Operations Reports

Forecasting is an essential tool the ISO uses in managing the variability and uncertainty of wind and solar generation. With any type of forecast, there is a difference between the actual value of a variable and its predicted value. Mean Absolute percentage Error (MAE) and Bias are industry standard statistics used to quantify the error in a forecast. MAE summarizes the accuracy (i.e., how closely the forecast matches observations) and Bias summarizes the amount of systematic error (i.e., whether the forecasts consistently over or under predict). Both MAE and Bias are normalized by the nameplate capacity. Ideally, both would be equal to zero; and typically, both generally increase with the forecast horizon. MAE and Bias for a fleet of resources are less than for individual plants due to offsetting errors. Positive bias means less wind or solar power was actually observed compared to forecast. Negative bias means more wind or solar power was actually observed compared to forecast. The MAE and Bias are developed for the Short Term, Medium Term, and Long Term forecasting timeframes. Statistics for the Medium Term (1 Hour Ahead to 48 Hours Ahead) and Long Term (49 Hours Ahead to 168 Hours Ahead) forecasts are shown together.