SCEC Award Number 22134 View PDF
Proposal Category Individual Proposal (Integration and Theory)
Proposal Title 2022 SCEC Proposal: Joint Inversion of GNSS and InSAR Data for Continuous 3-D Velocity and Strain Rate Fields in Southern California
Investigator(s)
Name Organization
William Holt Stony Brook University
Other Participants Mradula Vashishtha, Jeonghyeop Kim
SCEC Priorities 1a, 1b, 1e SCEC Groups Geodesy, CXM, SDOT
Report Due Date 03/15/2023 Date Report Submitted 12/07/2023
Project Abstract
 We develop a joint inversion algorithm for continuous surface 3-D velocities and associated horizontal strain rate fields. Using a thin-elastic-sheet model, we construct basis functions that represent a response to body-force equivalents embedded within grid cells. The goal of this inversion is to find the best-fit linear combination of these basis functions that predict both GNSS and InSAR measurements. We use the 10-fold cross-validation method and the trade-off curve to determine an “optimal” level of smoothing. The relative weighting for each data set is initially decided based on the number of data points and uncertainties in GNSS data, and it is iteratively updated until the weighted root mean squared error (RMSE) of GNSS data achieves a value ~2 mm/yr. We have used this joint inversion algorithm to provide an estimate of the interseismic strain rate field, rotation rates, vertical rates, and horizontal gradients of vertical rates for southern California. The results can be used to provide new estimates of fault locking depths and the solution and codes will be contributed to the Community Geodetic Model (CGM).
Intellectual Merit The Intellectual Merit of this work is to jointly invert InSAR and GNSS data for a secular velocity field and associated strain rates, using a physics-based approach on a spherical Earth. This is one of the major research priorities for the CGM. Through this work, we can contribute a new joint inversion algorithm and its product to the CGM. Testing this algorithm to obtain a robust secular field estimate will also set the stage for providing a tool that can be used to quantify time-dependent fields.
Broader Impacts The Broader Impact of this work involves a preparation for the current research strategies of the CGM, which is to “develop vector time series of crustal deformation at ~1km spatial resolution and better than seasonal temporal resolution,” using both GNSS and InSAR measurements. The proposed work will allow the refinement necessary for modeling these time-dependent strain changes inferred from InSAR and GNSS time series. Our joint inversion algorithm is a new type compared to other previous published work in terms of our physics-based response functions. We originally developed this algorithm for time-dependent elastic signals. The physics-based basis function responses, however, may also be viewed as fitting functions for nonelastic signals. We observed this flexibility when the algorithm recovered 4 LOS displacement measurements associated with groundwater recharging processes (Murray et al., 2021). We plan to keep testing this algorithm as a fitting technique for nonelastic signals such as plastic deformations and poroelastic processes. Our approach is thus an important tool and set of products to provide to the CGM community. A final important broader impact is that this work is that it supported a female graduate student in research involving joint analysis of InSAR and GNSS observations.
Exemplary Figure Figure 3. Results from joint inversion. (a)-(d) Input InsSAR data obtained from Xu et al. 2021 (e)-(h) Prediction of InSAR data (i) Input GNSS velocities from Blewitt et al. 2016 and NGL’s MIDAS in white arrows and predicted velocities in red arrows along with contour map of predicted vertical velocity solution (i)-(k) Estimated time averaged crustal velocities (l) Horizontal strain rate field. Background is the 2nd invariant and arrows indicate principal axes of strain rate (gray = compression, bold = extension). Figure is from Vashishtha et al. (2023a).