The Effect of Offshore Wind Capacity Expansion on Uncertainties in Germany’s Day-Ahead Wind Energy Forecasts

Abstract

Germany has experienced rapid growth in onshore wind capacities over the past two decades. Substantial capacities of offshore wind turbines have been added since 2013. On a local, highly-resolved level, this analysis evaluated if differences in wind speed forecast errors exist for offshore and onshore locations regarding magnitude and variation. A model based on the Extra Trees algorithm is proposed and found to be a viable method to transform local wind speeds and capacities into aggregated wind energy feed-in. This model was used to analyze if offshore and onshore wind power expansion lead to different distributions of day-ahead wind energy forecast errors in Germany. The Extra Trees model results indicate that offshore wind capacity expansion entails an energy forecast error distribution with more frequent medium to high deviations, stemming from larger and more variable wind speed deviations of offshore locations combined with greater geographical concentration of offshore wind turbines and their exposure to high-wind oceanic conditions. The energy forecast error distribution of onshore expansion, however, shows heavier tails and consequently more frequent extreme deviations. The analysis suggests that this can be rooted in the simultaneous over- or underestimation of wind speeds at many onshore locations.

Keywords

• wind energy forecasts
• offshore capacity expansion
• day-ahead wind energy uncertainties
• machine learning
• Extra Trees