In alpine valleys with strong urban development, river training works for flood safety become more and more difficult to implement because of economic and environmental constraints. Thus flood management has a great importance especially in river basins equipped with storage power plants having a large retention potential. To reduce the flood risk in the Upper Rhone River basin in the Cantons of Vaud and Valais, the MINERVE flood forecast system was developed. It aims an optimized flood management by taking advantage of the numerous existing high head power plants and reservoirs. The MINERVE flood forecast system was operational since 2006 with deterministic meteorological forecasts. Dr. Javier García Hernández improved and enhanced the system by implementing ensemble meteorological forecasts as well as an adapted decision making tool for preventive operations of the hydropower plants. This needed several scientific developments namely a combination of multi-attribute decision-making methodology with probabilistic forecasts for mathematical optimisation and a global procedure for solving a complex river basin with deterministic and probabilistic forecasts. The MINERVE system is now able to provide hydrological ensemble forecasts all over the Upper Rhone catchment area. Furthermore a new warning system tool was developed which allows producing warning reports. The warning system predicts the future time evolution of the hydrological situation at selected main checkpoints in the catchment area. Three warning levels during a flood event have been implemented depending on related critical discharge thresholds. Furthermore, in order to manage the multi-reservoir system during floods in an optimal way and to limit or avoid flood damages, optimization algorithms and procedures have been developed and tested. The most important scientific contribution of Dr. Javier García Hernández is the development of a decision support tool called MINDS (MINERVE Interactive Decision Support System), which allows real-time decision making based on hydrological forecasts. This tool suggests preventive operation measures of the hydropower plants such as turbine and bottom outlet releases in order to achieve an optimum economical use of the reservoirs, reducing the river discharge during the flood peak. The developed decision support system combines high-quality optimization of the system with a user-friendly interface that helps decision makers understanding the consequences of the preventive operation measures. Although MINDS has been specifically developed for the Upper Rhone River basin, the architecture of the system and its conceptual methodology can be applied to other cases in the field of water resources, flood warnings or reservoir management.