Evaluating Seasonal Deformation in the Vicinity of Active Fault Structures in Central California Using GPS Data

Central California is a tectonically active region in the Western United States, which encompasses segments of both the San Andreas and Calaveras Faults and centers around the town of Parkfield, California. Recently, statistical studies of microseismicity suggest that earthquake rates in this region can vary seasonally. Also, studies using data from modern GPS networks have revealed that crustal deformation can be influenced by seasonal and nontectonic factors, such as hydrological, temperature, and atmospheric loads. Here we analyze eight-years (2008 – 2016) of GPS data and build on this idea by developing a robust seasonal model of dilatational and shear strain in Central California. Using an inversion, we model each GPS time series in our study region to derive seasonal horizontal displacements for each month of the year. These positions are detrended using robust MIDAS velocities, destepped using a Heavyside function, and demeaned to center the time series around zero. The stations we use are carefully chosen using a selection method which allows us to exclude stations located on unstable, heavily subsiding ground and include stations on sturdy bedrock. In building our seasonal strain model, we first filter these monthly seasonal horizontal displacements using a medianspatial filter technique called GPS Imaging to remove outliers and enhance the signal common to multiple stations. We then grid these seasonal horizontal filtered displacements and use them to model our dilatational and shear strain field for each month of the year. We setup our model such that a large portion of the strain in the region is accommodated on or near the San Andreas and Calaveras Faults. We test this setup using two sets of synthetic data and explore how varying the a priori faulting constraints of the on and off-fault standard deviations in the strain tensor affects the output of the model. We additionally extract strain time series for key regions along/near the San Andreas and Calaveras Faults. We find that the most prevalent seasonal strain signal exists in the main creeping section along the San Andreas Fault in Central California. This region, which runs from Parkfield to Bitterwater Valley, shows peaks in contraction (negative dilatation) during the wet period (February/March) and peaks in extension (positive dilatation) during the dry period (August/September). The north transitional creeping section along the San Andreas Fault and the Calaveras Fault displays general similarities with the main creeping section trend. In sharp contrast, seasonality is virtually undetected in the locked section of the San Andreas Fault south of the town of Cholame. Additionally, the southern transitional creeping section shows two distinct patterns. For the most part this region, between Parkfield and Cholame, shows peaks in contraction during the wet period (February/March) and peaks in extension during the dry period (August/September), similar to the main creeping section. However, the segment of the southern transitional creeping section surrounding the town of Cholame opposes this trend with peaks in extension during the wet period and peaks in contraction during the dry period. We postulate several causes for this seasonal signal, which we plan to explore further in future work