A geographic ontology fusion method based on granular theory is presented. First, Formal concept analysis (FCA) is used to formalize different geographic domain ontologies. Then granular computing is introduced to geo\|ontologies fusion to implement concept lattice reduction in different granulations. Finally, experiments are conducted by combing fundamental geographic information data and spatial data in the hydraulic engineering domain. The results show that Geo\|ontologies fusion is a progressive process, and the proposed method is valid in multi\|source Geo\|ontologies fusion. As well as merits and limitations are discussed.