Comparing WRF simulated high winds to observed wind gust distributions in the Pacific Northwest

Assessing the risk of extreme wind gusts during wind farm planning helps builders prepare for infrequent but potentially destructive high wind events. Because the wait time to collect sufficient on-site observations to describe the extreme tail of the winds gust distribution can be prohibitively long, either generalized “rule of thumb” or weibull-fit approaches are often used to predict extreme wind event recurrence and intensity. These methods may be insufficient under certain conditions such as in areas of complex terrain. A more attractive alternative to describe high events is to use downscaled model simulations that can account for the regional terrain and climatology of the site. However, the model likely has biases and resolution limitations that yield a different wind gust distribution than what is observed, and these differences must be corrected in post-processing. We examine the ability of the WRF model to predict the return time and intensity of high wind events at sites in the Pacific Northwest that have long term observational records. The predicted frequency and intensity of the WRF-simulated high wind events are compared to observed wind gust distributions at the towers. Using Extreme Value Theory, we fit an appropriate general extreme value (GEV) distribution to the simulated windspeeds, and we examine the sensitivity of the predicted values to various GEV fits. The fits among the different towers are studied to determine if the proper fit can be predicted from model conditions and site characteristics alone.