The influence of ice nucleation mode and ice vapor growth on simulation of arctic mixed-phase clouds

Mixed-phase arctic stratus clouds are the predominant cloud type in the Arctic . Perhaps one of the most intriguing of their features is that they tend to have liquid tops that precipitate ice. Despite the fact that this situation is colloidally unstable, these cloud systems are quite long lived - from a few days to over a couple of weeks. Previous studies have suggested that this longevity may be due to a paucity of ice nucleating aerosols (ice nuclei, or IN) in the Arctic. Such studies have shown that small changes in IN concentrations can cause large changes in the amount of liquid water within a mixed-phase stratus deck. We use the Regional Atmospheric Modeling System (RAMS) to simulate the time period of October 9-11 from the Mixed-Phase Arctic Cloud Experiment (M-PACE) which was conducted in October of 2004.. Using heterogeneous ice nucleation parameterizations typical of most models (e.g. Meyers et al., 1992), the simulated clouds rapidly glaciate, unlike the observed clouds.. Measurements during M-PACE suggest that the IN concentrations predicted by typical parameterizations are an order of magnitude too large. Using data taken during M-PACE, the existing IN parameterizations in RAMS were modified. In this case, extensive decks of liquid clouds with smaller amounts of ice are now predicted, in better accordance with the observations. It is important to note that in the simulations described above IN are depleted due to precipitation of ice crystals. Without depletion even the MPACE-derived IN lead to rapid glaciation and loss of all liquid water. These results suggest that in order to more accurately simulate mixed-phase clouds models must not only correctly predict the number of heterogeneously nucleated ice but also the cloud processing and removal of IN through precipitation. Our simulations are also potentially sensitive to the way contact nucleation and pristine ice vapor growth are parameterized. Hence, we are conducting tests to examine the impact of these parameterizations, and whether they need improvement. At the meeting results for both meso-scale and cloud-resolving model simulations will be presented.