Using WRF-LES to Simulate Wind Plant In-Flow Conditions in a Region of Complex Terrain: Application to the Second Wind Forecast Improvement Project (WFIP2)

Using numerical simulations to resolve the finest turbulent scales present in the planetary boundary layer helps us to study the processes that occur within that layer, such as the turbulent inflow condition to the wind plant and the generation of wakes behind wind turbines. In this work we employ several nested domains in the WRF-LES framework to simulate conditions in a convectively driven cloud free boundary layer at an instrumented site within the Department of Energy's Second Wind Forecast Improvement Project (WFIP2) study domain. The innermost LES domain (30 m spatial resolution) receives the boundary forcing from two other coarser resolution LES outer domains, which in turn receive boundary conditions from two WRF-mesoscale domains. Wind and virtual temperature data from sonic anemometers mounted at two vertical levels (30 m and 60 m) deployed as part of the previous Columbia Basin Wind Energy Study (CBWES) are compared with the LES results in term of first and second statistical moments as well as power spectra and distributions of wind velocity. Simulations are run using two common boundary layer parameterizations (MYNN and YSU) tested in the WRF mesoscale domains, the MYNN scheme shows slightly better agreement with the observations for some quantities, such as time averaged velocity and Turbulence Kinetic Energy (TKE). However, LES driven by WRF-mesoscale simulations using either parameterization have similar velocity spectra and distributions of velocity. For each component of the wind velocity, WRF-LES power spectra are found to be comparable to the spectra derived from the measured data (for the frequencies that are accurately represented by WRF-LES). The LES simulations are then analyzed further to investigate the TKE budget and to identify the source and sink terms that are most important in an area of complex terrain. This information could be used to evaluate existing turbulence parameterizations applied in mesoscale models, or to develop improved representations that yield better predictions of hub-height winds and turbulence intensity.