Data Assimilation and Problems in the Quality Control of Doppler Winds

The use of radar winds in the numerical analysis is studied by starting from the idea that the radar winds are most useful in mesoscale systems, which often are convective. In such cases the assumption of a linear wind field does not hold. The focus is in a continental wide data exchange of radar winds needed for limited area models. Parameter values produced by the Fourier analysis as applied in the VAD method seem, in comparison to a direct use of the radial winds, to be optimal or nearly optimal for the application in the numerical analysis. The advantages of the VAD approach are as follows. The VAD method tends to smooth the data similarly as is done in the grid. Especially the subgrid scale is smoothed out. The method makes full use of the data, which is efficiently compressed. So one data record is enough for dissemination to cover requirements of different model applications. The wind components and deformation terms produced are independent from each other and therefore present an optimal data set for the numerical analysis. The divergence is applicable only at lower levels.