Research highlight
Modeling Warm and Mixed Phase Cloud Microphysics
Cloud microphysical processes refers to the processes involved in the formation and evolution of particles in clouds, for example, droplet condensation and evaporation, collision-coalescence among cloud particles, ice nucleation, and ice growth. They are crucial to prediction of cloud cover and precipitation.
Interactions of cloud, aerosol, and dynamics (turbulence) remain the largest source of uncertainty in the atmospheric models. I use high-resolution cloud models to study the impact of microphysics (aerosol, ice, drops) and dynamics (turbulence) on the formation of clouds and precipitation. In-situ measurements and laboratory experiments are incorporated to constrain the initial conditions of my simulations. |
Direct Numerical SimulationThe direct numeical simulation, or DNS, is a powerful computational fluid dynamics model. The model resolves every turbulent eddies within the model domain by numerically solving the Navier-Stokes equations. Therefore, information of the flow at any given time and at any given locations can be retrieved.
The particle-by-particle DNS, which applies the Lagrangian particle method to trace individual particles in turbulence, is an ideal tool for studying cloud microphysical processes. See our papers (2021, 2020, 2018b, 2018a, 2016) on its application in warm-phase clouds. We recently modified the model to include ice particles, which helps to investigate crucial processes in mixed-phase clouds. See our ACP paper Chen et al. (2023) for more details. |
Lagrangian Particle MethodThe Lagrangian particle method is arguably more accurate than the traditional Eulerian method in calculating the cloud microphysics.
Comparing to the Eulerian method (bin microphysics and bulk microphysics) where evolution equation of the droplet size distribution is solved in each grid box, the Lagrangian method follows cloud particles. In the DNS model, we are able to implement particle-by-particle tracking techniques and calculate the growth history of individual particles coupled with turbulence (e.g., see Chen et al. 2018b). In the large-eddy-simulation (LES), to reduce the computation, Lagrangian particles are treated as computational particles, also referred to as super-droplets. Each super-particle represents a group of particles with the same properties. |
Parcel Model |
Bridging the gap between modeling, observations, and experimentation |
A parcel model is a 1D model following the Lagrangian trajectory of an air parcel. It calculates the bulk tendencies of the variables (temperature, humidity, droplet spectra, etc.) inside an air parcel. The model is computationally cheaper than most other cloud models due to its simplicity and can be used for investigating DSD evolution affected by certain processes in an idealized cloudy environment.
It can be combined with a DNS parcel (i.e., a parcel-DNS hybrid framework) and used when effects of local fluctuations and spatial heterogeneity have a minimal impact on the simulated results to reduce computations, see Chen et al. 2020. |
Observations from flight measurements and Laboratory experiments are important to constrain the modeling and to verify the model output (Shaw et al. 2020).
In my research, flight measurements are frequently used for setting up the initial conditions of the simulations. (e.g., Chen et al. 2016 and Chen et al. 2018a In the meantime, the cloud chamber provide steady-state environments with controlled boundary conditions to conduct repeatable experiments which are impossible in field campaigns. Therefore, laboratory studies are ideal for comparing to theory and simulations (see our workshop summary in BAMS) |