The Walloon region is 16842 square kilometers large and most of it belongs to the Meuse and Scheldt basins. The regional hydrological service (MET/DGVH) operates an automatic rain gauge network including about 90 stations, which corresponds to a density of 1 gauge per 187 square kilometers. Radar observations are available from a C-band Doppler radar located in the south of the region and operated by the Royal Meteorological Institute of Belgium (RMI). The precipitation data ingested in the operational hydrological forecasting system are currently gauge observations onldy but a combined use of radar and gauge observations is planned in a near future. The use of radar data in real-time flow forecasting is presented in Leclercq et al. (2008). In the present study, we evaluate several techniques for merging radar and gauge observations with various degrees of complexity. The merging techniques which have been implemented include mean bias correction, static local bias correction aimed at correcting for visibility effects (beam blocking), range dependant adjustment, Brandes spatial adjustment and sophisticated merging based on geostatistical techniques. The verification of the precipitation field resulting from the merging will be based on the comparison with the 24haccumulation gauge observations from the climatological network operated by RMI. This network includes 150 stations in the area of interest. Most of them are manual stations. Verification results will be presented for one year with the aim of selecting the most appropriate technique for operational use in the hydrological forecasting system.