The aim of this project is to improve the current operational quality control of the ASCAT level 2 wind product, notably under rainy conditions. A new image processing technique will be tested to complement the current QC algorithm developments in the OSI SAF CDOP2, i.e., the inversion residual or MLE-based QC. The new technique, the so-called Singularity Analysis, refers to any technique capable of evaluating the local singularity exponents of a given function around each one of its points. The concept of singularity exponent extends that of differentiability to a continuous range of cases, across which the regular character of the function can steadily vary. Singularity exponents also allow characterizing non-regular behaviours such as discontinuities and even actual divergences of the function to infinity. When analyzing ASCAT operational wind maps, singularity fronts are induced (Turiel et al., 2012). This is mainly due to the fact that singularity analysis is applied to bi-dimensional maps of a given variable which is submitted to a process taking place in three dimensions. As such, convergence and divergence areas associated with circulation cell boundaries will show up as singularity fronts in ASCAT-derived maps, because they represent actual separation between two flow regimes as observed by the satellite. However, other effects not related to wind circulation induce spurious singularity fronts. For instance, recent studies show that errors in the ASCAT retrieved wind speed and/or direction lead to marked singularity fronts. Moreover, the presence of heavy rain induces clear spurious singularity fronts in ASCAT wind maps. Although separating rain-induced singularity fronts from wind-induced ones is challenging, preliminary results show the technique’s potential to assess the quality of the scatterometer retrieved wind fields.
To contribute to the current ASCAT operational QC, further SA developments are required. We propose to focus on analysing the relation between singularity fronts, for the ASCAT wind vector and each wind component (i.e., U, V, speed and direction) separately, and all geophysical phenomena which affect the radar backscatter signal, including rain, local wind variability, confused sea state, etc. Besides the retrieved wind, other ASCAT-derived parameters, such as the backscatter measurements and the inversion residuals (MLE), will be used to generate singularity maps. These may reveal further characteristics of the ASCAT data in convective areas, which can lead to improving both ASCAT wind retrievals and QC. Numerical Weather Prediction (NWP) model output, satellite derived rain data (e.g., the Tropical Rainfall Measuring Mission’s (TRMM) Microwave Imager or TMI, and the Meteosat Second Generation or MSG), as well as in-situ (moored buoys) rain and wind data will be used to assess the rain impact on ASCAT winds and the SA effectiveness.
Both the MLE QC-based and the singularity analysis methods are expected to be more effective when applied on higher-resolution ASCAT products, i.e., 12.5-km and coastal products (see OSI SAF product at http://www.knmi.nl/scatterometer/ ). On the one hand, ASCAT is expected to better resolve higher resolution wind phenomena (e.g., convergence and downbursts); on the other hand, the rain splashing signal, being patchy and intermittent, is expected to become more evident at smaller ASCAT footprints. As such, we will extend this study to the ASCAT full resolution products.