UC San Diego

In this dissertation, I use Interferometric Synthetic Aperture Radar (InSAR) and Global Navigation Satellite System (GNSS) to study time-dependent crustal deformation due to several recent large earthquakes (M > 7) at the margin of Tibetan Plateau and in the eastern California shear zone (ECSZ), in order to have a better understanding of earthquake triggering process, lithospheric rheological and frictional properties during the earthquake cycle. Chapter 1 is an introduction to the tectonic background and the data I used in each following chapter. Chapter 2 studies surface displacements due to the 2019 Ridgecrest earthquake sequences, and investigates stress transfer and possible triggering relationships between pre-mainshock seismicity and the M7.1 mainshock. Because historical studies reveal different lithospheric rheologies across different margins of Tibetan Plateau, chapter 3 focuses on the study of 5-year postseismic deformation following the 2015 M7.2 Sarez (Pamir) earthquake to place the constraint on the viscosity of the lower crust beneath the west margin of Tibet Plateau. Chapters 4 and 5, by contrast, aim at constraining the viscosity of the lower crust beneath the north-east margin of Tibet Plateau. Consistent with previous postseismic studies of Tibetan earthquakes, we did not find any evidence of a low viscosity channel (10^{16} ∼ 10^{17} Pa s) beneath Tibetan Plateau margins. Moreover, studies of M > 7 strike-slip earthquakes that occurred in Tibet and California all suggest ∼ 30% of coseismic shallow slip deficit compared to its peak slip occurred in the depth interval of 3 − 4 km. The deficit is insufficiently accommodated by both interseismic and postseismic slip, which indicates off-fault yielding over multiple earthquake cycles. Chapter 6 proposes an inversion optimization method that aims to use the least number of parameters to fit geodetic observations almost equivalent well. We designed an 1D inverse problem that used synthetic surface data to invert slip distribution beneath the surface. The method we proposed reduces more than 2/3 of unnecessary number of parameters but achieves a good fit required by a certain uncertainty threshold.