Assimilation of radar-based nowcast into a HIRLAM NWP model

The present study introduces a nowcast scheme that assimilates radar extrapolation data (RED) into a nowcasting version of the high resolution limited area model (HIRLAM) numerical weather prediction (NWP) model covering the area of Denmark. The RED are based on the Co-TREC (tracking radar echoes by correlation) methodology and are generated from cleaned radar mosaics from the Danish weather radar network. The assimilation technique is a newly developed method that increases model precipitation by increasing low-level convergence and decreasing convergence aloft in order to increase the vertical velocity. A selection criterion based on spatial coverage and temporal duration is applied to identify rain events for evaluation. Eight events from August 2010 comply with the criteria and additionally, an extreme event on 2 July 2011 is also evaluated. The RED assimilated NWP model is verified against a non-assimilated NWP reference run using the fractional skill score. In all cases, the predictability of the RED assimilated NWP model has been increased compared with the non-RED assimilated NWP model. After ending the RED assimilation, the positive effects continue until the skill converges towards the reference run. The level of improved predictability relies on the RED quality, which again relies on the type of event.