A scatterometer is microwave radar used to measure the backscatter signal over the earth’s surface. Over the ocean, the radar backscatter signal is dominated by the scattering from wind-generated capillary-gravity waves, which are generally in equilibrium with the sea surface wind. Therefore, the near-surface wind vectors can be determined using a geophysical model function (GMF) that relates wind vectors and scatterometer backscatters.
Ocean surface winds from the global numerical weather prediction (NWP) models are rather uniform in space and time. That is, small scale processes of mixing and convection are not well represented in NWP models. In contrast, the spaceborne scatterometers are able to resolve the increased wind variability near convection areas. This unique capability is essential for climate applications, since wind variability directly impacts air-sea fluxes and, as such, air-sea interaction. However, scatterometer-derived wind quality is known to be degraded by rain contamination effects (notably for Ku-band systems) and increased sub-cell wind variability. Quality control (QC) is therefore of outmost importance since it directly contributes to sampling errors, i.e., rejection of poor-quality winds while preserving as much wind variability information as possible is required. In this study, an improved QC procedure for the C-band Advanced Scatterometer (ASCAT) onboard the Metop satellite series is proposed. Moreover, the differences between in situ, ASCAT and modelled winds near convection areas are analyzed since they are of prime importance for understanding air-sea interaction. This results in triple collocation data sets that are analyzed.
At scatterometer scales, ECMWF wind errors are high and buoy errors considerably increase under large subcell wind variability, ASCAT errors are generally smaller than the other two wind sources. A method to convert 10-min buoy wind data to 25-km equivalent winds is proposed and used for QC verification. Temporally-averaged buoy winds do clearly better represent 25-km spatially-averaged winds. Since the time averaging reduces both the rejected and accepted variances in similar amounts, wind variances appear scalable and wind errors appear mainly due to enhanced wind variability for the rejected cases, i.e., representativeness errors dominate the differences between ASCAT and buoy winds.
In conclusion, analyses of air-sea interaction buoy, scatterometer and NWP data show large effects of differences in representation of physical and dynamical processes across the oceans, with both random and systematic effects on the air-sea interaction. These processes include mixing, turbulence and convection and appear over large ocean areas, thus affecting many ocean applications.
