| 07 Aug 2026
Assessing Future Wind Speed Variations: Methodology for Climate Model Integration in Wind Resource Assessment
Abstract. Current wind resource assessments primarily rely on historical wind climate data and typically assume persistence in the wind conditions for the future climate. However, climate change influences the large-scale atmospheric circulation and thus might impact the available wind resource at specific sites. These possible changes in the wind climate are represented in climate model simulations. In this work, we introduce a methodology that integrates climate model ensemble data into the long-term referencing process using the Measure-Correlate-Predict (MCP) framework to assess future wind speed variations at specific sites. Seven sites across Europe with measurements near modern wind turbine hub heights (100 m) are analyzed. The approach incorporates the ERA5 reanalysis dataset and an ensemble of ten global climate models from the Coupled Model Intercomparison Project (CMIP6) under intermediate and very high green house gas emission scenarios (SSP2-4.5 and SSP5-8.5). The methodology is validated against historical data and the results indicate that the approach reliably corrects climate model data to the specific site conditions. The bias correction reduces discrepancies between climate model outputs and observations, within all seasons. Depending on the site and climate model, future projections indicate a decrease in summer wind speeds ranging from −0.1 to −0.3 ms−1 by mid-century and up to −0.4 ms−1 by century’s end under SSP2-4.5, with slightly larger declines under SSP5-8.5. Conversely, a slight increase in winter wind speeds is observed, accompanied by greater uncertainty among models. Overall, the proposed methodology provides a robust, data-driven basis for improved future wind resource estimation, supporting investment decisions and enhancing understanding of risks under climate change.
The manuscript details a methodology of integrating climate model ensemble data for two emission scenarios into the long-term referencing process using incorporates the ERA5 reanalysis dataset, by applying the Measure-Correlate-Predict (MCP) framework to assess future wind speed variations at specific sites. Usual approach in the industry is using ERA5 data without consideration of future variations in wind speed. This work proposes an alternative method which assessment of future wind speed variations across Europe, providing better input for placement of future wind farm.
While I do support publication of this work, I have a few questions and suggestions:
Line 49: “Especially regional studies are based on the former generation of climate…” I’m not sure that “Especially” is the right word to be used here. I would just omit it.
Table 1: In addition to Table 1. It would be useful to “borrow” maps from the work done by Borowski et al. (2026), showing geographic location of selected sites, so that reader has a better idea about any local geographic features that might affect past/present/future wind resources.
Table 2: It would be nice to include a sentence or two about if and how horizontal resolution of different global climate models from the CMIP6 ensemble might affect projections of wind speed.
Line 240: “These uncertainties were investigated in detail in Borowski et al. (2026) and are in the range from 1 %(Falkenberg) to approx. 11 % (Obninsk).” I believe it would be useful to include a bit more info about where this big range of uncertainty is coming from (why is Obninsk so special to have such a big uncertainty compared to Falkenberg, in addition to “Obninsk, where ERA5 overestimates high wind speeds and underestimates those below the mean” (line 230)
Line 259: “The differences between the wind speed distributions of the individual climate models are exemplified for the site Hegyhatsal (Fig. 5b). “ Why is this the case for this site, but not for other(s)?
Line 342: “sensitivity is particularly evident at Obninsk (Fig. 10), where CanESM5 exhibits a stronger decrease in wind speeds compared to the remaining ensemble members, leading to a reduction in ensemble spread of approximately 0.4 ms−1 by the end of the century when the outlying model is excluded. A similar behavior is observed at Hamburg during the historical and early future periods, where one model deviates from the remaining ensemble members.” Any idea why? It would be nice to include some ideas/speculations explaining this.