AAAR 36th Annual Conference October 16 - October 20, 2017 Raleigh Convention Center Raleigh, North Carolina, USA
Abstract View
Interplay of Mobile Air Monitoring and Distributed Samplers to Study Intracity Spatiotemporal Variation
HUGH LI, Peishi Gu, Qing Ye, Naomi Zimmerman, Ellis Shipley Robinson, Joshua Apte, Allen Robinson, Albert A. Presto, Carnegie Mellon University
Abstract Number: 434 Working Group: Aerosol Exposure
Abstract Long-term exposure to particulate matter (PM) is the major contributor to air pollution related death in 21st century. Intracity exposure patterns can be more complex than intercity ones because of local traffic, land use, and industrial facilities. We use a hybrid sampling network to characterize spatiotemporal variations of multiple pollutants in Pittsburgh, PA. This network incorporates a mobile sampling platform, supersite, and distributed sampling together to investigate a wide range of pollutants (CO, NO, NO2, VOCs, O3, PM2.5 mass and composition, and ultrafine particles UFP). The 1st key question to address is the effective mobile sample size to generate representative long-term averages. We randomly subsample from the mobile data, and compare it to the stationary site data. We compare mobile and stationary data collected in the same 30m (or 100m) grid box so that we get a "true" comparison of the two sampling methods at the same point in space. A mobile sampling size of 20 measurements inside a 30m grid box has such statistical power to estimate long-term trends for most target pollutants. Pollutants such as NO, BC and UFP have significant intracity spatial variation after removing within day and day to day temporal variability.
Second finding focuses on spatial representativeness of long-term distributed monitors. Locations closer to monitors are more likely correlated with monitor measurements. We first use a 20% deviation criteria to define measurements agreement between two sampling platforms (mobile V.S. stationary) and identify cutoff distance for target pollutants across various types of site. And then we use the geospatial approach -- semivariogram to find the same correlation distance threshold. The final product will be correlation distance for different gas and particle pollutants, and possible metric (traffic density, dispersion characteristics) to describe the spatial range informed by stationary monitors.