PNNL

Abstract

This short report provides a time-sensitive update on the 20-year wind resource data set (named “CA20”) produced in 2020 by the National Renewable Energy Laboratory (NREL) for the outer continental shelf (OCS) off the coast of California. Following the deployment of two floating lidars in the region in late 2020, NREL validated the CA20-modeled wind speeds against the lidar measurements using two different approaches: 1. Comparison of lidar observations against a 1-year extension of the CA20 dataset that overlapped with the lidar period of record (October 2020 – September 2021). 2. Comparison of the original 20-year CA20 dataset against a long-term adjusted lidar dataset that matched the CA20 period of record (2000-2019). The two validation approaches yielded different results. The first approach showed average biases in the 150-meter (m) modeled wind speeds of around 1.8 m s-1 at Humboldt and 0.7 m s-1 at Morro Bay. However, the lidar period of record at Humboldt was only 7 months, was seasonally biased, and had highly anomalous wind conditions. Therefore, bias results found in this short-term validation should be interpreted cautiously. The second approach showed average biases of 1.3 m s-1 at Humboldt and 1.0 m s-1 at Morro Bay for 150-m wind speeds, which NREL believes to be more representative of long-term CA20 performance. Despite these high overall biases, NREL found that the impact on long-term energy estimates is relatively small. Long-term gross energy estimates at both sites - calculated using 20 years of hourly timeseries data and the IEC 15 MW power curve - were found to be only 3% higher when using CA20 compared to the long-term adjusted lidar data. This lower energy bias occurs because most of the extreme CA20 wind speed biases are found in high wind speed regimes, where turbines are already operating at rated power. In the more critical 3-10 m s-1 wind speed range where wind speed biases are most impactful, the CA20 bias is lower. Based on this validation work, NREL concludes that use of the CA20 dataset is reasonable for annual energy production estimates and for assessing the economic value of the Humboldt and Morro Bay lease areas. However, on-site measurements will always be more accurate than any purely modeled data, and stakeholders are encouraged to leverage the publicly available lidar observations in addition to modeled wind products for their lease value assessments. Further research is ongoing to better understand the reasons behind the CA20 bias. Preliminary analysis suggests that the choice of the planetary boundary layer (PBL) scheme within the numerical model could be connected to a large portion of the observed bias, with the YSU PBL scheme providing a better match with the lidar observations compared to the MYNN scheme, that was used in CA20. A more in-depth analysis to confirm the issue and identify the physical drivers of the observed bias is underway. The results of this analysis, as well as a revised CA20 dataset, will be published by NREL in due course. In the meantime, NREL cautions against the use of CA20 for more detailed energy analyses (e.g., seasonal and diurnal trends) until the sources of bias are better investigated and understood.