Hydropower Forecasting in Brazil

Most of the electric power in Brazil comes from hydropower, and the short-term power production at each of the major power plants in Brazil is optimized using streamflow forecasts of lead times up to 14 days. These forecasts were usually obtained using stochastic models based only on the last observed streamflow values. During the last few years, rainfall–runoff models that use predicted rainfall as the main input start to replace the stochastic models. However, this new model generation still uses deterministic precipitation forecasts and does not take advantage of the ensemble precipitation forecasts that are already available in Brazil from regional and global meteorological models. Based on recent research results, it is likely that ensemble streamflow forecasts outperform deterministic forecasts in application to short-term reservoir management for objectives such as energy generation and flood mitigation. This chapter presents an assessment of 4 years of ensemble inflow forecasts to a major hydropower reservoir in Brazil, the Três Marias dam, on the São Francisco River. A 14 member ensemble obtained from a global numerical weather prediction model of the Brazilian Center for Weather Prediction is used, and results are evaluated in terms of ensemble applicability for a period between 2008 and 2012. Results are encouraging, and due to this it is believed that ensemble inflow forecasts to major reservoirs in Brazil will be used in a near future as input to the optimization of the national electric power producing system.