Global optimization algorithms for field-wise scatterometer wind estimation

Field-wise scatterometer wind estimation determines the vector wind at many resolution elements simultaneously by estimating the parameters of a wind field model. According to simulations, it results in more accurate estimates than traditional point-wise estimation, which estimates the vector wind one resolution element at a time. Further, field-wise estimation produces fewer ambiguities than point-wise estimation. Field-wise estimation necessitates locating the local minima of a high-dimensional, non-linear objective function. Conventional optimization techniques can be employed if initial search points within the capture regions of the local minima can be found. The authors develop and evaluate two novel approaches for determining initial search points for field-wise estimation. The accuracy of each algorithm is evaluated using simulated NASA Scatterometer (NSCAT) data