Skip to Main content Skip to Navigation
New interface
Journal articles

Skill forecasting from ensemble predictions of wind power

Abstract : Optimal management and trading of wind generation calls for the providing of uncertainty estimates along with the commonly provided short-term wind power point predictions. Alternative approaches for the use of probabilistic forecasting are introduced. More precisely, focus is given to prediction risk indices aiming to give a comprehensive signal on the expected level of forecast uncertainty. Ensemble predictions of wind generation are used as input. A proposal for the definition of prediction risk indices is given. Such skill forecasts are based on the spread of ensemble forecasts (i.e. a set of alternative scenarios for the coming period) for a single prediction horizon or over a look-ahead period. It is shown on the test case of a Danish offshore wind farm how these prediction risk indices may be related to several levels of forecast uncertainty (and potential energy imbalances). Wind power ensemble predictions are derived from the conversion of ECMWF and NCEP ensemble forecasts of meteorological variables to wind power ensemble forecasts, as well as by a lagged average approach alternative. The ability of prediction risk indices calculated from the various types of ensembles forecasts to resolve among situations with different levels of uncertainty is discussed.
Document type :
Journal articles
Complete list of metadata

Cited literature [27 references]  Display  Hide  Download
Contributor : Magalie Prudon Connect in order to contact the contributor
Submitted on : Friday, April 12, 2013 - 9:19:07 AM
Last modification on : Saturday, October 22, 2022 - 3:29:40 AM
Long-term archiving on: : Saturday, July 13, 2013 - 4:02:15 AM


Files produced by the author(s)



Pierre Pinson, Henrik Aa. Nielsen, Henrik Madsen, Georges Kariniotakis. Skill forecasting from ensemble predictions of wind power. Applied Energy, 2009, 86 (7-8, July-August 2009), pp.Pages 1326-1334. ⟨10.1016/j.apenergy.2008.10.009⟩. ⟨hal-00812349⟩



Record views


Files downloads