Fuzzy geospatial information processing is an important topic in geospatial semantic retrieval (GSR). Representing fuzzy information in a machine understandable way and reasoning them efficiently can make a GSR system more practical and robust. This paper proposes an innovative fuzzy geospatial ontology model, FGSO, to represent and reason fuzzy geospatial information. In FGSO, the complex fuzzy semantic of spatial relations are described explicitly by a fuzzy spatial relation ontology which also carries out fuzzy spatial reasoning; a fuzzy geospatial ontology based on the fuzzy ontology language, f-SHIF(D), is developed to represent a variety of geospatial information besides spatial relations in a uniform format, and its reasoning module integrates the two ontologies to mine more information. Based on FGSO, a fuzzy geospatial semantic retrieval system, FGSRS, is constructed. Experiment results show that FGSRS has good performance on Recall and Precision.