The Variable Relationship between Observed and Reanalysis Wind Speeds: Effects on P50 Bias

During the measure-correlate-predict (MCP) process, data from an on-site meteorological tower are correlated to a long-term reference dataset. This correlation is then used to estimate the historical on-site wind resource. A key assumption of MCP is that the relationship between the target and reference site determined during the training period is valid for other time periods. However, if this relationship changes, wind speeds estimated from MCP can be significantly biased.

In this presentation, the effects of a changing target/reference relationship on MCP bias are demonstrated for 20 meteorological towers in the United States. For ~1/3 of these towers, the changing relationship between observed and re-analysis wind speeds is related to interannual wind speed variability. At these sites, re-analysis datasets do not adequately capture high and low wind speed anomalies related to large-scale climate oscillations, leading to temporal variations that can bias a site’s energy prediction.