Please use this identifier to cite or link to this item:
Title: Improved earthquake slip distribution inversion with geodetic constraints
Authors: Wang, Chisheng
Degree: Ph.D.
Issue Date: 2014
Abstract: Earthquake is a natural disaster that can lead to a significant loss of lives and properties. Earthquake studies have been relying on the inversion of ground observations, including geodetic and seismological data, due to a lack of underground measurement. The development of modern geodetic technologies such as Interferometric Synthetic Aperture Radar (InSAR) and Global Positioning System (GPS) have enabled ground displacement to be measured with unprecedented spatial and temporal coverage. The earthquake slip inversion with geodetic constraints has become a routine method to facilitate the understanding of the earthquake mechanism. The existing studies on earthquake source inversion offer a general approach to obtaining earthquake slip characteristics from geodetic observations, but the methods employed are by no means sufficient or optimal. For example, the utilization of the low-coherence InSAR interferogram is still an unsolved issue. Also, there are hardly any evaluation methods to quantitatively describe the performances of different InSAR downsampling criteria. In addition, in most studies, different fault patches are regularized equally without considering their diversity. One of the objectives of this thesis is to refine the existing methods. First, we propose a multiple-starting-point method to properly utilize the low-coherence InSAR interferogram. This can better deal with the low-coherence interferogram in terms of involving all isolated regions in the inversion. Studies used simulated data and Izimit earthquake data both show that this method can derive more detailed slip distribution results than the previous methods. Second, we can use the matrix perturbation theory to quantitatively describe the influence of InSAR data downsampling on earthquake inversion. A formula is devised to assess the extent of perturbation on the inversion results brought by downsampling. Based on this formula, we propose an improved equation-based downsampling algorithm which generates slip results closer to full data inversion. Our simulation study shows that this new algorithm outperforms previous sampling algorithms in maintaining earthquake slip information. Third, we analyze the performance of variable regularization in fault patches, and find that optimal regularization should be related to the roughness of the fault slips. We propose a two-step regularization method, which derives preliminary fault slips, and adjusts the regularization matrix according to the derived slips to obtain more refined results. Our experiment using simulated data from four different slip patterns shows that the two-step method can obtain results with smaller Mean Square Error (MSE).
This thesis also aims to study the sequence of the 2011 Van-Ercis and Van-Edremit earthquakes. The Van earthquakes occurred on unknown faults, and some research findings on them conflict with each other. We use a joint geodetic dataset, including GPS, InSAR, Multi-aperture InSAR (MAI) technique and SAR offset-tracking measurements, to constrain the slip distributions. The proposed multiple-starting-point unwrapping, equation-based InSAR downsampling and two-step regularization are applied to the earthquake slip inversion process. The results suggest that two nearly W-E striking segmental faults broke during the Van-Ercis event. Two main slip concentrations are found buried 5 km to 18 km underground. The Van-Edremit earthquake is shown to have been sourced from a dextral, north-dipping, and nearly E-W striking fault, and have had the largest slip of about 0.5 m at the depth of 6.4 km. The stress change analysis shows that the Van-Ercis earthquake imposed an up to 4 bars stress load on the causative fault of the Van-Edremit earthquake.
Subjects: Earthquakes.
Faults (Geology)
Hong Kong Polytechnic University -- Dissertations
Pages: xii, 130 leaves : color illustrations ; 30 cm
Appears in Collections:Thesis

Show full item record

Page views

Last Week
Last month
Citations as of May 19, 2024

Google ScholarTM


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.