Validation of satellite SAR offshore wind speed maps to in-situ data, microscale and mesoscale model results

摘要

A validation study has been performed in order to investigate the precision and accuracy of the satellite-derived ERS-2 SAR wind products in offshore regions. The overall project goal is to develop a method for utilizing the satellite wind speed maps foroffshore wind resources, e.g. in future planning of offshore wind farms. The report describes the validation analysis in detail for three sites in Denmark, Italy and Egypt. The site in Norway is analyzed by the Nansen Environmental and Remote SensingCentre (NERSC). Wind speed maps and wind direction maps from Earth Observation data recorded by the ERS-2 SAR satellite have been obtained from the NERSC. For the Danish site the wind speed and wind direction maps have been compared to in-situobservations from a met-mast at Horns Rev in the North Sea located 14 km offshore. The SAR wind speeds have been area-averaged by simple and advanced footprint modelling, ie. the upwind conditions to the meteorological mast are explicitly averaged in theSAR wind speed maps before comparison. The comparison results are very promising with a standard error of ± 0.61 m s-1, a bias around 2 m s-1 and R2 around 0.88 between in-situ wind speed observations and SAR footprint averaged values at 10 m level. Windspeeds predicted by the local scale model LINCOM and the mesoscale model KAMM2 have been compared to the spatial variations in the SAR wind speed maps. The finding is a good correspondence between SAR observations and model results. Near the coast is an800 m wide band in which the SAR wind speed observations have a strong negative bias. The bathymetry of Horns Rev combined with tidal currents give rise to bias in the SAR wind speed maps near areas of shallow, complex bottom topography in some cases. Atotal of 16 cases were analyzed for Horns Rev. For Maddalena in Italy five cases were analyzed. At the Italian site the SAR wind speed maps were compared to WAsP and KAMM2 model results. The WAsP model captured the local wind speeds very well especiallynear the coast and up to around 5 km offshore. Further offshore the KAMM2 model results seemed more reliable than the WAsP model. This is likely due to the effect of high orography of the island Corsica located North of the study area. The mountains wereincluded in the KAMM2 model domain but not in the WAsP model domain. The mountains had a significant impact on the wind field far offshore. In the Gulf of Suez the winds are very strong but there exists large spatial wind speed gradients and this makesthe site challenging for SAR wind speed validation studies. Only three cases were analyzed for the Gulf of Suez in Egypt. A study on how many wind speed maps would be needed for wind resource estimation showed that around 60-70 randomly selectedsatellite images are required to characterize the mean wind speed and Weibull c parameter, while of the order of 150 images are required to obtain a variance estimate, and nearly 2000 are needed to obtain a robust estimate of energy density (or Weibullk). This is under the assumption of no error in the SAR wind speed maps and for an uncertainty of ± 10% at a confidence level of 90%. Around 100 satellite SAR scenes may be available for some sites on Earth but far few at other sites. Currently the numberof available satellite SAR scenes is increasing rapidly with ERS-2, RADARSAT-1 and ENVISAT in orbit. Hence the technique holds promise for future utilization in offshore wind resource assessment.