Towards an objective performance measure of spaceborne wind sensors: ERS test case

In order to look for a generic scatterometer inversion method, the ERS inversion is revisited in this paper. First, the use of measurement noise information, which includes a realistic geophysical noise term, is tested. It turns out that the shape of the solution surface is more important than the noise information for inversion. Such shape has to be symmetric or circular as a function of wind direction in order to produce realistic and accurate wind direction distributions. A method, which looks for the measurement space transformation that produces such symmetric properties is then presented. In contrast with some previous work performed along these lines, the proposed method is generic. The results show indeed improved wind direction skill compared to the more traditional (measurement-noise normalization) inversion for ERS.