The data assimilation method Latent Heat Nudging assessed with the Dynamic State Index

In April 2007, the German Weather Service (DWD) added the non-hydrostatic limited area model COSMO-DE to its model chain. COSMO-DE covers mainly Germany and bordering countries and has a horizontal resolution of 2.8 km (0.025°). An advantage of the COSMO-DE is that deep convection is expected to be resolved explicitly. In order to improve the initial state of COSMO-DE an assimilation of radar derived precipitation rates is applied by using "'Latent Heat Nudging'" (LHN). The aim of LHN is to adjust the model state so that the model will respond by producing a rain rate close to the observed value. In this study, the influence of LHN on diabatic processes is investigated by evaluation of the Dynamic State Index (DSI). The DSI is calculated on both isentropic (DSI θ ) and model levels (DSI θ ). To analyse the influence of LHN, model analyses with and without LHN are examined for summer 2009 (June to August). The influence of LHN is also examined in more detail in a case study of a heavy rainfall event on July 18, 2009 where a strong rainfall area was developed on the front side of a trough. Results, based on a statistical investigation and the case study, show that the model with LHN can reproduce the observed rainfall better than the runs without LHN. This is also suggested by a higher correlation between DSI and observed precipitation. The release of latent heat enhances the diabatic processes associated with the formation of additional potential vorticity anomalies resulting in stronger DSI-signals.