The Impact of the Ice Phase on Orographic Mixed-phase Clouds and Surface Precipitation in the Swiss Alps

Abstract

In mixed-phase clouds (MPCs), supercooled liquid water and ice particles can co-exist at subzero temperatures warmer than ~ −38 °C. Therefore, MPCs are thermodynamically unstable enabling the rapid growth of ice crystals in conditions where liquid drops evaporate. Consequently, precipitation formation in MPCs occurs faster than in liquid water clouds. The ice phase contributes up to 63% to surface precipitation globally and up to 50% in the continental mid-latitudes. At temperatures warmer than ~ −38 °C, primary ice production occurs through heterogeneous nucleation, which involves ice nucleating particles (INPs). However, aerial and ground-based observations show that the ice crystal number concentrations (ICNCs) in winter and summertime orographic MPCs are typically 1 to 3 orders of magnitude larger than the concentration of INPs. It was hypothesized that secondary ice production, specifically ice multiplication through rime splintering, could be the reason for this discrepancy. However, theoretical calculations and numerical simulations of rime splintering could not account for the high observed ICNCs. Also, the rapid glaciation of observed MPCs is suggested to occur faster than what can be achieved in laboratory experiments that considered rime splintering. The evidence is clear that other processes than the presence of INPs enhance ice formation in MPCs. Representing the ice phase correctly in models is crucial for simulating the lifetime of MPCs and precipitation formation accurately. Therefore, in this thesis, I explored processes involving secondary ice production (SIP) and the seeder-feeder process that could enhance ICNCs using the regional climate model COSMO. Secondary ice production occurs through (1) the breakup of ice particles during ice-graupel collisions and (2) frozen droplet shattering upon freezing. To address the discrepancy, we conducted COSMO simulations based on data collected during the RACLETS campaign, in the Swiss Alps in February and March 2019. We hypothesized that the ICNCs could be enhanced through collisional breakup and droplet shattering which could explain the INPs vs ICNCs discrepancy. On 7 March 2019 near Davos, we simulated the passage of a cold front to assess the role of secondary ice production on wintertime MPCs. Surface measurements of ICNCs, using the HoloGondel platform, were between 9±3 and 19±4 per liter during the afternoon. Simulations of droplet shattering underestimated the surface ICNCs by at least 2 orders of magnitude because the cloud base temperatures were colder than what is needed for optimal conditions for droplet shattering. However, the simulated ICNCs as a result of collisional breakup were generally in agreement with the HoloGondel observations. The explosive breakup in the mid levels of the cloud caused a drastic decrease in the liquid water fraction favoring depositional growth of ice particles. Consequently, the microphysics predominantly controlled the reduction of precipitation in localized regions of high precipitation compared to the control simulation which did not include any SIP processes. SIP show promise in narrowing the gap between observed and simulated ICNCs. Enhanced ICNCs compared to INP concentrations can also occur when ice particles sediment from an upper seeder cloud into a lower laying feeder cloud (external seeder-feeder) or when ice particles sediment from the cirrus region into warmer regions of the same cloud (internal seeder-feeder). We hypothesize that: (1) the external and internal seeder-feeder process could significantly enhance the ICNCs and (2) when the seeding particles are removed in the simulations, the surface precipitation will decrease significantly. On 18 May 2016 in the Swiss Alps, a high number of multi-layered clouds were observed by the DARDAR-CLOUD satellite. We simulated this case to elucidate the occurrence of seeder-feeder events and their impact on surface precipitation. The external seeder-feeder events occurred 10.3% of the time when multilayered clouds were observed. On average over the domain, ice particles seldom could survive a fall greater than 2.5 km. When they survived 58.3% of the ice particle mass was lost as a result of melting or sublimation. When we removed the seeding particles from the external and internal seeder-feeder processes, the depositional growth rate was reduced, corresponding to a slower glaciation of the clouds. The precipitation was reduced in the internal seeder-feeder process by up to 7.5% compared to the control simulation. The external seeder-feeder process was shown to impact surface precipitation significantly through the enhancement of ICNCs in the feeder cloud, however their infrequent occurrence can most likely not explain the frequently observed discrepancy (from literature) between ICNCs and INPs. However, the internal seeder-feeder processes enhances the ICNCs significantly in the lower cloud resulting in significant enhancements in surface precipitation. Increasing the concentration of INPs can aid in the formation of more ice particles, but this does not address the discrepancy between ICNCs and INPs. However, we hypothesised that enhanced aerosols can be linked with enhanced homogeneous freezing of liquid droplets. Therefore, these homogeneously formed ice crystals can sediment into the cloud’s mixed-phase region, enhancing the ICNCs and consequently surface precipitation. In summer 2017, a flooding event occurred in Switzerland which coincided with a Saharan dust outbreak during which elevated aerosol concentrations were observed ahead of the cold front. We hypothesized that there was a link between the aerosols and the flooding event. A one-moment cloud microphysics scheme, similar to the one used by MeteoSwiss, unsuccessfully predicted the flooding event. We speculated that the two-moment scheme could have predicted the flooding event. We therefore compared an aerosol perturbed two-moment scheme to a control simulation using an unperturbed control two-moment cloud microphysics scheme. The choice of the microphysics scheme had the largest impact on surface precipitation. The simulations perturbed with higher cloud condensation nuclei (CCN) showed a shift towards the lee of the Alpine ridge, while the INP perturbations enhanced precipitation on the windward side. In the INP perturbations, the reduction in the total cloud liquid and increase in total cloud ice only caused a shift in the growth mechanism to depositional growth relating to a shift in precipitation, but not a significant change in the amount of precipitation. Therefore, the enhanced aerosols could not explain the flooding event. In this thesis, we demonstrated the importance of both collisional breakup and the seeder-feeder mechanism in enhancing ICNCs in clouds, resulting in significant changes in growth regimes of ice particles yielding significant changes in surface precipitation. Employing SIP processes in models is a way forward in reducing the discrepancy between observed and modeled ICNCs.