2026 Volume 7 Issue 1 Pages 261-266
The River Basin Data Platform (DPF), developed by MLIT, has been established to centrally manage various types of basin-related data. To support the realization of a river basin digital twin, this study identifies key requirements for data integration technologies and experimentally implements an ETL (Extract, Transform, Load)-centered framework in a River Basin Digital Testbed deployed on the DPF, evaluating its effectiveness and challenges. Specifically, metadata were attached during rainfall data extraction and format conversion processes to improve data reusability. Data lineage functions were also introduced to visualize and trace processing workflows, thereby ensuring transparency and reliability. Experiments conducted within the testbed environment demonstrated that API-based integration enables automated execution of runoff analysis. Furthermore, comparisons of different vCPU and RAM configurations in flood inundation simulations clarified their relationships with computation time and operational cost. The results demonstrate the effectiveness of ETL pipeline implementation using Microsoft Fabric and data catalog and lineage management using Microsoft Purview, while highlighting the importance of secure API integration for reliable system operation.
