Parallel Aerosol Air Quality Models

摘要

Accurate aerosol modeling requires tracking the time evolution of the aerosol size and composition distributions, which is very compute intensive. Massively parallel computing provides an avenue for managing this workload. Towards this end, a parallel version of the modal aerosol code has been developed using a messaging passing paradigm. Details of this implementation, portability considerations, performance measures and cross-platform comparisons, as well as model predictions of aerosol mass concentrations and size distributions will be presented.