Examining the Role Played by Meteorological Ensemble Forecasts, Ensemble Kalman Filter Streamflow Assimilation, and Multiple Hydrological Models within a Prediction System Accounting for Three Sources of Uncertainty
Abstract:Building a hydrological ensemble prediction system (H-EPS) from an operational deterministic one may look like an easy task. Indeed, one has only to issue many forecasts at each time step instead of a single one. The problem gets much more complicated when that same person seeks the predictive distribution to be interpretable (reliable).
This presentation examine the role played by meteorological ensemble forecasts (meteorological uncertainty), Ensemble Kalman Filter (EnKF) streamflow assimilation (initial conditions uncertainty), and multiple hydrological models (structural uncertainty), combined in a nearly reliable H-EPS. The EnKF is shown to contribute largely to the ensemble accuracy and dispersion, indicating that the initial condition uncertainty is dominant. However, it fails to maintain the required dispersion throughout the entire forecast horizon and needs to be supported by a multimodel approach to take into account structural uncertainty. Moreover, the multimodel approach contributes to improve the general forecasting performance and prevents from falling into the model selection pitfall since models differ strongly in their ability. Finally, the use of probabilistic meteorological forcing was found to contribute mostly to long lead time reliability.
The H-EPS was implemented on 38 catchments (Québec, Canada) characterized by a dominant spring freshet and tested for 9-day ahead forecasts over a 2-year period. Twenty lumped hydrological models, chosen for their structural and conceptual diversity, were available to the project as well as ECMWF probabilistic meteorological weather forecasts.