QuikSCAT normalized radar cross section noise characterization for coastal wind field retrieval

An assessment of the noise affecting the QuikSCAT Normalized Radar Cross Sections (σ0 s) is carried out in this study. The estimation of Kp (K̂p ) is compared to the median of the Kp values (K̃p ) provided in the Level 1B Full Resolution (L1B) file with orbit number 40651, dated 10th of April 2007, and the main differences are discussed. A sensitivity analysis aiming at assessing the presence of any dependencies with respect to (w.r.t.) different wind regimes, the kind of scattering surface, the scatterometer view and the polarization of the signal is carried out. In addition, the presence of any biases is assessed and discussed. Finally, a theoretical σ0 distribution model is proposed and validated against the true measurements. The main outcomes of this study demonstrate that H-Pol measurements are noisier than those V-Pol and that the noise lowers with increasing σ0 levels, in line with the expectations. Furthermore, K̂p may be significantly larger than K̃p , especially for the H-Pol peripheral slices w.r.t. to the footprint (egg) centroid. In addition, the K̂p values estimated over the sea surface are lower than those estimated without making any distinctions among the scattering surfaces. This trend is not seen for K̃p , for which the differences are almost absent. In addition, there are no remarkable differences between the fore and aft views, even if the outermost H-Pol fore acquisitions seem to be slightly noisier than those aft. Furthermore, some inter slice biases up to 0.8 dB are present for H-Pol acquisitions while they are up to 0.3 dB for V-Pol ones, in both cases increasing with the relative distance between the slices, in line with the general Geophysical Model Function (GMF) sensitivity as a function of incidence angle. These biases have a non flat trend w.r.t. the acquisition azimuth angle for both polarizations. These small variations may be due to the changes in wind speed and direction sample for each bin. The theoretical σ0 model proves to be effective. It can be used both for simulation studies and for checking the accuracy of the σ0 noise.