10th International Aerosol Conference September 2 - September 7, 2018 America's Center Convention Complex St. Louis, Missouri, USA
Abstract View
Towards High-Resolution Air Pollution Mapping: Fusing Mobile PM Measurements with Data from a Dense Low-Cost Sensor Network
SARAH CHAMBLISS, Kyle Messier, Chelsea V. Preble, Julien Caubel, Ramon Alvarez, Brian LaFranchi, Melissa M. Lunden, Thomas W. Kirchstetter, Joshua Apte, University of Texas at Austin
Abstract Number: 1358 Working Group: Low-Cost and Portable Sensors
Abstract Emerging techniques for measuring urban air pollutant concentration at high spatial and temporal resolution have the potential to greatly improve estimates of pollution exposure for urban populations. Two such techniques are (1) the deployment of a city-wide network of low-cost sensors and (2) mobile monitoring campaigns in which reference-grade instruments are routinely driven along all roads in a specified sampling area. Current research in the field seeks to answer how either technique may be optimally deployed to yield representative long-term exposure estimates with lower measurement density, and how a combination of techniques may provide complementary information to further improve exposure estimates. This study examines mobile and fixed black carbon (BC) measurements collected during an intensive two-month field campaign in summer 2017 in West Oakland, California. A dense network of 100 fixed sites was established within a 5 km2 neighborhood. Time-resolved (1 min) BC measurements were collected using a novel low-cost sensor based on light absorption through a filter medium. Concurrent mobile measurements within this same domain were collected during a total of 341 sampling hours by two Google Street View cars specially equipped with photoacoustic extinctiometers measuring at 1 Hz.
The study compares 1 Hz and time-averaged on-road measurements to lower frequency measurements from nearby low-cost sensors to identify biases that may exist in either measurement method and to assess the comparative sensitivity of these measurements to localized emissions sources and short-term emissions events (e.g., vehicle plumes). On-road measurements include measurements taken while driving and collocation events, during which a Street View car was parked along a public street in proximity to a low-cost sensor. Thirty collocation sites were selected to provide an even distribution of sampling locations within the neighborhood and to represent a range of land uses and road types. Distances between parking and location sites ranged from 3 to 100 m. Measurements were collected for a minimum sampling duration of 5 minutes, with an average event duration of 15 minutes. The campaign included 183 full collocation events for a total duration of 47 sampling hours.
Preliminary analyses of the collocation data indicate modest correlation between BC concentration means calculated for the duration of each stop, and little to no correlation between one minute measurements. This loss of correlation with higher time resolution, along with high within-event variation and intermittent peaks observed in measurements taken at the parked car, may indicate the degree of increased sensitivity of on-road measurements to high frequency changes in emissions. Preliminary results also suggest that land use type may influence the level of agreement between a parked mobile monitor and off-road fixed site sensors. Further analysis will include comparisons with in-motion mobile measurements.
The findings of this study offer insight into how pollution monitoring campaigns may effectively integrate complementary sampling techniques to improve both the spatial and temporal resolution of exposure estimates. These findings may also suggest methods of improving the interpretation of mobile monitoring measurements to assess off-road urban exposure.