Exploring migration patterns through uncovering migratory birds' habitat information is very important in biology, which has scientific significance in animal habitat conservation and avian influenza control. In this paper, we convert the traditional biology problem into a computational study and use data mining techniques to analyze the spatial and temporal distribution of bird-watching data in China. First, we present an improved hierarchical clustering algorithm (IHDBSCAN) to identify the habitats/stopovers of migrant birds. Then, we use a kernel smoothing method to fit the temporal distribution of bird observation in each spatial cluster. A hierarchical cluster tree is generated where the leaf nodes indicate different bird habitats/stopovers. Finally, the results is visualized on the map of China. Experimental results show that the proposed algorithm can effectively find the spatial and temporal distribution of Anseriformes' habitats.