Strain Rate
The 3-D velocity maps derived from the InSAR/GNSS combination have small spatial scale variations that are associated with hydrological processes and residual atmospheric error. These models must be smoothed to extract the underlying tectonic strain rate [e.g., Weiss et al.,2020]. This is accomplished by using a smoothing interpolator that has constraints from elasticity to couple the two horizontal velocity components [Haines and Holt, 1993] as implemented in the program gpsgridder in GMT [Sandwell & Wessel , 2016]. We first subsample both the GNSS and InSAR/GNSS horizontal velocity fields shown in Fig. 6 at a spacing of 2.5 km resulting in 40,016 2-D velocity estimates. The gpsgridder algorithm uses singular value decomposition of the elasticity Greens functions and we select only those associated with the largest 800 eigenvalues. We then compute the three horizontal strain rate components from the smoothed horizontal velocities. The comparison of the GNSS only and InSAR/GNSS for several components of strain rate (second invariant, maximum shear and dilatation) are shown in Fig 7.
The San Andreas fault is well delineated by the concentrated high-strain (> 200 nanostrain/yr). Compared to the prediction of the GNSS-only strain rate model, the integrated InSAR/GNSS model reveals larger strain concentration, especially at locations where there is known surface creep (e.g., creeping section of SAFS and Hayward fault). Maacama and Bartlett Springs both have much larger strain concentrations in the InSAR/GNSS model although this may be a consequence of the assumption that the InSAR deformation is parallel to the GNSS deformation. Dilatational strain predicted by the integrated InSAR/GNSS model shows extension in the Central Valley associated with the subsidence from groundwater extraction. The shear component is also larger in areas of known hydrological signals. There is a prominent shear strain anomaly in the InSAR/GNSS strain rate maps (Fig. 7e, f) that is to the east of and parallel to the creeping section. However, there is not enough GNSS data in the region to determine if this is a real strain-rate anomaly or artifact from the interpolation approach. Overall, the strain is more widely distributed in the integrated InSAR/GNSS model, with significant components being off-fault. The remaining questions are how much of these are from tectonic motion and how much from hydrological activities, and whether these off-fault components are steady over time or just transients. Answering these will be the key leading to accurate assessment of the strain/moment accumulation and the associated seismic hazards.