Global wind power assessments present an open, high-resolution simulation workflow

An unedited manuscript introduces ETHOS.RESKit, an open-source global wind power simulation workflow validated against millions of meteorological and turbine measurements and reporting low errors at site and country scales.

An unedited manuscript describes ETHOS.RESKit, an open-source workflow for high-resolution global wind power simulation. It uses the new Global Wind Atlas 4 and the fifth iteration of the European Centre for Medium-Range Weather Forecasts reanalysis to simulate time-resolved turbine output for onshore and offshore designs. The authors report extensive validation using over 18 million meteorological mast measurements and 8 million wind turbine-site measurements from 2002–2021 across six countries. The manuscript is provided for early access and may be edited before final publication. Key findings: - ETHOS.RESKit is presented as a transparent, open-source global wind power simulation workflow capable of high spatial resolution and customizable turbine designs. - The workflow uses Global Wind Atlas 4 and the fifth-iteration ECMWF reanalysis as primary inputs for time-resolved simulations. - Validation included more than 18 million meteorological mast measurements for wind speed bias correction and over 8 million measurements from turbine sites across six countries during 2002–2021. - Compared with measured energy yield at turbine sites, the authors report a global average capacity factor mean error of 5.6% and a Pearson correlation of 0.844. - Evaluations against aggregated operational wind farm production and country-level wind generation statistics show a mean error of 0.6% after a final calibration step aligning simulations with real-world statistics. Summary: The workflow aims to improve consistency and transparency in wind resource and power estimates across spatial and temporal scales. The reported validation shows close agreement with measured site yields and national statistics, and a final calibration step is used to align simulations with observed generation. The manuscript is unedited and will undergo further editing before formal publication.