AGGREGATION IN LAND-COVER DATA GENERALIZATION CONSIDERING SPATIAL STRUCTURE CHARACTERISTICS

Abstract. Aggregation is an important operation for the generalization of land-cover data. However, current research often entails aggregation on a global perspective, which is not conducive to capturing the spatial characteristics of geographic objects with significant spatial structures, i.e., structured geographic objects. Hence this paper proposes an area aggregation method that can maintain the boundary characteristics of the structured geographic objects. First, we identify the structured geographic objects based on the description parameters of the spatial structure. Second, a Miter-type buffer transformation is introduced to extract the boundary of each structured geographic object, and area elements inside the boundary are processed with corresponding aggregation operations. Finally, the boundary of the structured geographic objects and the aggregation result of the area elements are inserted back into the aggregated result of the original land-cover data using the NOT operation. The proposed approach is experimentally validated using geographical condition census data for a city in southern China. The experimental result indicates that the proposed approach not only reasonably identify the typical characteristics of structured geographic objects but also effectively maintains the boundary characteristics of these objects.