Assimilation of ERS-2 scatterometer winds using the Canadian 3D-var

The goal of this study is to evaluate the impact of incorporating the marine surface winds retrieved from the ERS-2 scatterometer in the Canadian three-dimensional variational analysis system, (3D-var). The aspects of the 3D-var most relevant to the assimilation of surface wind observations and a general method for resolving the directional ambiguity of the retrieved scatterometer winds are first described. A comparison with 6-h forecasted winds is then made to demonstrate that these data are of high quality, but exhibit a speed bias that can be removed by increasing their amplitudes by about 5%. The analysis increment from a single scatterometer wind observation is calculated to illustrate the response of the 3D-var to surface wind observations. As a consequence of the forecast error covariance model, the assimilation of surface wind observations produces meteorologically consistent increments for both the rotational and divergent wind components and the mass field. The results from a series of cross-validation experiments using ship-based wind data demonstrate a positive impact of assimilating scatterometer winds and the effectiveness of a simple method for estimating and removing the speed bias. The impact of assimilating scatterometer data within a short assimilation cycle is also evaluated. Overall, the results show that including scatterometer data in the analysis decreases the 6-h forecast error of surface wind by 13%. Over the northern extra-tropics the improvement is only 4% and for the southern extra-tropics it is 16%. Results from a series of two-day forecasts produced using the analyses from the assimilation cycles with and without retrieved scatterometer winds included are also presented. Using radiosonde observations at 850 hPa, 500 hPa, 250 hPa and 100 hPa for verification, the impact on the forecasts is nearly neutral in the northern hemisphere and the tropics. Conversely, a significant positive impact is found on both wind and mass fields in the southern hemisphere over the entire two-day forecast.