From measurement to model: ERS-1 scatterometer data assimilation

Discusses the requirements to be fulfilled before observations can be assimilated in numerical models. More specifically, the authors discuss the most important characteristics of ERS-1 scatterometer data and its importance for Numerical Weather Prediction (NWP). They found that measurement noise is extremely low, and that uncertainties in the use of the data are merely a consequence of the interpretation of the measurements in terms of geophysical quantities on the spatial and timely scale of the NWP model. They show that the measurements of radar backscatter, σ⁰, can in first order be interpreted as observations of the surface wind vector. Collocation statistics of scatterometer retrieved winds show that these are more accurate than operationally available conventional surface wind observations. The low measurement noise makes it more practicable to use retrieved winds rather than σ⁰´s as input for variational NWP data assimilation schemes