WFIP2 Experiment: Impact of Different Spatial Model Resolutions on Forecasting 80m Wind Speed Ramp Events

The Wind Forecast Improvement Project 2 (WFIP 2) is a DOE and NOAA led multi-agency project being conducted in Columbia River Gorge and Basin in the Oregon and Washington states. The major goal of this project is to understand and to improve the forecast skill of numerical weather prediction (NWP) models in complex terrain. The WFIP2 study region is well-known for its excellent wind resource, and many wind farms are installed there. Thus, improvements in NWP forecast skill in this region will have a direct and immediate impact on wind power generation and have the potential for improving grid balancing costs.

The WFIP2 project included the deployment of many types of instruments including wind profiling radars, sodars, lidars, microbarographs and others collecting measurements for a period of 18 months, from October 2015 through March 2017.  Because improved wind power production is the goal, winds in the atmospheric layer that spans the turbine blade heights are analyzed.

The special WFIP2 observational data are used for assimilation into the real-time operational and experimental models, and are also used for validation purposes. For our analysis we use only two experimental models, the NOAA/ESRL/GSD 3-km High Resolution Rapid Refresh (HRRR) and its nested version (HRRR_NEST) with a resolution of 750 m. They use almost identical physical parameterization schemes but their own corresponding topography. Because of this the principal differences in the wind flow should mostly depend on the detailed realization of the complex terrain in these models.

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.  A Ramp Tool and Metric (RT&M) was developed as part of the first WFIP experiment, to measure the skill of NWP models at forecasting ramp events. We use this tool on hub-height wind speed from available observed systems (mostly from the sodars) and the model data from HRRR and HRRR_NEST to confirm if the detailed realization of the complex topography improves the model skill of ramp forecasting.  Model skill is evaluated for different periods of the year, different locations within the WFIP2 domain, and for different meteorological conditions.