The verification and application of satellite scatterometer winds

Abstract

Scatterometers are active microwave radars that provide surface wind vector observations over the global oceans. The data are especially useful in the studies of oceanic synoptic scale and meso-scale cyclones/storms, which can not be adequately described by sparse conventional observations. This dissertation tackles a fundamental question: what is the accuracy of the scatterometer winds? It is a straightforward question for moderate wind speeds: the scatterometer winds can be simply compared with buoy wind observations. However, an equally direct answer is impossible for high wind speeds because conventional wind data are known to be inaccurate at high winds mainly due to high sea states. A unique approach that relates surface wind to sea level pressure is applied to alleviate this problem. It is demonstrated that the scatterometer winds exhibit a significant low bias of around 2 ms$\sp{-1}$ at speeds greater than 18 ms$\sp{-1}$. An empirical correction function is proposed to correct for the speed bias.The dissertation then focuses on the applications of scatterometer winds to weather analysis, especially in the vicinity of storms. An innovative method that derives central pressures of oceanic storms from scatterometer winds is devised. The central pressures are compared with the analyses of the state-of-the-art operational weather prediction model. They generally agree well in the northern hemisphere, but significant difference exists in the southern hemisphere. A second application involves data assimilation, particularly why the impact of scatterometer wind on numerical weather prediction is mostly neutral. Displacement error in the model forecast fields is identified as one of the major reasons. A simple variational approach that take into account displacement errors is developed and tested in a controlled environment. Favorable results are produced by this approach.