Understanding how hydrological forecast quality impacts the management of hydroelectric reservoirs

The evaluation of inflow forecast quality and value is essential in hydroelectric reservoir management. Forecast value can be quantified by the economic gains obtained when optimizing hydroelectric reservoir operations informed by weather and hydrological forecasts. This study [1] investigates the impact of 7-day streamflow forecasts on the optimal management of hydroelectric reservoirs and the associated economic gains. Flows from ten catchments in France are synthetically generated over a 4-year period to obtain forecasts of different quality in terms of accuracy and reliability. These forecasts define the inflows to ten hydroelectric reservoirs, which are conceptually parametrized. Each reservoir is associated to a downstream power plant with yield 1 which produces electricity valued with a price signal. The system is modelled using linear programming. Relationships between forecast quality and economic value (hydropower revenue) show that forecasts with a recurrent positive bias (overestimation) and low accuracy generate the highest economic losses when compared to the reference management system where forecasts are equal to observed inflows. The smallest losses are observed for forecast systems with under-dispersion reliability bias, while forecast systems with negative bias (underestimation) show intermediate losses. Overall, the losses (which amount to millions of Euros) represent approximately 1% to 3% of the revenue over the study period. Besides revenue, the forecast quality also impacts spillage, stock evolution, production hours and production rates. For instance, forecasting systems that present a positive bias result in a tendency of operations to keep the storage at lower levels so that the reservoir can be able to handle the high volumes expected. This impacts the optimal placement of production at the best hours (i.e. when prices are higher) and the opportunity to produce electricity at higher production rates. Our study showed that when using biased forecasting systems, hydropower production is not only planned during more hours at lower rates but also at hours with lower median prices of electricity. The modelling approaches adopted in our study are certainly far from representing all the complexity of hydropower management under uncertainty. However, they proved to be adapted to obtaining the first orders of magnitude of the value of inflow forecasts in elementary situations.

[1] https://doi.org/10.5194/hess-2020-410