Spatiotemporal analysis of land use/cover change patterns in the new coastal district of Tianjin, China

Abstract

Land use/cover change is one of the most sensitive factors that show the interactions between human activities and the ecological environment. Since 1992, the government urged the development of the region of Bohai Gulf, the new coastal region of Tianjin has experienced great changes in land use/cover. It's urgent to detect the land use/cover change pattern to provide more explicit information on the further development of the new coastal district, which often requires to recover the history of land cover change and relates the spatio-temporal pattern of such change to other environmental and human factors, rather than merely relying on the change of areas or indices. This study makes Spatiotemporal analysis of trajectory land use/cover change patterns in the New Coastal District of Tianjin, after classifying the study area by Feature Extraction(FE) which first segments the whole area into objects and then classifies them based on the spectrum of the objects instead of pixels with Support Vector Machines (SVM) classifier. Multi-temporal and multi-source images of about one and a half decades were acquired for change detection, including Landsat TM, ETM+ and HJ1A images, and some land use maps were also used to supply training samples. First, all the images were geometrically corrected and registered on the same WGS84 coordinates. A unified land use classification plan was set up for all the images to classify land use type. The classification results were then utilized in the trajectory analysis of land use/cover change through the given three time nodes. Trajectories at every pixel were acquired to trace the history of land use/cover change for every location in the study area. Landscape metrics of change trajectories are also analyzed to detect the categorical change of different classes.