. Impact of improved parameterizations in reforecast runs from the HRRR and HRRRNEST models on 80-m wind speed bulk and ramp statistics during the second Wind Forecast Improvement Project (WFIP2)

Abstract
The second Wind Forecast Improvement Project (WFIP2) is a U.S. Department of Energy (DOE) and National Oceanic and Atmospheric Administration (NOAA) funded program, with private-sector and university partners, which aims to improve the accuracy of numerical weather prediction (NWP) model forecasts of wind speed in complex terrain for wind energy applications. During WFIP2 (Oct 2015 - Mar 2017, Columbia River Gorge and Basin area) several improvements to the parameterizations applied in the High Resolution Rapid Refresh (HRRR - 3 km horizontal grid spacing) and the High Resolution Rapid Refresh Nest (HRRRNEST - 750 m horizontal grid spacing) NWP models were tested during four 6-week reforecast periods (one for each season). For these tests the models were run in control (CNT) and experimental (EXP) configurations, with the EXP configuration including all the improved parameterizations. Since the WFIP2 study region is well-known for its excellent wind resource, many wind farms are installed there. For this reason in this study we focus the attention on the 80-m wind speeds (hub height), using observations collected by 19 sodars and 3 profiling lidars for verification. First, the impacts of the experimental parameterizations on the forecast of 80-m wind speeds from the HRRR and HRRRNEST models are assessed, using standard bulk statistics such as Mean Absolute Error (MAE) and Mean Bias Error (bias). Improvements due to the experimental physics (EXP vs CNT runs) versus those due to finer horizontal grid spacing (HRRRNEST vs HRRR), and the combination of the two are compared. The improvements are evaluated as a function of the model's initialization time, forecast horizon, time of the day, season of the year, site elevation, and meteorological phenomena, also looking for the causes of model weaknesses. Second, since one of the biggest challenges for wind power production is the accurate forecast of wind ramp events (i. e. large changes of generated power over short periods of time) in this study we will also measure the skill of the models in their CNT and EXP configurations explicitly at forecasting wind ramp events, using a Ramp Tool and Metric (RT&M) developed as part of the first WFIP experiment, held in the U.S. Great Plains (Sep 2011 - Aug 2012). The seasonal and diurnal distribution of ramp events in both the observations and models are analyzed, as well as model skill at forecasting ramp events, including a comparison between the CNT and EXP runs.