Verification of Wind Forecasts from the High-Resolution Rapid Refresh

Wintertime high wind events are common over the high plains of the United States. These events have significant societal impacts, including truck blowover risks, road closures, and blowing snow, which significantly decreases visibility. These events are difficult to forecast over the high plains in part due to the coarse resolution of Numerical Weather Prediction (NWP) models, which cannot adequately capture the complex terrain of this region, as well as the relatively sparse network of observing sites used to initialize these models. However, the High-Resolution Rapid Refresh (HRRR) model has a higher resolution than most NWP models at 3 km and can thus better represent the complex terrain over the high plains. Its forecasts are commonly used as guidance to NWS offices with frequent high wind events. However, its ability to accurately predict these events has not been analyzed. To validate the model, Mesowest data are compared with HRRR-predicted winds and wind gusts for three winter season (2016-2017, 2017-2018, and 2018-2019) which correspond to two different versions of the HRRR model (version 2 and version 3). The preliminary findings of this work indicate that the HRRR model overestimates low wind speeds and gusts and significantly underestimates stronger wind speeds and gusts. We explore how these biases change as a function of forecast lead time, and we also examine the potential of other HRRR-predicted variables as proxies for high wind events.