Wind bias from sub-optimal estimation due to geophysical modeling error

Inherent in all empirical estimates of the geophysical model function (which relates the wind to the normalized radar cross section, NRCS, of the ocean surface) is uncertainty caused by parameters not included in the empirical model. This can be thought of as variability in the NRCS for given wind conditions. When the estimated variability is included in the maximum likelihood wind retrieval algorithm, complicated interdependencies arise between the estimated wind and the estimated variability. Techniques for simplifying these interdependencies results in sub-optimal wind estimates. The model function variability can be ignored, or the log-variance term in the maximum likelihood estimation could be eliminated. The authors consider the impact of these suggestions on NSCAT compass simulations. As expected, each method results in biased wind estimates. However, the biases are small, particularly when compared to the added complexity of the full solution