Assessing inter-model agreement in convection-permitting simulations of extreme winds for wind energy applications

Abstract. Convection-permitting models (CPMs) have great potential for wind energy applications such as wind energy planning, turbine design loads, and operational safety in a climate change context. In fact, compared to coarse-resolution models, they have an improved representation of atmospheric processes and surface characteristics that directly influence winds at heights relevant for turbines. Evaluating how well CPMs perform at reproducing extreme wind events is crucial for wind energy applications. Inter-model comparisons provide insights into uncertainties and enhance the credibility of CPM-based applications. Here, we use a new framework to examine the agreement among three CPMs from the CORDEX Flagship Pilot Study in simulating extreme wind speeds in central Europe. This framework combines surface-based spatial categorisation with Principal Component Analysis and with the non-asymptotic Simplified Metastatistical Extreme Value (SMEV) method, able to estimate rare return levels, such as the 50-year wind speeds U50 required for wind turbine design, from the short CPM simulations. Our results show large agreement between the models, with the first Principal Component explaining 74.2 % of the total variance and indicating a strong consensus in extreme wind patterns, despite systematic differences in magnitudes. Stronger agreement emerges during the winter, when extreme winds are driven by synoptic conditions, and less concordance during summer, when localised convective events cause most extremes. Our research emphasises the value of using CPM ensembles over single-model assessments of extreme winds, and provides the wind energy community with baseline information on CPM capabilities and limitations in estimating wind speed extremes.