Simulating cold pools with ICON during the FESSTVaL period

<p>During the Field Experiment on Submesoscale Spatio-Temporal Variability in Lindenberg (FESSTVaL) just under forty distinct cold pool cases were observed within the area covered by the observational network (see contribution by B. Kirsch on observed cold pool morphology in UP1.5). Here, we investigate to what degree the Icosahedral Nonhydrostatic (ICON) model is able to reproduce the observed cold pool properties.<br>ICON is run on a limited area domain centered around the observational facilities used during FESSTVaL and covers most of eastern Germany and western Poland. Similar to the operational &#8220;D2&#8221; setup, the horizontal resolution is approximately 2km such that deep convective transport is largely resolved, while shallow convection remains a subgrid-scale, parameterized process. The model is tested with two options for the shallow convection scheme (Tiedtke-Bechtold with/without stochastic perturbations) and an optional cold pool perturbation (CPP) scheme aimed at improving convection triggering along cold pool boundaries.<br>The cold pools produced by the model share many qualitative similarities with the observations: ICON successfully produces cold pools on days when they are observed. As in the observations, the growth of the cold pool area appears to be directly associated with precipitation: When precipitation ceases, the cold pool area stops growing. The intensity of the cold pool (measured as the 2m temperature decrease) however tends to reach its peak before precipitation and area growth cease. It appears that the model may lose the cold pool signature rather more quickly than is observed.<br>As per design, the CPP scheme enhances the vertical velocity along cold pool boundaries, which leads to more intense precipitation associated with the cold pool forming convection as well as enhanced cold pool intensity.</p>