Computer Science Ph.D. Student in the Computational Sciences Group at KAUST Torsten Hädrich talked about generating realistic cloud dynamics.
My name is Torsten Hädrich. I am a Computer Science Ph.D. Student in the Computational Sciences Group at KAUST. In my research, I am interested in the simulation of natural phenomena in the field of computer graphics. This involves for example the physically-based simulation of vegetation, the simulation of fire, or erosion.
The simulation of clouds was interesting for us for several reasons. Firstly, clouds are one of the most common weather phenomena we can observe, and clouds formations, such as thunderstorms, be visually very impressive. Moreover, being able to accurately capture clouds is essential for simulating the climatic conditions of an environment, as clouds are for example responsible for precipitation or influence the ground temperature by blocking sunlight. A physically-based model for clouds can also be used for the simulation of vegetation in complex ecosystems.
Our goal in particular for this project was to develop a physically accurate model that is able to simulate cloud formation for a variety of different cloud types, as well as their evolution over time.
Challenges and Approach
The main challenge was that the simulation of clouds requires modeling a complex system where the interplay of many different variables is responsible for clouds to form. In this system, we integrated, for example, fluid dynamics, which governs the motion of air, and thermodynamic processes which involve for example the condensation of water vapor which leads to cloud formation.
Some of the most essential parameters that influence cloud formation are the temperature and humidity in the air. An increase in temperature causes humid air to rise. Clouds start to form at a certain altitude when a parcel of humid air is cooled below the saturation point of its contained water vapor.
Our first-principles-based formulation for buoyancy allows us to simulate the variations of atmospheric density and varying temperature gradients, which includes temperature inversion profiles, in the atmosphere. Unlike existing models, this enables us to create conditions necessary to simulate so-called cumulonimbus cloud formations with their characteristic flattened top.
Types of Clouds
In this project, we were interested in simulating the atmospheric conditions in the troposphere up to about 13km. We focus on simulating Cumulonimbus, Cumulus, Stratus, and Stratocumulus clouds which are commonly present at these altitudes.
To control the formation of clouds we identified a lightweight parameter for the conditions on the ground. These parameters control the emission of heat and vapor, as well as their distribution on the surface.
We are also able to simulate large-scale clouds called supercells. Supercells are types of clouds that are often present during thunderstorms when continuously rotating updrafts are present. We are able to simulate three different types of supercells: the classic supercell, low- and high- precipitation supercell.
We are able to simulate clouds at interactive rates for areas of up to 20km x 20km.
Our model currently does not allow simulating ice cloud formations, such as cirrus clouds. However, these types of clouds can also be supported by extending our approach. Moreover, we do not yet model the feedback of heat transfer between the ground surface and air. This means that for example shadowing effects are not considered.