Interpretation of radar information is usually obtained through geophysical inversion of radar measurement, such as Z-R relationship or VAD technique. An alternative approach to interprete and quantitatively exploit radar data is data assimilation. It enables direct assimilation of remote sensing data. In essence, the observation is modelled and the model counterpart of the observation is calculated from the NWP model variables. Each remote sensing observation type requires specific modelling.

This paper presents a model, also called an observation operator, for Doppler radar radial winds. The observation operator has been developed and implemented to the High Resolution Limited Area Model HIRLAM. It consists of (1) interpolation of the model horizontal wind components to the observation location, and (2) their projection on the slanted line of the radar beam. Further refinements include modelling of (3) the vertical beam broadening by a range dependent Gaussian weight function, (4) the obscuring effect of the radar horizon and (5) the upper integration limit for the scattering distribution. Implemented but not yet fully tested refinement is to take into account (6) the bending of the radar beam by calculating the local refraction index from the model (T, q, p) -profile and to modify the observation height and the elevation angle accordingly.

Assimilation experiments revealed a range dependent bias where the model counterpart is stronger than the observed wind the longer is the range. This can be reduced by refinement (3). Refinements (4) and (5) have only a minor effect on bias but make the weight function more realistic. Preliminary results show that accounting for the bending of the radar beam (6) reduces the detected bias minutely.

Impact studies indicate that radar radial wind information has a positive impact on forecast quality of HIRLAM model. This is especially true for the winds and temperatures (through the multivariate background error constraint formulation of the HIRLAM data assimilation system) in the low and middle troposphere.