Impact of using Doppler radar radial wind data in a winter cyclone period

High-resolution observing networks are required to fulfill the needs of the present-day mesoscale numerical weather prediction (NWP) models. In many countries, like Finland and Sweden, the radar network has an excellent geographical coverage and Doppler radars provide radial wind observations with good temporal and spatial resolution.

The analysis method in the High Resolution Limited Area Model (HIRLAM) is 3-dimensional variational assimilation (3D-Var). The HIRLAM 3D-Var includes a detailed observation operator for producing the model counterpart for Doppler radar radial winds. Typically the HIRLAM model is run with horizontal resolutions from 10 to 20 kilometers whereas the horizontal resolution of the Doppler radar radial wind data is approximately one kilometer. Preprosessing of the radar wind observations is required in order to minimize representativeness error when compared to the model with coarser resolution. Radar radial wind data is input to the HIRLAM model as spatial averages, so-called superobservations. Superobservation generation averages out the random errors from the high resolution radial wind observations quite effectively.

A two week winter period is studied to asses the impact of Doppler radar radial wind superobservations on model analyses and forecasts. The period is characterized by deep cyclones passing over the Baltic sea area. HIRLAM model with 9 km horizontal resolution is used in the experiment. Results from earlier assimilation experiment with 22 km grid size indicate that using Doppler radar winds have positive impact on wind and temperature forecasts in the low and middle troposphere.