Abstract: Big data have brought innovations as well as challenges to Earth science research. However, due to inconsistent data description standards, unclear data sharing mechanism and significant semantic heterogeneity, there are significant difficulties existed in big data integration, sharing and reuse in Earth science research. The knowledge graph can be used to explicitly represent concepts and their complex semantic relationships in a machine-understandable way. Therefore, it has been widely applied for semantic translation, data integration and data reuse. In order to establish Earth science knowledge graph, the current paper analyzes the characteristics of the existing knowledge graphs and investigates the main construction methods. Relationships of data dictionary, knowledge system and knowledge graph are also illustrated and analyzed. Current Earth science thematic databases and domain ontologies have also been reviewed, and future perspectives on Earth science knowledge graph applications are provided.

Key words: Earth science, knowledge graph, knowledge system, big data, artificial intelligence