Estimating daily run-of-river hydropower generation from open hydrometeorological data: Limits of hydrology-based modelling

Hydropower is important for energy system modelling, where run-of-river hydropower is commonly represented as prescribed non-dispatchable generation. This study uses open hydrometeorological datasets to estimate the aggregated daily run-of-river generation and to assess the explanatory limits of hydrology-based modelling. ERA5, GEOGLOWS V2, and GloFAS v4.0 reanalyses are evaluated against reported generation for six European countries as a case study: France, Austria, Germany, Finland, Portugal, and Romania. Different input sets are compared within a consistent modelling framework, and the best-performing baseline model is used to diagnose remaining estimation errors. The results show that GloFAS river discharge provides the most suitable input for reproducing reported run-of-river generation, as it captures the dominant seasonal hydrological variability more robustly than the models with ERA5 and GEOGLOWS. The remaining error is the uncaptured short-term variation in reported generation. Additional tests indicate that accounting for reservoir generation provides limited improvements in model performance at the aggregated daily country level, while part of the residual is associated with electricity-price deviations, potentially reflecting short-term operation enabled by unclassified pondage within reported run-of-river generation. Accordingly, representing reported run-of-river generation as fully non-dispatchable in large-scale energy system models may overlook the potential short-term flexibility contained in the reported profiles The hydrology-based run-of-river estimates should therefore be interpreted as weather-dependent availability profiles, and the reproducible workflow in this study can produce such profiles for large-scale energy system studies across historical weather years beyond the period covered by reported data.