On the quality of high-resolution scatterometer winds

High-resolution wind products based on space-borne scatterometer measurements by ASCAT and SeaWinds are used widely for various purposes. In this paper the quality of such products is assessed in terms of accuracy and resolution, using spectral analysis and triple collocation with buoy measurements and NWP model forecasts. An experimental ASCAT coastal product is shown to have a spectral behavior close to k(-5/3) for scales around 100 km, as expected from theory and airborne measurements. The NWP spectra fall off more rapidly than the scatterometer wind spectra starting at scales of about 1000 km. Triple collocation is performed for four collocated data sets, each with a different scatterometer wind product: ASCAT at 12.5 km and 25 km, and SeaWinds at 25 km processed in two different ways. The spectral difference between scatterometer wind and model forecast is integrated to obtain the representation error which originates from the fact that global weather models miss small-scale details observed by the scatterometers and the buoys. The estimated errors in buoy winds and model winds are consistent over the data sets for the meriodional wind component; for the zonal wind component consistency is less, but still acceptable. Generally, enhanced detail in the scatterometer winds, as indicated at high spatial frequencies by a spectral tail close to k(-5/3), results in better agreement with buoys and worse agreement with NWP predictions. The accuracy of the scatterometer winds is about 1 ms(-1) or better. The calibration coefficients from triple collocation indicate that on average the ASCAT winds are slightly underestimated.