Estimating threshold confidence for wind energy production

In planning wind projects, the question would inevitably arise in some form as to the likelihood that total wind energy production for an arbitrary period of time in the future will exceed a defined threshold. With time periods of interest extending far beyond the utility of numerical weather prediction, determining that confidence level mainly requires the application of statistical tools and a detailed knowledge of the variability of wind speed at time scales from minutes to many years.

Unfortunately, meteorological records for almost all discrete locations lack sufficient sampling to directly capture inter-annual wind climate variability and interpolated, long-record wind fields and climate forecasts fail to capture significant small-scale wind variations in space and time.

This paper describes a statistical method incorporating iso-factorial models to estimate threshold confidence for long-term wind energy production without resorting to parametric estimates of small scale wind speed probability density functions. In particular, these models are used to produce estimates of the probability of exceeding specific wind power density values within selected future time periods.