Spatial clustering considering spatio-temporal correlation

Geographic time-series data have the characteristics of time correlation and spatial correlation. Similarity/Dissimilarity measurement is a key problem to measure these correlations. Based on spatio-temporal correlation analysis of geographic time-series data, the paper proposes a kind of spatial clustering method which considers spatio-temporal correlation. The paper first puts forward a method of dissimilarity measure which includes both time-series dissimilarity and spatial dissimilarity, and then it proposes the spatial clustering method by incorporating the proposed dissimilarity into fuzzy C-means clustering. At last this method is applied to GDP clustering of 31 mainland provinces in China. The theoretic analysis and experiments validate the proposed spatial clustering method.