Development and Assessment of an Operational Hydrological Forecasting System for Yacyretá Hydropower System

Hydrological forecasting is a fundamental service necessary for an effective reservoir operation. It enables anticipatory decisions that improve the management of the system considering operating goals, constraints and requirements. The operational forecasting systems rely on a cascade of processes that involve importing meteorological forecasts, forcing the hydrological models with meteorological variables, routing the hydraulic wave through the river channel, correcting errors, among others. The system should also pre-process, transform, and validate data on a daily operation.

In this study, we present the developments of an operational hydrological forecasting system for Yacyretá reservoir using Delft-FEWS platform. The system takes information from the deterministic global numerical weather forecasts (NWF) GFS, WRF, as well as the local NWF SMN, and produces expected reservoir inflows up to 15 days ahead. It also includes the use of ensemble forecasts derived from the NWF GEFS. The hydrological models implemented in the forecasting system include the well-known HEC-HMS and HEC-RAS, as well as an internal operation model developed by Entidad Binacional Yacyretá. The system makes use of information from upstream reservoirs such as Itaipú and Salto Caxias to determine the reservoir inflows of Yacyretá.

We assess the performance of the forecasts using verification metrics such as the Continuous Ranked Probability Score (CRPS), Brier Score (BS) and rank histograms, and present the results for a testing period. Future improvements of the system include the integration of a real-time optimization of the reservoir using anticipatory control as well as the integration of upstream and downstream hydrological models.