Presentation Description: One of the key deliverables from a wind resource assessment is an estimate of the P50, the median annual generation expected during the lifetime of the project. However, generation during specific years or quarters of the project’s lifetime can deviate significantly from expectations. While part of this deviation can be the result of model error, part of it can also be the result of a wind resource that’s significantly different from “normal”. In order to place these deviations in context, wind anomalies are often computed at the quarterly and annual levels which show the wind resource expressed as a percent deviation above or below normal. These calculations are also often performed across large regions and presented as wind anomaly maps. These wind anomaly estimates are helpful when discussing realized plant revenue with investors or removing the effects of windiness from the results of P50 accuracy benchmarking studies.
Despite the large emphasis placed on wind anomalies, significant uncertainty surrounds these estimates. In particular, the magnitude and even the sign of wind anomalies can vary depending on how the “normal” wind resource is defined. In this presentation, the sensitivity of wind anomaly calculations to a variety of factors will be demonstrated, including the selection of re-analysis dataset and the averaging period used to calculate the normal wind resource. The limitations of using both modeled and observed data to calculate anomalies will be demonstrated using re-analysis data and observed data from operational wind farms and meteorological towers. Finally, a proposal will be presented wherein wind anomalies are expressed with uncertainty bounds that indicate the degree of certainty in the anomaly estimate.
Methodology: Interactive polling software will be used to poll the audience on how they typically estimate wind anomalies.