The purpose of this presentation is to present an icing frequency map for the US and southern Canada produced by using a model-derived climatology adjusted and validated through surface-based observations from long-term climate stations. Icing inflicts large economic losses throughout the U.S. and Canada, and can be responsible for upwards of 6% of energy production losses at existing wind farms, especially at higher elevations that are frequently in cloud. Thus, an accurate estimate of icing frequency is crucial in estimating overall energy production losses at proposed project sites.

Current methods for estimating icing frequency suffer from a lack of spatially representative observational data. Sites where icing is frequently observed are not generally near wind project sites, and may not even be at a similar elevation. It is consequently necessary to employ a physical model of some kind to estimate icing frequencies where measurements do not exist.

AWS Truewind's mesoscale model, MASS, provides the basis for such estimates. We recently completed 11 years of historical weather simulations on a 20 km resolution grid covering the United States and southern Canada. These simulations provide a baseline climatology of weather statistics at a higher resolution than is available from public sources such as global the North American Regional Reanalysis (NARR). Freezing rain and freezing drizzle frequencies can be extracted directly from the model runs. Likewise, rime ice frequency (frequency of deposition of ice in sub-freezing clouds) can be estimated from model-derived cloud frequency and cloud base temperature.

Here, we present a map of the icing frequency at the nominal mesoscale grid resolution for a range of heights above ground for the U.S. and Canada.