Towards the assimilation of radar reflectivities: improving the observation operator by applying beam blockage information

The assimilation of weather radar measurements into numerical weather prediction (NWP) models benefits from advanced quality information. Within the framework of the NORDMET Activity on Radar Applications (NORA) a beam propagation model (BPM) has been applied to correct topographical beam blockages, to assess potential radar locations, and to improve the radar observation operator of the ALADIN (Aire Limitée Adaptation dynamique Développement International) forecast model.

The BPM simulates the radar's field of view considering the scan geometry, the topography, and the atmospheric conditions. Topography is described by a digital terrain model with 1 km horizontal resolution. Atmospheric profiles are either based on standard conditions, radiosoundings or NWP forecasts. BPM output fields are (i) the degree of beam blockage, (ii) the corresponding correction factor which can be applied to operational radar reflectivity products, and (iii) the visibility for a given elevation angle and half-power beam width. Studies for radars of the BALTEX Radar Network (BALTRAD) reveal that beam blockage correction for standard propagation reduces the gauge-radar scatter and bias in average. An operational implementation of the correction matrices for BALTRAD is planned for this year.

In data assimilation the interpolation of variables from model to radar space is part of the so-called observation operator. Currently, ALADIN assumes a Gaussian-shaped radar beam when it comes to vertical interpolation of model variables. Topographical shielding is not considered. By using BPM's visibility maps for standard propagation the vertical interpolation becomes more realistic where the radar beam is partly blocked. It is expected that the suggested modification of the observation operator will have a positive impact on the ALADIN 3DVar data assimilation system and furthermore will improve short range weather forecast at mesoscale.