Selective simulated seeding on hailstorms – a summertime case study over Switzerland

Abstract Ice Nucleating Particles (INPs) are known to affect the microphysical properties of hailstorms. However, their role on surface precipitation amount and hail size distribution is still unclear. The reason is that hailstorms are rare events that consist of many non–linear interacting processes. In this study, we are analyzing the effect of INPs perturbations on hailstorms observed on 6th of July 2019 over the Swiss plateau in terms of precipitation and hail size distribution. The simulations are performed by the regional weather and climate model COSMO(Consortium for Small-Scale Modeling) (Steppeler et al., 2003; Baldauf et al., 2011) on a rotated latitude-longitude grid with 0.01° horizontal resolution (which corresponds to approximately 1.1km grid spacing), 80 hybrid vertical levels with the upper limit being at approximately 23 km and  6 s temporal resolution. One prerequisite of our experiment is to create a thunderstorm that is in agreement with one of the observed thunderstorms, using initial and boundary conditions of the given day. We define this agreement when the following two conditions are met: i) the simulated cell is within 40 km and 3 hours of the observed cell and ii) when the hail diameter of the simulated cell is larger than 1 cm at the cloud base in one or more grid points. When this is achieved, we seed these hailstorms the time period that they are at the cumulus until the early mature stage (onset of precipitation). The seeding is conducted in the updraft close to the cloud base. The additional INPs from the seeding are treated as a prognostic variable following the studies by Possner et al. (2017) and Eirund et al. (2018). In order to go back in time and locate the wider area in which the hailstorm was at these stages, we use the mean updraft 700-500 hPa and precipitation rate. The mean updraft indicates the area where there is upward motion until 3 m/s while the precipitation rate assures the existence of a deep convective cloud which doesn’t precipitate heavily -- higher than 5 mm/hr.  The whole process is repeated 10 times following the time-lagged ensemble approach (Vogel et al. 2013), in order to analyze the effect of seeding and its significance in the context of model variability.