AAAR 36th Annual Conference October 16 - October 20, 2017 Raleigh Convention Center Raleigh, North Carolina, USA
Abstract View
Development of Novel Model Fusion Method for Simulating Spatially Resolved (250-m) Air Pollutant Concentration Estimates
JOSEPHINE BATES, Audrey Pennington, Xinxin Zhai, Mariel Friberg, Francesc Metcalf, Matthew Strickland, Lyndsey Darrow, James Mulholland, Armistead G. Russell, Georgia Institute of Technology
Abstract Number: 325 Working Group: Urban Aerosols
Abstract Air pollutant concentrations can vary greatly in time and space, especially near strong emission sources like roadways. Modelling spatially resolved air pollutant concentrations is critical for accurately capturing urban air quality influenced by numerous local emission sources. Two novel model fusion methods were developed to estimate gaseous and particulate matter air pollutant concentrations at a fine spatial resolution without losing chemistry or emission source information. Methods were applied to the Atlanta, GA region to obtain daily estimates of 24-hr averaged PM2.5 and 1-hour maximum CO and NOx from 2002-2011 at a 250m grid resolution. The methods combine 12km CMAQ estimates, which are temporally resolved and contain comprehensive chemistry and source information but are spatially coarse, and 250m annual average RLINE estimates, which are spatially resolved but temporally coarse and only account for primary, inert emissions from vehicles, in a computationally efficient manner using linear combinations and a mass conservative bilinear interpolation algorithm. The model fusion methods can be applied to different grid resolutions, inputs from different model types, and different pollutants, providing flexibility. Results for Atlanta, GA show steep spatial gradients in pollutant concentrations near roadways, accurately capturing intraurban variability due to vehicle use. Model fusion estimates were evaluated against data from available monitoring sites with withholding and results show an improvement in both spatial R and temporal R2 compared to CMAQ and RLINE, emphasizing the ability of the model fusion results to capture not only fine-scale spatial variation but also comprehensive chemistry leading to significant secondary formation and other local and regional emission sources. Simulating gradients near roadways while also capturing effects of regional sources that can influence daily variation in concentration can potentially reduce exposure errors in health studies.