Outdoor beam tracing over undulating terrain

Constructing noise maps for large areas is a tedious and CPU intensive task, even if engineering approximations to outdoor sound propagation models are introduced. This paper reports on an effort to optimize beam tracing in order to reduce the number of generated paths between source and observer, while still guaranteeing the required accuracy. Beam tracing is a ray based technique that groups rays into spatial coherent beams. These beams are restricted to the geometry at hand. Due to its nature it is very efficient in highly occluded areas. However, the proposed modifications allow the algorithm to be used efficiently in sparse outdoor scenes with complex undulating terrain as well. The beam tracing algorithm gradually turns into a Monte Carlo ray tracing algorithm while it propagates through the scene. Beams are assigned a chance of survival depending on the distance traveled. To conserve energy the remaining beams are assigned a weight that grows with propagation distance. Statistically the algorithm redistributes energy perfectly but Monte Carlo noise is still visible if no counter measures are taken. Therefore super-sampling is used which suits the beam tracing algorithm well. Another technique that reduces the variance on the noise significantly is to distribute the energy of highly weighted beams to receivers in the neighborhood. Efficient data structures are needed to support these operations. The improved beam tracing algorithm has been implemented and tested in the mountainous region of the Brennerpass (Austria) with good results. The calculation time of the simulation can easily be traded with the variance and resolution of the results. Noise mapping and ray tracing algorithms