horizontal shape consistent with the observed outline of the seismic reflector at a depth of 19 km, and the observed uplift pattern. We explore a range of plausible crustal rheologies, and different emplacement times and histories of pressure within the chamber. Observables that are brought to bear on the model predictions are the historic uplift rate, and spatial pattern of surface deformation. Preliminary results indicate that the observed uplift history may be compatible with a viscoelastic crustal rheology provided that the effective viscosity is high (1019 Pa s or greater). The shape of the uplift is sensitive to viscosity, with lower viscosities requiring smaller rates of pressure increase and leading to broader patterns of uplift.
G52A-07
Measuring vertical deformation in the Seattle, WA urban corridor with satellite radar interferometry time series analysis
Satellite radar interferometry (InSAR) time series analysis (e.g., Lundgren et al., 2001) can reveal rich patterns of deformation in both time and space. As the technique is sensitive to mm-scale vertical deformation over large and spatially extensive regions, it provides a useful geodetic tool where satellite coverage and radar phase coherence permit. Here we apply InSAR time series techniques based on the Small BAseline Subset Algorithm (SBAS) (Berardino et al., 2002) using data from three satellites (ERS 1, ERS2, and RADARSAT) to the urban corridor between Tacoma, Seattle and Everett, WA, over the time period 1992 - 2007. The target of our work is to better characterize the nature of active faulting and deep-seated landsliding within the densely populated study area. Additionally, we seek to independently quantify how localized short-wavelength deformation is contaminating data collected from the ~ 12 GPS stations in the eastern Puget Sound region. Comparisons of InSAR time series inversions to data from 4 GPS stations temporally and spatially overlapping the available InSAR observations reveal that surface displacement computed from InSAR matches the GPS deformation within the range of error reported for vertical GPS data (~ 4mm). Contemporaneous surface velocity maps generated via linear regression to two independent time series inversions from overlapping ERS satellite tracks 428 and 156 show striking agreement in the pattern of surface velocity, and effectively resolve rates as low as 1 mm/yr. Based on the results of our velocity mapping, we provide new constraints on surface deformation in the Seattle metro region. First, between 1992 and 2007 we document subsidence (~ 1-3 mm/yr) over much of the region characterized by Holocene infilling of the Puget Sound by lahar and floodplain sedimentation. This deformation is consistent with subsidence due to sediment compaction and de-watering. Second, between 1992 and 2007 we document no slow landslide deformation on any of the numerous mapped slide complexes within Seattle. Regions of known active landsliding, such as along Perkins Lane in Seattle, exhibit radar phase de- correlation. These observations are therefore consistent with relatively infrequent and rapid landslide deformation within Seattle. Finally, we note a NW-SE striking, sharp linear gradient in deformation near Federal Way, WA. As this feature is located just north of the Tacoma Fault Zone, it may mark the location of a previously unmapped fault splay that is serving as a barrier to local groundwater flow.
G52A-08
Estimate and Modeling of Vertical Ground Displacements by Space and Terrestrial Techniques in the Northern Adriatic, Italy
The Northern Adriatic area, in Italy, including coastal regions, is affected by high rates of natural subsidence greatly enhanced, during last century, by anthropogenic components mainly due to ground fluid exploitation. To measure and model vertical crustal movements is of major importance for the presence of natural lowlands and coastal lowlands which are exposed to the risks of flooding events due to river overflow or to the occurrence of extreme storm events. Sea-level rise resulting from increasing global temperature is an additional threat to this region. A network of permanent GPS stations, extending from the southern Po River Plain to the area around Venice, is in place since several years. At most of the stations, in addition to GPS, other geodetic observation techniques such as gravity and InSAR allow for validation of the individual time series, for continuous monitoring in space and time of vertical movements by combining GPS and InSAR, and for a better understanding of the geodynamical processes in the area. Three tide gauge stations with centennial time series, co-located with the GPS systems, are available. The vertical land movements can be removed from the tide gauge records in order to derive absolute sea-level time series. A comparison with satellite altimetry data is performed at the three coastal stations, although this is not an easy task because the satellite data are limited due to the narrow shape of the Adriatic basin. The GPS network data have been re-analyzed by using ITRF2005 and the vertical rates have been estimated. These results are compared to those provided by absolute and relative gravimetry and InSAR data.