Land-use/land-cover change detection by using the extended change-vector analysis

Abstract

Detecting land-use/land-cover (LULC) changes in rural-urban fringe areas (RUFAs) timely and accurately using satellite imagery is essential for land-use planning and management in China. Although traditional spectral-based change-vector analysis (CVA) can offer an effective method of LULC-change detection in many cases, it encounters difficulties in RUFAs because of deficiencies in the spectral information of satellite images. To detect LULC changes in RUFAs effectively, this paper proposes an extended CVA approach that incorporates textural change information into the traditional spectral-based CVA. The extended CVA was applied to the Haidian District, Beijing, China. The results demonstrated the improvement of the extended CVA compared to the traditional spectral-based CVA: overall accuracy increased from 90.67% to 95.33%, and the kappa coefficient increased from 0.81 to 0.91. The advantage of the extended CVA lies in its integration of both spectral and textural change information to detect LULC changes, allowing for effective discrimination of LULC changes that are spectrally similar but texturally different in RUFAs. The extended CVA has great potential to be widely used for LULC-change detection in RUFAs, which are often heterogeneous and fragmental in nature, with rich textural information.