The results of the national investigation and evaluation of mountain flood come in various data types and large data volume, with complicated data structure and high degree of specialization. The national data volume reaches 100T, and the average provincial-level data volume reaches TB scale. It is difficult to manage these data effectively. In this paper, we designed a data model for national investigation and evaluation of mountain flood based on administrative division and small watershed, and established an entity model of the objects and analyzed the relationships of the objects. We used ETL technology to realize multi-level aggregation and integration of massive data. On this basis, we designed and developed a data management platform based on the service-oriented software architecture, and realized various functions such as differentiated information organization of different management levels, multi-dimensional information correlation analysis, and online analysis and summary. The construction of this platform solved some problems in national mountain flood investigation and evaluation such as multi-source massive data management and differentiated demand of different business levels. It can provide reference for management and sharing of mountain flood investigation data at all levels.