Investigation of Change Detection using Sentinel-1 Data for Identifying Illegal Occupations Upon Government Land in Hong Kong

Lands Department (LandsD) of Hong Kong SAR Government is now using an Object-based Change Detection (OBCD) with aerial photos to detect land cover changes for monitoring of land use and land administrative purposes. However, limitations are found in the current practice which made regular change detection over a large area not possible. To overcome these limitations, a feasibility study for using Synthetic Aperture Radar (SAR) images to conduct change detection in Hong Kong is conducted. A workflow which is combining Amplitude Change Detection (ACD) and Coherence Change-Detection (CCD) by using Sentinel-1 data is developed and its overall accuracy and Kappa Statistics for detecting changes are 90.98% and 0.6930 respectively. Vegetation is found as the major limitation of this workflow. Comparison between OBCD and the SAR workflow, and suggestions for enhancing the efficiency of identifying illegal occupations in Hong Kong are also given in this thesis