Wind resource assessment from C-band SAR

Abstract

Using accurate inputs of wind speed is crucial in wind resource assessment, as predicted power is proportional to the wind speed cubed. First, wind speeds retrieved from a series of 91 ERS-2 SAR and Envisat ASAR images, at moderate wind speeds (2–15 m s− 1), were validated against in situ measurements from an offshore mast in the North Sea. The wind direction input, necessary for SAR wind speed retrievals, was obtained from the meteorological mast and from a local gradient analysis of wind streaks in the SAR images. A wind speed standard deviation of not, vert, similar 1.1 m s− 1 was found when in situ wind directions were used. The use of local gradient wind directions yielded a standard deviation of not, vert, similar 1.3 m s− 1. Wind speeds retrieved from three geophysical model functions (CMOD-IFR2, CMOD4, and CMOD5) were compared. The best approximation to the in situ measurements of wind speed was found for CMOD-IFR2, despite a bias on the order of − 0.3 m s− 1. CMOD4 retrievals also underestimated the wind speed, whereas the bias on CMOD5 retrievals was negligible. Then, wind resource assessments were made from the SAR-based wind observations to show how errors in wind speed from the different SAR wind retrievals were reflected in the wind statistics. The mean wind speed, obtained for all of the 91 SAR scenes, was linked closely to the bias of SAR wind retrievals. Agreement to ± 15% of the in situ measurements was found for all the wind retrieval methods tested. The accuracy of power density estimates for the entire data set was evaluated by the standard deviation of SAR wind retrievals relative to the in situ measurements. SAR wind fields retrieved with CMOD-IFR2, using in situ wind direction inputs, exactly yielded the power density predicted from in situ measurements alone. The SAR-based wind resource assessment also corresponded well to predictions from longer time series of in situ measurements. This indicates that a reliable wind resource assessment may be achieved from a series of randomly selected SAR images. The findings presented here could be useful in future wind resource assessment based on SAR images.