One of the challenging tasks in data management and interoperation is efficient data integration. Data integration in scientific community is affected by heterogeneities of the data and diversity of the participating data sources. We have developed a web-based data integration framework, HIDE, for hydrology datasets and data sources. The system is built on an extended XML mediator architecture with modular methodology and software engineering practices of simplicity, extensibility and flexibility. HIDE promotes on-demand data integration and carries functionalities such as search and data access, data integration and remote data visualization. Registration of a new data source to HIDE for integration involves minimal effort of adding a few XML metadata files. The system can help hydrologists to access and share, and dynamically integrate data from diverse data sources quickly. HIDE was primarily built for integrating data from National Weather services with datasets from various autonomous organizations.