Description
An adequate estimate of IEC turbine siting parameters like 50-year return winds (U50s) is crucial in determining extreme environmental conditions for turbine siting both onshore and offshore. Highly reliable long-term wind speed time series are however rarely available for a measurement-based assessment, and therefore model-based estimates are required. Products like the Global Atlas for Siting Parameters (GASP, [1]) try to address this need by providing freely available atlases of, among others, 50-year return winds of high spatial resolution and near-global coverage downscaled from meteorological input from the Climate Forecast System Reanalysis (CFSR) product [2].While model-derived U50 estimates can already be highly valuable, complex modeling strategies as applied in GASP are impacted by internal model assumptions and external input data. It is thus crucial to also investigate model sensitivities to assess the reliability of the model-derived estimates. This is the aim of this study, which focuses on investigating the impact of external reanalysis input on model-derived U50s within the GASP workflow. In addition to CFSR, the European Centre for Medium-Range Weather Forecasts Reanalysis 5 (ERA5) and the Modern-Era Retrospective Analysis for Research and Applications version 2 (MERRA2) have been used as model input to calculate downscaled global U50s, and to evaluate the sensitivity of these U50 estimates. By introducing a metric for spread and a model divergence index of the three estimates, the study is able to identify critical regions of low/high variability or regions with implications on the choice of wind turbine classes. To quantify the performance, measurement-derived 50-year return winds from a range of quality-assessed long-term measurements from the global Tall Tower Data set [3] as well as literature values are used.
It is shown that model-derived U50s can be highly sensitive to the underlying reanalysis product whereby model spread depends on geographical location. Large deviations are found for complex sites with particular surface characteristics (among others large forests) due to deviations in surface representations and atmospheric conditions in the reanalysis products as well as in tropical cyclone affected areas in the Atlantic and Pacific ocean. The comparison with measurement-derived U50s shows a decent agreement at most stations given the uncertainty of the Gumbel-fit. The global atlases downscaled from the three reanalysis data can thus provide an estimate for a given location with better estimate of uncertainty.
[1] Larsén, X. G., Davis, N., Hannesdóttir, Á., Kelly, M., Olsen, B., Floors, R., Nielsen, M., & Imberger, M. (2021). Calculation of Global Atlas of Siting Parameters. DTU Wind Energy. DTU Wind Energy E No. E-Report-0208
[2] https://science.globalwindatlas.info/#/map
[3] Ramon, J., Lledó, L., Pérez-Zanón, N., Soret, A., & Doblas-Reyes, F. J. (2020). The Tall Tower Dataset: a unique initiative to boost wind energy research. In Earth System Science Data (Vol. 12, Issue 1, pp. 429–439). Copernicus GmbH. https://doi.org/10.5194/essd-12-429-2020
| Period | 26 May 2023 |
|---|---|
| Event title | Wind Energy Science Conference |
| Event type | Conference |
| Conference number | 4 |
| Location | Glasgow, United KingdomShow on map |