The Sparse Matrix Operator Kernel Emission model (SMOKE) was developed by the Environmental Modeling Center, MCNC, to integrate emissions data processing in preparation of input data for Air Quality Models (AQM). The development took place under cooperative research agreements and contracts with the U.S. EPA. SMOKE has been integrated with the U.S. EPA Models-3 system for which it may be used to provide emissions input data.

The SMOKE system facilitates support for decision making on emissions control for urban and regional models. It is primarily an emissions processing system and not an emission inventory system. However, for the specific case of mobile sources, SMOKE can compute emissions inventory from mobile source activity data.

Emissions inventory data are accepted by SMOKE in the Inventory Data Anlayzer (IDA) format. Speciation for selected gas phase lumped species mechanisms is included. Also supported is speciation of fine particulate emissions in six particulate categories to support the aerosol chemistry portions of CMAQ. SMOKE supports area, mobile, and point source, emissions processing. Hourly biogenic and on-road mobile emissions data are modeled directly using the Biogenic Inventory System (BEIS 3.09) and Mobile 5b, respectively, that are embedded in SMOKE. BEIS 3.09, in the former case, uses either county level resolution, or 1km resolution North American BELD land cover data.

The algorithms used in SMOKE supercede the less efficient ones of previous systems and, as a consequence, processing is both rapid and flexible. For example, a change of control strategy in SMOKE requires only the control and merge steps to be processed again. Legacy systems required the repetition of multiple steps such as temporal projection, controls, chemical speciation, temporal and spatial allocation, etc. Furthermore, the use of dynamic memory allocation with Fortran 90 in SMOKE eliminates the requirement in legacy algorithms for recompilation with changes in inventories, grids, number of states/counties, etc.