AAAR 37th Annual Conference October 14 - October 18, 2019 Oregon Convention Center Portland, Oregon, USA
Abstract View
High Resolution Chemical Transport Modeling of Ultrafine Particles over Pittsburgh
SHAYAK SENGUPTA, Pablo Garcia, David Patoulias, Provat Saha, Wei Ma, Christopher Tessum, Iannis Kioutsioukis, Sean Qian, Spyros Pandis, Inês Azevedo, Peter Adams, Carnegie Mellon University
Abstract Number: 361 Working Group: Urban Aerosols
Abstract Ambient ultrafine particles (UFPs), solid or liquid particles in the atmosphere with diameters less than 100 nm, pose poorly understood human health impacts relative to the well-understood impacts of PM2.5. Numerous studies have documented health effects related to roadway proximity, with UFP emissions from traffic as a possible culprit. However, high spatial variability and the lack of widespread monitoring complicate exposure quantification of UFPs. The goal of this work is to develop and evaluate high resolution (1 km) chemical transport model (CTM) simulations to quantify UFP concentrations as a step towards quantifying UFP exposure in an urban area. This study uses PM-CAMx-UF to predict UFP concentrations in the Pittsburgh metropolitan area at 1 km spatial resolution for February 2017 and July 2017. PM-CAMx-UF is a state-of-the-science CTM which simulates the production and destruction of UFPs in the atmosphere by explicitly tracking both particle number and mass concentrations and solving the general dynamic equation for aerosol microphysics. Model inputs include traffic emissions at 1 km resolution, spatially resolved using a traffic model for Pittsburgh. Baseline simulations indicate February 2017 particle number concentrations (PNC) in Pittsburgh vary by more than a factor of two, with a mean concentration of 9,000 cm-3. Comparisons to a network of 27 long-term (~1 month) winter observation sites in Pittsburgh show model spatial agreement with MFB = -15% and MFE = 22% at 13 urban background and local sites. At 14 sites influenced by local sources or topography, there is poorer agreement with MFB = -39% and MFE = 39%, indicating the importance of local sources and variability that are not resolved even at 1 km model resolution. Temporally, the model matches winter diurnal variability in PNC at 12 sites with r2 values exceeding 0.29. Sensitivity simulations to quantify source apportionment show on-road traffic contributes to 29% of predicted PNC within Pittsburgh followed by both stationary wood combustion and waste disposal at 17%. We estimate natural gas combustion to contribute 15% of predicted PNC. Using a traffic model to spatially allocate traffic emissions inputs captures the traffic patterns better: predicted NO2 at 46 distributed NO2 observation sites across Pittsburgh yields higher agreement (r2 = 0.39, MFB = -13%, MFE = 21%) than simulations at default EPA spatial surrogates (r2 = 0.29, MFB = -27%, MFE = 51%). Primary emissions largely govern the spatial variability in predicted PNC at the 1-km scale. Mean predicted particle lifetime due to coagulation is 10 hours within the city with only 7% of particles lost due to coagulation. Modeling at 1-km resolution resolves more variability in human exposure. The 5th-95th percentiles of population were estimated to be exposed to 3,400-8,700 cm-3 at 1 km resolution, but 3,500-7,000 cm-3 at 4 km resolution.