Many water users are affected by weather/climate variations (at short to seasonal ranges) and therefore there is a need to manage such variations to their advantage through better understanding. The production of hydrological forecasts involves a number of components (i.e. selection of meteorological/climate forecasts, pre-processing, hydrological model(s) and setup, calibration and initialization, updating, and generation, evaluation and visualization/communication of forecasts), which are strongly subject to multiple sources of error/uncertainty. To improve the hydrological forecast system, each of these components has to be evaluated and considered for refinements. In addition, there is a need to better understand the key factors influencing the spatial and temporal variation in hydrological predictability; bearing also in mind that the user needs, e.g. for water management or security, are mostly local. To improve the usefulness of forecasts, we are currently moving towards impact-based forecasts, in which the predictions are linked to local vulnerability and/or profit of the users.
To ensure usefulness, hydrological forecasts need to be unbiased, reliable and coherent. One way to achieve this is by refining pre-processing methods using multiple-data sources in hydrological forecasting. Use of multi-model approaches is another way, as different (climate or hydrological) models have strengths in capturing different aspects of reality. Finally, there is a gap between experts and users on the potential value in forecasts, which is mainly driven by the lack of knowledge of user needs. This points towards the need to consider new ways to communicate forecasts and their uncertainty.
What is the scientific gap?
- What are the limits of predictability in the hydrological forecasting systems and which are the drivers affecting the skill both in space and time for different water variables?
- How can we improve the local accuracy and link the forecasts to local user needs?
- How are the forecasts perceived by different users and how to better communicate hydrological forecasts at various spatial and temporal resolutions?
What are we doing?
- Evaluate the hydrological added value by using new products either for hydrological model state initialisation (e.g. assimilation of in-situ data, earth observations for different variables) or as meteorological forcing (e.g. probabilistic forecasts, high resolution NWP models).
- Analyse hydrological forecasts along the world’s hydro-climatic gradient to improve the understanding of the drivers that control the predictive skill.
- Assess the hydrological forecasting skill through a multi-model approach, including different meteorological forecasting systems and/or different hydrological models, and further identify best approaches for multi-model averaging.
- Improve the link to users by identifying their needs at different spatial and temporal resolutions, and improve visualisation tools for communicating uncertainty.
- Evaluate impact-based hydrological forecasts and provide the information about level of confidence in the forecast that would convey additional information for better decision-making.