Extracting Horizontal Ground Displacements And Time-series Information From Optical Images

Surface deformation monitoring provides an intuitional and reliable way to observe and study geophysical phenomena. With development of optical satellite technology, a new method emerges for deformation detection, which is alternative to SAR interferometry. The observation capability of optical satellites is greatly enhanced, which broadens views of people for recognizing phenomena on the Earth. This thesis focuses on investigation of this new methodology for measuring displacement, and addressing how to extract time-series information of horizontal displacements.The Landsat 8 satellite was successfully launched on 11 February 2013. It allows to measure large deformation in addition to traditional applications of optimal satellite observations. It compensates the limitation of single-direction deformation detecting of InSAR data, and also offers new information about temporal evolution of horizontal ground displacements. This thesis studies the surface deformation of the West Kunlun Mountain glaciers in the northwestern Tibetan plateau as an application case. The purpose is to investigate how to extract time-series information using the new Landsat-8 data based on sub-pixel cross-correlation of optical images and the time-series analysis technique.Firstly, this thesis presents the principles of extracting surface deformation and its time-series information from optical images. This thesis explores the methodology of extracting horizontal ground displacements and its time-series information from optical images and distinguishes it from InSAR technique. Then in the case study of active glacier motion of the West Kunlun Mountains, this work obtains periodically acquired Landsat-8 images of the study area, and coregisters them using the sub-pixel cross-correlation technique at an accuracy of 0.01 pixel, implying a horizontal displacements detecting precision of the optical image 0.15m. By cross-correlation analysis of the Landsat-8 images acquired from July 2013 to August 2014 and time-series inversion, this thesis constructs the displacement and velocity fields of two glaciers in the West Kunlun Mountains. The results are summarized below.(1) It is effective to extract horizontal ground displacements and improve the measuring accuracy of spatial-temporal deformation fields based on Landsat-8 images by the using image registration method on amplitude information, the sub-pixel cross-correlation technique and SVD time-series inversion.(2) This study obtained the time-series information of horizontal displacements from July 2013 to August 2014 of the West Kunlun Mountain glaciers by using Landsat-8 images for the first time. It demonstates that the glaciers in the study area were slipping fast and uniformly (no obvious acceleration and deceleration) at a rate of 6m/month-18m/month.(3) The results of this thesis have reference significance on glacier dynamics and climatic oscillation. This study also explores a method and its potential application in measuring large horizontal ground displacement and crustal deformation caused by earthquakes, landslides, and glaciers using Landsat-8 images.