This paper describes an ongoing effort to develop and integrate an empirical cloud model within a Distributed Interactive Simulation (DIS) environment in support of high-fidelity training and simulation applications. TASC is developing a cloud model (known as the Cloud Scene Simulation Model) to simulate, realistic high-resolution cloud features within domains defined by larger-scale weather conditions. The cloud model generates four-dimensional (three spatial and time), multi-layer cloud fields using a combination of stochastic field generation techniques and convection physics, where internal model parameters are tuned to fit observed cloud data. One data set is generated for each specified output time and contains cloud water density values arranged on a regular volumetric grid. A typical output field contains tens of thousands of data points covering simulation domains of 50 km or more.

Because these data sets are too dense to be transferred across the DIS network or rendered in real-time, we have developed an approach that approximates the complex cloud formations generated by the model as a series of geometric primitives. The cloud data sets are filtered to the level-of-detail appropriate for a particular simulator. The approach uses a planar-wise approximation of a volumetric phenomenon that takes advantage of today's state-of-the-art image generator hardware. The cloud model runs in real-time, allowing for smooth transitions as the weather conditions evolve over the simulation domain.

In this paper, we present an overview of the Cloud Scene Simulation Model (CSSM); its inputs, outputs, and overall methodology. We describe a DIS architecture which enables distributed real-time calculation of large cloud fields, and address usage of and extensions to the standard DIS network protocol. We follow with a description of the volumetric rendering techniques employed in this effort. Finally, we summarize and briefly discuss the application of our methodology to other atmospheric phenomena in future implementations. We conclude our oral presentation with a video tape showing real-time cloud field generation and visualization within a DIS training environment.