High Resolution Interseismic Velocity Data Along the San Andreas Fault from GPS and InSAR

Xiaopeng Tong, David Sandwell and Bridget Smith-Konter




High-Resolution Interseismic Velocity Map along the San Andreas Fault from ALOS/PALSAR and GPS
Figure 1. Interseismic deformation of the SAF derived from integrating the GPS observations with ALOS radar interferograms (2006.5-2010). The radar flying direction and look direction are marked. Positive velocities (reds) show the ground moving away from the satellite. The shading highlights the gradient in the velocity field. The areas with low coherence and large standard deviation (> 6 mm/yr) are masked. The black lines shows the geological fault traces. Satellite images were collected by the L-band synthetic aperture radar aboard the ALOS spacecraft that is operated by the Japanese Space Agency - JAXA.
Click the image and Zoom In to view the details of this map. A full resolution version of this LOS velocity map and its relationship to faults and cultural features can be downloaded as a KML-file for Google Earth from here: ALOS_ASC_masked.kmz


Crustal velocity model in line-of-sight (LOS) velocity based on regional GPS velocity field
Figure 2. Crustal velocity model in line-of-sight (LOS) velocity based on regional the GPS velocity field [Smith-Konter and Sandwell, 2009] in oblique Mercator projection. The colors represent the LOS velocity field along 13 ALOS ascending tracks. The radar flying direction and look direction are marked in Figure 1. Positive velocities (reds) show the ground moving relatively away from the satellite. The small triangles are the GPS stations used to constrain the velocity model. The black lines shows the geological fault traces.


High-pass filtered residual velocity (2006.5-2010) along ALOS ascending tracks
Figure 3. High-pass filtered residual velocity (2006.5-2010) along ALOS ascending tracks. This residual velocity reveals the discrepancy between the InSAR observations and GPS model prediction at short wavelength. For example, we found that the residual are significant along the creeping sections, the Garlock fault, and the LA basin. A fine-tuned interseismic velocity model based on both InSAR and GPS observations should have smaller high-pass filtered residual velocity. Note that the residual could also be caused by non-tectonic effects, such as ground water.


Standard deviation of the average LOS velocity (2006.5-2010) along ALOS ascending tracks
Figure 4.Standard deviation of the average LOS velocity (2006.5-2010) along ALOS ascending tracks. Larger uncertainties are found north of the San Francisco Bay area in northern California, near the San Bernadino Mountain. The uncertainties could be due to unwrapping errors, atmospheric noise or deviations from steady-state ground motion. The standard deviation provides a measure of uncertainty of the high-resolution LOS velocity data and can be used in modeling the interseismic deformation.


InSAR/GPS integration
Figure 5. Illustration of the InSAR/GPS integration method. The thin black lines are the average of the 6 pairs of coherence spectra from GPS velocity models. The coherence spectrum for 6 pairs of 4 GPS velocity models are compared here: [Meade and Hager, 2005]; [McCaffrey, 2005]; [Zeng and Shen, 2010]; [Smith_Konter and Sandwell, 2009]. Current interseismic velocity models based on GPS measurements alone cannot resolve features with short wavelengths (<15-40 km). L-band InSAR data is contaminated by errors at longer wavelengths from ionosphere, orbit (plane), and the atmosphere. To remedy these inadequacies, we recovered the interseismic deformation along the entire San Andreas fault at a spatial resolution of 200 meters by combining GPS and InSAR observations using a Sum/Remove/Filter/Restore (SURF) approach. The integration uses a dislocation-based velocity model to interpolate the Line-Of-Sight (LOS) velocity at the full resolution of the InSAR data in radar coordinates. The residual between the model and InSAR LOS velocity were stacked and high-pass filtered, then added back to the model.


Fault creep rate from InSAR and ground-truth comparison along SAF
Figure 6. Creep rate comparison with an independent data set compiled by The Uniform California Earthquake Rupture Forecast, Version 2 (UCERF2). The red circles are the creep rate along SAF from InSAR in the period from 2006.5 to 2010 (this study). The error bars show the standard deviation of the measurements. The triangles and other symbols are independent creep measurements compiled by UCERF2. AA means alignment array; CM means creep meters; Cult means cultural offset. a) Creep rate along the entire SAF from north to south. The inset on the upper right corner shows the linear regression method to determine the surface creep rate across fault. b) A zoomed-in view at the creeping section in central California.