Streamflow Prediction Validation with the Application of Large

Streamflow Prediction Validation with the Application of Large Basin Rainfall-Runoff Model Department of Civil and Environmental Engineering Universidade Federal do Paraná Curitiba, Paraná Brasil Alexandre Guetter Second EUROBRISA Workshop - Dartmoor, Devon, UK - 21 -24 July 2009

Overview of the Brazilian Hydropower System • Total annual power production: 445 TWh • Annual rate of increase: 6% • State-run companies: 90% • Demand: Hydropower (>30 MW): 72. 6% Hydropower (<30 MW): 1. 7% Brazil Thermal: 14. 7% Nuclear: 2. 5% Import: 8. 5% World Second EUROBRISA Workshop - Dartmoor, Devon, UK - 21 -24 July 2009

Overview of the Brazilian Hydropower System • Electricity Demand per Sector: Industrial: 47% Residential: 22% Services: 14% Other: 17% • Hourly Demand Curve Peak at 6 pm (residential) Base: hydro+thermal Peak: hydro Second EUROBRISA Workshop - Dartmoor, Devon, UK - 21 -24 July 2009

Power production Integrated System • Different regional climates grouped on a continental scale • Installed Capacity Southeastern/Midwest: 63% Southern: 17% Northeastern: 14% Northern: 6% • Seasonal Precipitation Forecasting Skill Adequate: SU-NE-NO Low: SE Second EUROBRISA Workshop - Dartmoor, Devon, UK - 21 -24 July 2009

Hydropower Integrated System • Well-defined rainy season Southeastern/Midwest: SON-DJF Northeastern: FMAM (largest storage) • Capacidade instalada Installed Capacity Quasi-uniform rainfall Southern (smallest storage) • What do we want to predict: Southeastern: in August if SON will have little rainfall – energy production Southeastern: in November if DJF will be very rainy – flood control Energy Storage Energia armazenada máxima Northeastern: in January if FMAM will have little rainfall Second EUROBRISA Workshop - Dartmoor, Devon, UK - 21 -24 July 2009

The size of our problem: • About 60 large hydropower stations: require seasonal streamflow prediction Second EUROBRISA Workshop - Dartmoor, Devon, UK - 21 -24 July 2009

Pilot Study: Streamflow Prediction Validation • 20 prediction points in 11 large basins • 9 basins with 2 points: the most upstream/downstream hydroplants Second EUROBRISA Workshop - Dartmoor, Devon, UK - 21 -24 July 2009

Pilot Study: Activities • Naturalized streamflow analysis (1931 -2007) • Diagnostic study of the association between high/low flows and SST anomalies (what is the climate model skill for such forcings? ) • Data collection: raingauge mean-areal rainfall (variable period) • Large-scale rainfall-runoff model calibration • Streamflow assimilation: hydrologic model state updating • Validation study: use monthly precipitation forecast as input data for the state-updated rainfall-runoff model (1981 -2005) Second EUROBRISA Workshop - Dartmoor, Devon, UK - 21 -24 July 2009

Naturalized Streamflow Analysis – Iguaçu River Second EUROBRISA Workshop - Dartmoor, Devon, UK - 21 -24 July 2009

Naturalized Streamflow Analysis – Iguaçu River Second EUROBRISA Workshop - Dartmoor, Devon, UK - 21 -24 July 2009

High Flows and SST anomalies – Iguaçu River High Flows Low Flows Second EUROBRISA Workshop - Dartmoor, Devon, UK - 21 -24 July 2009

High Flows and SST anomalies – Iguaçu River High Flows Low Flows Second EUROBRISA Workshop - Dartmoor, Devon, UK - 21 -24 July 2009

Streamflow Forecasting Using Either Resampled or Predicted Rainfall Ensemble Streamflow Forecast Dynamic Model Forecast ENSO Southern Atlantic Run Updated Hydrologic Model Select Precipitation Data from Historic Record Associated with Forecast Climate Category

Rainfall – Runoff Model Second EUROBRISA Workshop - Dartmoor, Devon, UK - 21 -24 July 2009

Rainfall-Runoff Model With State-Updating Second EUROBRISA Workshop - Dartmoor, Devon, UK - 21 -24 July 2009

Rainfall-Runoff Model With State-Updating Hydrologic Model State Updating αu e αp : fatores de ponderação, a calibrar U : covariance matrix of input errors W : covariance errors of model parameters M(t) : input data sensitivity matrix N(t) : model parameter sensitivity matrix Second EUROBRISA Workshop - Dartmoor, Devon, UK - 21 -24 July 2009

Streamflow Prediction Using Resampled Rainfall Second EUROBRISA Workshop - Dartmoor, Devon, UK - 21 -24 July 2009

Streamflow. Prediction Using Climate Model Rainfall Streamflow Using Resampled Rainfall * This results are not for the Iguaçu Basin Highest Simulation (1981 -2005) Forecast Ensemble Lowest Simulation (1981 -2005) Second EUROBRISA Workshop - Dartmoor, Devon, UK - 21 -24 July 2009

Conclusions and Recommendations • The diagnostic study of the relationship between high/low flows and SST anomalies was accomplished in year-1; • The collection of raingauge precipitation data and the rainfall -runoff model calibration was accomplished in year-2; • The streamflow prediction validation using seasonal rainfall prediction has not been completed yet; • The sensitivity analysis of energy production to rainfall prediction skill is a task for year-3; • Selection of period and basin for joint papers. Second EUROBRISA Workshop - Dartmoor, Devon, UK - 21 -24 July 2009