A baseline approach for downscaling the Euro-CORDEX data for wind risk assessment of the Metsovo village in Greece

A methodology is presented for downscaling the wind projections of Euro-CORDEX in order to derive temporally and spatially correlated region-wide wind fields that can be used for assessing the wind risk for cultural heritage sites. The coarse spatial and temporal resolutions of the Euro-CORDEX projections prohibit their use as a direct input for such purposes, especially for cultural heritage assets that are spatially distributed within the Euro-CORDEX grid and dynamically respond differently to wind. To improve the temporal resolution of the Euro-CORDEX data, we leverage machine learning tools and weather station measurements, aiming to generate composite “Frankenstein” days at the locations of the weather stations that comprise 144 jigsaw pieces of actually measured 10min wind time-series that are matched together to form a continuous daily record. The “Frankenstein” days are expanded spatially to all locations where critical assets can be found by employing spatially distributed wind fields that are computed via high-fidelity computational fluid dynamics simulations and provide contemporaneous wind values at all locations of interest. This process allows generating “Frankenstein” days and wind fields with a temporal resolution of 10min and spatial resolution that allows assessing the wind risk for spatially distributed assets. As a case study, the Euro-CORDEX wind projections are downscaled for the cultural heritage village of Metsovo that is found at the Western part of Greece. Most of the buildings within this village are made of stone masonry and tiled roofs and are vulnerable to extreme wind actions as wind can cause damages e.g., on the tiled roofs thus making the buildings vulnerable to rainfall, or even lead to their partial or complete failure. Thus, the Frankenstein days and wind fields are employed for assessing the wind risk for the cultural heritage buildings of Metsovo both on an event-basis and in the long-term.