Monitoring the impacts of weather radar data quality control at the continental scale

At ECCC, the first impact studies of assimilating weather radar data into a numerical weather prediction (NWP) model used aggressively configured quality controls with data from the Canadian network of 31 radars and all available NEXRAD radars from the continental United States and Alaska. The radar data assimilation approach used with these impact studies was latent heat nudging (LHN). Owing to LHN’s sensitivity to the presence of non-precipitation echoes (false positives), the radar data quality control was carried out with the understanding that removing the desired amount of non-precipitation echoes came at the cost of also removing a certain amount of real precipitation information. While this gave us a starting point for radar data assimilation, we were confident that it could be subsequently improved upon using better data quality control. However, the question facing us was: how do we know when we have an improvement?