Wind power forecasting is pivotal in promoting a stable and sustainable grid operation by estimating future power outputs from past meteorological and turbine data. The inherent unpredictability in wind patterns poses substantial challenges in synchronizing supply with demand, with inaccuracies potentially destabilizing the grid and potentially causing energy shortages or excesses. This study develops a data-driven approach to forecast wind power from 30 min to 12 h ahead using historical wind power data collected by the Supervisory Control and Data Acquisition (SCADA) system from one wind turbine, the Enercon/E92 2350 kW model, installed at Casa Nova, Bahia, Brazil. Those data were measured from January 2020 to April 2021. Time orientation was embedded using sine/cosine or cyclic encoding, deriving 16 normalized features that encapsulate crucial daily and seasonal trends. The research explores two distinct strategies: error prediction and error correction, both employing a sequential model where initial forecasts via k-Nearest Neighbors (KNN) are rectified by the Extra Trees Regressor. Their primary divergence is the second model’s target variable. Evaluations revealed both strategies outperforming the standalone KNN, with error correction excelling in short-term predictions and error prediction showing potential for extended forecasts. This exploration underscores the imperative importance of methodology selection in wind power forecasting.
Brazil's offshore wind resource is a strategic asset for the energy transition, yet its future variability remains insufficiently characterized. This study employs dynamical downscaling with WRF and MPAS to evaluate the spatial distribution and future changes in Brazil's offshore WPD from 2015 to 2050 under CMIP6 SSP2-4.5 and SSP5-8.5 scenarios. Simulations were evaluated against the CCMP wind dataset using standard statistical metrics for the CMIP6 historical period (2004–2014). Both models perform closest to CCMP in southern Brazil's EEZ, with the largest biases in the northern and northeastern regions. WRF exhibited lower Hellinger distances than MPAS (0.17–0.25 vs. 0.20–0.33), with lower values for WRF in 16 of 17 EEZs, indicating greater similarity to CCMP's WPD distribution. Both models' climatological mean fields indicate an increase in WPD, though they differ in magnitude: WRF shows a minimum increase of +20\%, while MPAS shows a maximum of +10\%. Still, both yield similar WPD magnitudes in the main hotspot regions-approximately 700–800 Wm⁻² off Piauí, Rio Grande do Norte, and southern Brazil, and 700 Wm⁻² off Rio de Janeiro. Fitted slopes for EEZ time series in hotspot areas were positive for RN and RJ (SSP2-4.5) and RS (SSP5-8.5), and negative for MA, PI, CE, and RS (SSP2-4.5) and MA, PI, CE, RN, and RJ (SSP5-8.5). Given MPAS's potential as a successor to WRF, this study offers the first assessment of its application to Brazil's energy sector, though further work is needed on hub-height wind representation and sensitivity to atmospheric conditions over northern and northeastern Brazil.