In recent years, ancient sites in various locations have been frequently threatened by flood disaster, but there is no unified monitoring strategy; the monitoring methods using sensors, remote sensing, and other methods as well as the related monitoring data obtained are complicated and difficult to manage. Existing monitoring systems do not accurately reflect the relationship between the components of the system and the data. Knowledge graphs have attracted attention as semantic networks that can intuitively reflect the relationship between knowledge entities. In this study, a method of constructing a flood disaster risk knowledge graph for ancient sites was proposed and realized. First, the definition of a risk monitoring index system for ancient sites for flood disaster is proposed. Then, the knowledge graph structure of the monitoring index system is reorganized using semantic reasoning techniques, and the monitoring index bodies are extracted. The proposed method was used to monitor the Pujindu site in China. The results show that the knowledge graph has the advantages of visualization and a clear structure that can intuitively represent the relationship between entities, manage monitoring methods and data such as sensor and remote sensing data, and be effectively applied to the flood disaster risk monitoring of ancient sites.