How is renewable energy generation forecasted?

Renewable forecasting combines: weather prediction models (NWP - numerical weather prediction for days ahead), satellite data (cloud tracking for hours ahead), ground measurements (real-time conditions), machine learning (improve NWP, learn local effects), ensemble methods (combine multiple models), and persistence (assume recent conditions continue). Time horizons: intraday (dispatch, trading), day-ahead (unit commitment, markets), week-ahead (maintenance planning). Accuracy varies: solar 2-5% RMSE day-ahead, wind 5-15%. Better forecasting reduces balancing costs and enables higher renewable penetration. Probabilistic forecasts quantify uncertainty.