Time-varying Displacement Field Mapping And Earthquake Three Elements Inversion Using CORS Network Measurements

Global Positioning System (GPS) seismology, a new subject developed at the end of the 20th century, has the unprecedented advantage in researching crustal deformation caused by strong shallow earthquake. Firstly, the crustal deformation spaning from seconds to years can be mapped by this technology. Secondly, time-varying displacement fields can be directly obtained, without the clipped and drift phenomena happened as the traditional instrument monitoring the strong earthquake. In recent years, Continuously Operating Reference System (CORS) observation networks consisted of many fixed GPS sites are developed widely and fast in terms of crustal deformation. While CORS still have broad development space on comprehensive and systematical research in seismology, and its potentiality remain to be found.In this research, we take 2008 Ms8.0 Wenchuan earthquake and 2013 Ms7.0 Lushan earthquake occurred in Sichuan area as my main research objects. My research work can be divided into three main parts as follows:Exploring the plate movement characteristics of Sichuan Basin before and after Wenchuan earthquake; Extracting seismic signal from high rate GPS coordinate time series, and investigating its features; Quasi-real time determination of earthquake three elements (the earthquake trigger time, hypocenter location and magnitude).The main contributions of this dissertation are summarized as follows:1. The movement characteristics of CORS stations in Sichuan basin before and after Wenchuan earthquake are analyzed. GAMIT/GLOBK software is utilized to obtain the coordinate time series of CORS stations. By establishing the time series models, velocity fields and vertical amplitudes from pre-seismic and post-seismic position time series are derived and compared. Preliminary results suggest that Wenchuan Earthquake potentially changed the movement characteristics of Sichuan Basin. These changes are embodied in the Common Mode Error (CME), the velocity field and the annual amplitude.The first principal component of CME is reduced obviously by 20%-40% after Wenchuan earthquake due to the effect of the post-seismic deformation; The velocity field changes obviously, especially at the PIXI, CHDU and MYAN sites, while YAAN and QLAI sites seem to be locked; The average annual amplitude of Sichuan Basin is the largest at the year before Wenchuan earthquake, and it becomes smaller graduallyafter that year.2. Taken the Ms7.0 Lushan earthquake occurred on April 20,2013 as example, the time-frequency analysis of high rate GPS coordinate time series indicates that the frequency band of earthquake signals is ranged from 0.02 Hz to 0.2Hz, and the PSDs (Power Spectral Density) are obviously larger due to earthquake. With combining 1OHz high-rate GPS displacement and 200Hz strong motion accelerometer data using a Kalman filter, one can obtain 200Hz displacement waveforms. It is obviously more advantageous than their individual observations and minimizing their weaknesses.3. The coordinate time series near source are analyzed by S transformation and three-dimensional searching method are presented in this work. The earthquake’s epicenter and trigger time are fast determined in 60 seconds. Moreover, the surface wave magnitudeis are estimated using the regression method, with horizontal peak amplitudes from the CORS sites in nearly 300 seconds. The moment magnitude is derived by fastCMT algorithm within several hours on condition that static displacements of sufficient CORS stations are evaluated, which is almost agreement with the result of Unite States Geological Survey (USGS).4. The impact of common mode errors on the coordinate time series of CORS networks can not be ignored when differential positioning strategy is used. Principle common analysis (PCA) filter can effectively remove it. The root mean squares (RMS) values of both daily coordinate time series before Wenchuan earthquake and high-rate coordinate time series of CORS stations are improved 30%-60% by employing this method.5. The optimized noise model building-up could improve the reliability of parameters estimated in the function model of time series. The method of maximum likelihood estimation (MLE) is utilized to choose the optimal color noise model of about seven years (2006-2012) continuous CORS time series. We found the optimal noise model in Sichuan area is flicker noise (FN)+ white noise (WN) and Power Law (PL)+ WN in total account for 78%. The spectral index estimated from high-rates CORS station position time series is from 1.6 to 2.0. It states clearly that their optimal noise model is random walk (RW)+ WN model.