Evaluating the Methodology of Assessing Interannual Variability of Wind Speed

Since wind resources vary from year to year, the inter-annual variability (IAV) of wind speed is a key component of the overall uncertainty in the wind resource assessment process and causes challenges to wind-farm operators and owners. As wind turbine power generation is a function of wind speed, IAV of wind resources has important implications for power production. The length of the IAV analysis period also influences the resulting IAV estimate. As many approaches of calculating IAV exist, the goal of this study is to identify the methods most appropriate for different situations. We review and compare the various IAV methods, such as the Coefficient of Variation (CoV) and the Robust Coefficient of Variation (RCoV). We use the net monthly energy production of wind farms from the Energy Information Administration (EIA) and the wind speed data from NASA’s MERRA2 reanalysis dataset. After filtering the energy data via the R² from linear regression, we assess the accuracy of the IAV calculations by comparing the wind speed IAV estimates to the IAV of actual wind-farm power production. Results from statistically robust and resistant methods yield higher correlations between wind speed IAV and energy production IAV, as robust methods do not assume Gaussian wind speed distributions and resistant methods are insensitive to wind speed extremes. Finally, we recommend a systematic approach for estimating IAV. Depending on the exact method chosen, data for 3-11 years are generally required for accurate IAV estimates.

Figure 1: Map of wind farms: The pre-filter EIA wind farms are in blue, the R²-filtered wind farms are in red, and the correlation-filtered wind farms are in white.