It is clear that substantial uncertainties exist regarding the evolution of ice in atmospheric cold clouds of all types. The uncertainties regarding ice processes are broad, and cover almost every aspect of cold-cloud development and evolution: Ice formation, growth from the vapor, sedimentation, riming (ice collecting liquid), aggregation (ice collecting ice) and even radiative scattering and absorption all depend substantially on the evolution of particle habits in cold clouds, and all are uncertain to various extents. These uncertainties have demonstrably important consequences for simulations, and remotely sensed properties, of cold clouds of all types and in all geographic regions. Ice formation processes have posed a substantial challenge, but continuing progress is being made. After formation, crystals develop a range of shapes through growth from the vapor, but the growth rates are strongly affected by how those shapes evolve. Making matters more difficult, vapor growth is enhanced by ventilation during crystal sedimentation meaning that particle fall speeds must be accurately tied to the evolution of crystal shape. The evolution of shape and fall-speed are critical for estimating phase in mixed-phase clouds and the lifetime of cirrus. Even if particle habit evolution can be predicted, our equation for vapor growth breaks down at low saturations like those that occur in cirrus or the Arctic. Alas, no general method exists in the current literature for “correcting” the growth equations at low saturations for numerical models. Finally, riming (ice collecting drops) and aggregation (ice collecting ice) are other uncertain processes by which ice particles grow and modify the mass and thermal budget of a cold-cloud layer.

All of the aforementioned growth processes are inextricably tied together through the evolution of ice particle shape, and capturing these connected processes in a numerical model is challenging. Many prior numerical models use simplified approximations for the evolution of particle shape, often with somewhat arbitrarily chosen boundaries between different types of ice crystals. These somewhat artificial boundaries and constraints can have undesirable impacts on the numerical simulations of cold clouds. In an effort to avoid such problems, a new method has been developing in numerical modeling that predicts particle properties, such as the fraction of rime (frozen liquid) on ice particles, ice particle shapes, and so forth. Our ice habit prediction method follows this new approach, it is based on the physical mechanisms that control ice shape allowing for a natural evolution of size, shape, and fall-speed. However, our (and all) cloud models presently have no general way to correct for growth at low saturations. Moreover few, if any, cloud models correctly capture the habit-dependent effects associated with riming and aggregation growth, which are critical for mixed-phase and cirrus cloud evolution. The goal of our proposed research is to generalize the ice habit prediction method for the missing processes discussed above: Low ice saturation states, riming and aggregation. We plan to use DOE data products from past and recent field campaigns, for both low and high clouds, along with lab data to test and refine our model.