The smoother the better? A comparison of six post-processing methods to improve short-term offshore wind power forecasts in the Baltic Sea

Abstract. With a rapidly increasing capacity of electricity generation from wind power, the demand for accurate power production forecasts is growing. To date, most installations have been onshore and thus studies on focused conditions. However, as offshore becoming increasingly popular it also important to assess forecast quality in locations. In this study, high-resolution numerical weather prediction model AROME was used analyze performance an site Baltic Sea. improve forecasts, six post-processing methods were investigated their individual analyzed general well different speed ranges, boundary layer stratifications, synoptic situations low-level jet general, performed forecasting production, but applying smoothing or using random forest algorithm increased skill. Smoothing improved at all speeds, stratifications classes, method skill during jets. achieve best performance, we recommend selecting which use based forecasted Combining neighboring grid points, combining recent with yesterday linear regression correct earlier not fruitful increase overall quality.