Rendering and Modeling of Stratus Cloud Using Weather Forecast Data

This paper presents a method for physically plausible modeling and real-time rendering of large scale cloud scene from weather forecast data which includes stratus clouds and cumulus clouds. Our method uses particles as primitives, but differs from existing methods, the particles are generated from the data using a new optical model, and then are splatted onto multi-resolution axis-aligned slices for rendering at interactive frame rates. In the cloud optics model, multiple light sources are used and multiple anisotropic Mie scattering and Rayleigh scattering are employed according to the particle optical depth. After particles are generated from weather forecast data, multi-resolution axis-aligned slices are generated. The rendering process can use these slices and be easily implemented on graphics hardware. The experiment results show that realistic cloud can be modeled and rendered at interactive frame rates.