Identifying meteorological drivers for errors in modeled winds along the Northern California Coast

An accurate wind resource dataset is required for assessing the potential energy yield of floating offshore wind farms that are expected along the California outer continental shelf (OCS). The National Renewable Energy Laboratory (NREL) has developed and disseminated an updated wind resource dataset for the OCS, using the WRF model, referred to as the CA20 dataset. As compared to buoy lidar measurements that have become available recently, the CA20 dataset showed significant positive biases for 100-m wind speeds along northern California wind energy lease regions. To investigate the meteorological drivers for the model errors, we first consider two 1-year simulations run with two different planetary boundary layer (PBL) parameterizations: the Mellor-Yamada-Nakanishi-Niino (MYNN) PBL scheme (chosen configuration in the CA20 dataset) and the Yonsei University (YSU) PBL scheme (which significantly reduces the bias in modeled winds). By comparing the 1-year simulations to the concurrent lidar buoy observations, we find that errors are larger with the MYNN PBL scheme in warm seasons. We then dive deeper into the analysis by running simulations for short-term (3-day) case studies to evaluate the sensitivity of initial/boundary condition forcings on model results. By analyzing the short-term simulations, we find that during synoptic scale northerly flows driven by the North Pacific High (NPH) and inland thermal low, the coastal land-sea thermal contrast further accelerates the northerly winds superimposed on the generally northerly flow via the thermal-wind mechanism. The resultant low-level jet is often observed along the coast. The diurnal variation of the sea-breeze forcing alters the location and strength of the jet, which leads to the diurnally varying wind speed regime observed offshore Humboldt California in warm seasons. A coastal warm bias in the MYNN simulation is mainly responsible for the modeled wind speed bias by altering the boundary layer thermodynamics and accelerating the northerly winds further, especially from morning to early afternoon. Consequently, the diurnal variation in wind speed is muted, and a large positive bias in MYNN-simulated wind speed is observed. Our findings suggest having access to observed hub-height wind speed is essential to ensure an accurate model validation, as well as validating other wind-related variables such as surface temperature where coastal thermal circulations could matter. The results of our analysis will help guide the creation of an updated version of the CA20 dataset.