Assessing the uncertainties of using land-based wind observations for determining extreme open-water winds

For the assessment of the safety of the Dutch flood defences extreme open-water winds need to be computed. There are, however, no sufficiently long and reliable in-situ data available. On the other hand, there is a rich dataset of decades of measurements at certain coastal and relatively close by inland stations. A commonly used two-layer model for neutral atmospheres was thought to provide reasonably accurate open-water winds from the available data, given that the model assumptions seemed plausible for the extreme winds of interest. However, the model results were deemed inaccurate and not usable. Given that this was unexpected, many of the model assumptions were analysed and, with the gained further insight, their validity and contribution to the invalidity of the deemed simple model approach assessed. Our conclusion is that the quality of the model results is significantly affected by at least two aspects: the assumption of neutral stability in the model, and – equally important – the assumption of independence between the surface roughness and the wind speed.

► Spatial ratios of extreme 10 m wind speeds depend on the thermal stability.
► Decreasing trends in extreme wind speed over land are due to changes in surface roughness.
► Translating wind velocities to open-water requires stability and wind speed dependent roughness data.
► The 10 min standard deviation is a better proxy from roughness than the 1 h gust.