AAAR 33rd Annual Conference
October 20 - October 24, 2014
Rosen Shingle Creek
Orlando, Florida, USA
Abstract View
A Reduced-Complexity, Variable Grid Resolution Model for PM2.5 Transport and Transformation
CHRISTOPHER TESSUM, Jason Hill, Julian Marshall, University of Minnesota
Abstract Number: 30 Working Group: Linking Aerosols with Public Health in a Changing World
Abstract Background: Air pollution modeling is usually done with either gaussian plume models (e.g., AERMOD, ISC) or with temporally explicit chemical transport models (e.g., CMAQ, CAMx, WRF/Chem). While both model types can be useful, the former neglects important physical and chemical processes, and the latter is commonly limited by computational costs.
Aims: To create a novel chemical transport model that includes the physical and chemical processes most important in determining PM$_(2.5) regulatory compliance and human exposure, that can simulate both intra-urban concentration gradients and long-range transport of pollution, and that can be run on a single desktop computer.
Methods: The model is based on a steady-state Eulerian framework with a population-based variable resolution computational grid. Meteorological and background chemical information is derived from the output of a 12-month, 12-km spatial resolution WRF/Chem simulation for the continental United States. The model includes mechanisms for advection, turbulent diffusion, pollutant removal by wet and dry deposition, and for partitioning between gas and particle phases for sulfur, nitrate, ammonia, and organic matter.
Results: We compare annual-average results of 12-month, continental U.S., 12-km spatial resolution WRF/Chem simulations for 12 emissions scenarios to predictions by our simplified model for the same emissions scenarios and spatial grid. For domain-average total PM$_(2.5) concentrations: R$^2=0.97, mean fractional bias=-43%. For population-weighted total PM$_(2.5) concentrations: R$^2=0.97, mean fractional bias=-2%. In general, agreement was better for primary PM$_(2.5) and particulate sulfate concentrations than for concentrations of other compounds. Each simplified model run takes ~30 processor hours, compared to ~120,000 processor hours for the comparable WRF/Chem simulation.
Conclusion: Our modeling approach can provide useful pollution simulations at several orders of magnitude less computational cost than comprehensive chemical transport models.