High Performance Algorithms In The Sparse Matrix Operator Kernel Emissions (smoke) Modeling System

Emissions modeling is one of the major bottlenecks limiting the usefulness of mesoscale Eulerian air quality modeling for environmental planning. Frequently, emissions modeling has computational and storage requirements as large as the air quality models themselves. Improving the turnaround for processing an emissions control strategy from days (or even weeks) to minutes or seconds will have major impacts on the usability of air quality models for environmental decision making. As emissions models are presently formulated, not even supercomputers can help this problem—the models hide the parallelism necessary to use such machines effectively.