Probing the Rheology of Tibet using Postseismic Motion from Large Earthquakes

Figure 1. Interferograms for the three earthquakes considered in this study. Postseismic interferograms are shown for the 1997 Manyi earthquake and the 2001 Kokoxili earthquake. For the Manyi case, ERS-2 SAR data covering an 8-month time interval are used, and for the Kokoxili case, Envisat data covering a 1-year time interval are used. A coseismic interferogram is shown for the 2008 Yutian earthquake, with each fringe representing about 12~cm of line-of-sight displacement.

Project Summary For the last three decades, a rather vigorous debate has been going on about the most fundamental aspects of the style of continental deformation in the India-Eurasia collision zone. While one school of thought envisions Tibet to be a thickened, weak and fluid-like zone (e.g. Cook and Royden, 2008), others consider the tectonics in the region as that of essentially rigid microplates bounded by major lithospheric faults (e.g. Thatcher, 2007). Until this issue is resolved, it will be difficult to gain a clear idea of what forces drive the deformation of the upper crust in this region. Large earthquakes initiate rock mechanics experiments of lithospheric dimensions in which a sudden stress change leads to stress relaxation at depth. Surface measurements of the resulting deformation provide a basis for testing structural and rheological models of the lithosphere. Our main objective in this project is the elucidation of lithospheric rheological structure in northern Tibet, chiefly through InSAR investigation of postseismic motion following three recent major earthquakes. These events are the 1997 M 7.6 Manyi earthquake (Funning et al., 2007; Ryder et al., 2007), the 2001 M 7.9 Kokoxili earthquake (Lasserre et al., 2005) and the 2008 M 7.2 Yutian earthquake. By analysing the spatio-temporal characteristics of the postseismic signals, and testing the data against various candidate models, we hope to determine whether deformation in the mid to lower crust is localized or distributed, and constrain rheological parameters associated with our preferred model. Our broader objective is to inform the geophysical debate concerning the nature of the Tibetan Plateau: specifically, whether it behaves more like a viscous fluid or a series of rigid blocks.

InSAR Observations We use the satellite geodetic technique of Interferometric Synthetic Aperture Radar (InSAR) to observe postseismic deformation for several years following each of the three major earthquakes. For the Manyi case we use data from the European Space Agency's (ESA) ERS-2 satellite, and for the Kokoxili case we use radar scenes from ESA's Envisat. The Yutian earthquake occurred after the launch of Japan's L-band ALOS satellite, so we will be able to utilize postseismic data from this as well as the C-band Envisat. Postseismic interferograms for both the Manyi and Kokoxili earthquakes reveal several centimeters of line-of-sight transient displacement in both cases. The wavelength of the postseismic signal suggests that relaxation processes occur at a depth of 15-20 km, which is approximately the seismogenic depth in this area. A deformation time series for the Manyi event can be constructed from multiple interferograms, and yields a relaxation time of 0.7~years. It is not possible to make early postseismic interferograms for the Kokoxili case, and so making a time series is difficult, but GPS data from a network of 65 sites either side of the rupture will ultimately give temporal information about the postseismic transient. We are currently waiting for postseismic data for the Yutian earthquake; cosesismic interferograms have been processed, and show clear deformation fringes despite the ice cover in the area. Figure \ref{ryderfig1} shows interferograms for each of the three events.

Viscoelastic Models In order to interpret the observed surface deformation in terms of sub-surface rheological structure, we implement models of postseimic stress relaxation in a viscoelastic medium. The idea here is that stress changes near the fault induced during the earthquake are relaxed due to viscous flow beneath an elastic upper crust. The thickness and viscosity of a Maxwell viscoelastic layer are varied in the models, and we seek parameters which best fit the data. For both the Manyi and Kokoxili cases, the best-fit viscosity is between $5\times10^{18}$ and $1\times10^{19}$~Pa\,s; a single thick viscoelastic layer is preferred, suggesting that the entire mid to lower crust flows, with no stratification in viscous properties. A Maxwell viscoelastic medium is the simplest type of linear viscoelastic rheology, and it is reasonable to run initial models using such a rheology. However, the Manyi time series is not fit consistently throughout the entire 3.5 year observation period by a single Maxwell viscosity; rather, the effective viscosity increases over time. Another linear rheology which could produce this effect is a Burgers body, which consists of a Maxwell and a Kelvin element in series. We are currently running further models, varying the transient and steady-state viscosities as well as the ratio of Maxwell and long term shear moduli. Another candidate relaxation mechanism is localized afterslip on an extension of the coseismic rupture plane at depth. The data will also be tested against models simulating this mechanism. For strike-slip earthquakes such as Manyi and Kokoxili, surface deformation due to distributed viscous flow and localized afterslip can look very similar. The postseismic signature of the normal-faulting Yutian earthquake should be particularly useful for distinguishing between mechanisms, since the surface deformation from the two processes looks very different for dip-slip events.

Tools InSAR processing and analysis, GPS data analysis, forward and inverse elastic and viscoelastic modeling

Geographic Location Northern Tibetan Plateau

Group Members Involved Isabelle Ryder, Roland Bürgmann

Project Duration In Progress: start of 2008 through 2009