EMS Annual Meeting Abstracts
Vol. 20, EMS2023-283, 2023, updated on 06 Jul 2023
EMS Annual Meeting 2023
© Author(s) 2023. This work is distributed under
the Creative Commons Attribution 4.0 License.

Wind speed maps for Austria: An artificial-intelligence approach

Anna-Maria Tilg1, Irene Schicker1, Annemarie Lexer1, Konrad Andre1, and Martina Heidenhofer2
Anna-Maria Tilg et al.
  • 1GeoSphere Austria, Wien, Austria (anna-maria.tilg@geosphere.at)
  • 24ward Energy Research GmbH, Graz, Austria

The crucial role of wind power in the future energy generation demands to investigate the influence of climate change on wind speed and the associated consequences for wind power. Especially in areas with complex terrain, there are still open research needs for wind energy in future climate (Clifton et al., 2022). To assess future wind climate, one requirement is the availability of a climatological dataset of wind speed with high spatial and temporal resolution to downscale wind speed from global and regional climate models (GCMs, RCMs). So far, no such dataset has been available for Austria. To fill this gap, a wind speed atlas for the past 30 years (1991 – 2020) is compiled for the diverse landscape of Austria, having flat and mountainous terrain, using a set of artificial intelligence (AI) methods.

A comparison of gridded wind speed data considering three baseline methodologies will be presented: The ‘Integrated Nowcasting through Comprehensive Analysis’ (INCA) model combining ERA5 and station observations, a generalized additive regression model (GAM) and a deep neural network (DNN) model with adapted loss function. All approaches are used to create a wind speed analysis for the years 1991 to 2020 with a spatial resolution of 1 km to 1 km for the territory of Austria. Furthermore, they consider the same station-observation dataset of 10-m wind speed measurements of the national weather service in Austria. Preliminary results from the validation against wind speed observations not used in the observation dataset will be shown as well. First results of the GAM and DNN baselines are promising. The uncertainty in interpolation given through the methodologies is, so far, within the expected range.

The availability of a wind atlas for Austria allows downscaling of climate projections and thereby the investigation of the climate change impact on future wind speeds and future wind power potential in Austria.


Clifton, A., Barber, S., Stökl, A., Frank, H., and Karlsson, T.: Research challenges and needs for the deployment of wind energy in hilly and mountainous regions, Wind Energ. Sci., 7, 2231–2254, https://doi.org/10.5194/wes-7-2231-2022, 2022.

How to cite: Tilg, A.-M., Schicker, I., Lexer, A., Andre, K., and Heidenhofer, M.: Wind speed maps for Austria: An artificial-intelligence approach, EMS Annual Meeting 2023, Bratislava, Slovakia, 4–8 Sep 2023, EMS2023-283, https://doi.org/10.5194/ems2023-283, 2023.