Evaluating the quality of scenarios of short-term wind power generation

Scenarios of short-term wind power generation are becoming increasingly popular as input to multistage decision-making problems e.g. multivariate stochastic optimization and stochastic programming. The quality of these scenarios is intuitively expected to substantially impact the benefits from their use in decision-making. So far however, their verification is almost always focused on their marginal distributions for each individual lead time only, thus overlooking their temporal interdependence structure. The shortcomings of such an approach are discussed. Multivariate verification tools, as well as diagnostic approaches based on event-based verification are then presented. Their application to the evaluation of various sets of scenarios of short-term wind power generation demonstrates them as valuable discrimination tools.

► Presentation of the desirable properties of wind power generation scenarios.
► Description of various evaluation frameworks (univariate, multivariate, diagnostic).
► Highlighting of the properties of current approaches to scenario generation.
► Guidelines for future evaluation/benchmark exercises.