Multi-Temporal Polarimetric SAR and Optical Data Fusion for Land Cover Mapping in Southwest China

Abstract

Zhazuo area located in Guizhou Province of southwest China was selected as the test site of this study. Based on six polarimetric RADARSAT-2 images, this paper presents the analysis on polarimetric and temporal variation backscattering behavior of typical targets such as residential areas, farmlands, forests and water bodies within the test site. The multi-temporal polarimetric SAR data fusion, as well as SAR and optical data fusion was conducted to improve the land cover classification. The analysis shows that there are significant differences of polarimetric backscattering and temporal variation behavior among different land cover types. The fusion of the multi-temporal polarimetric SAR data highlights the significant reduction of speckle noises, better land cover discrimination and improved visual appearance of SAR image. The fusion of SAR images with QuickBird data efficiently improves the land cover classification. This study demonstrates that high-precision land cover thematic mapping with SAR data in combination with optical remote sensing image in cloudy and rainy areas is feasible.