Assimilation of Nowcast Objects for Rapid Update Cycling

<p>Within the SINFONY project at DWD, we have developed a framework for assimilating convective objects from the cell-identification and -tracking scheme KONRAD3D into NWP forecasts using ICON-LAM. The main idea behind this approach is to additionally constrain modeled convection -- that is, generating missing convection and/or removing spurious convection in the model forecast -- by considering radar reflectivities as coherent objects with associated features, rather than 3-d fields of pixelated information.&#160; This approach is appealing because it harnesses the object-oriented approach taken in nowcasting in order to bring additional information into the NWP forecast, while also reducing the complex 3d radar scans down to a set of 2d fields.</p> <p>However, it is inherently difficult to evaluate the added value of convection-related observations, because they tend to be effective only in special cases, while long term statistics are still needed to see substantial effects in an assimilation system that is already highly constrained by observations. Here we present a set of experiments that examine what happens when we add different types of object-style observations to a baseline assimilation during the summer 2021 active period. We show the longer-term statistical effect of adding object-type observations, while also illustrating how the state update changes in individal cases.</p>