Error Factors Analysis of Detecting Terrain Deformation with PS-InSAR Technology in Mountain Area of Taiwan

  • 陳 郁琪

Student thesis: Doctoral Thesis


Recently many geodetic surveys have been widely used in Taiwan to monitor land surface deformation such as Interferometric synthetic aperture radar (InSAR) preliminary survey and GPS measurement InSAR technology is widely used including the original development of D-InSAR the later extended PS-InSAR and the proposed TCP-InSAR in 2011 SAR is now widely used in monitoring interpretation and classifying landslide types in mountainous areas This study used 46 ascending Sentinel-1A images and 29 descending images with SNAP/StaMPS to get LOS velocity and displacement The GPS data is from opendata of GPS LAB Taking the southern mountainous area for an example analysis the differences of site velocities between InSAR and GPS under different terrain factors The results show that after higher coherence threshold value filtering the smaller the RMS of GPS and InSAR difference In the ascending case the original data (unfilter) RMS is 14 39 mm/yr down to 13 62 mm/yr (0 9 coherence); in the descending case the original RMS is 5 45 mm/yr down to 5 02 mm/yr In the height factor analysis it is found that in the data of flat terrain where the elevation is less than 100 meters the ascending data is more suitable On the contrary if elevation is above 100 meters the descending data is more suitable In slope factor analysis the results show that if the angle is less than 5 degrees the ascending data is more suitable On the other hand it is more suitable to use the descending data with slope of 5~10 degrees Besides when the slope is greater than 10 degrees the degree of dispersion is large in both the ascending and descending data In aspect analysis overall the reliability of the slope facing ray direction is greater than that of the back slope The ascending data between aspect 170 to 260 degrees and the descending data between 10 to 100 degrees the RMS and standard deviation are lowest and the correlation coefficient is highest
Date of Award2019
Original languageEnglish
SupervisorTing-To Yu (Supervisor)

Cite this