Evaluating the utility of high-resolution air pollution data: Comparing the importance of temporal and spatial variability in estimating local air pollution exposures in California from 2015-2018

Abstract

Background: For many years, air quality researchers have focused heavily on improving the spatial resolution of outdoor pollution estimates. This has been accomplished through improved satellite remote sensing technologies combined with complex models, currently producing surface-level estimates down to the kilometer and sub-kilometer scales. However, as resolutions grow increasingly refined, it is important to evaluate how much new information additional resolution provides and compare the relative importance of spatial and temporal variability in assigning pollution exposures to study populations. Methods: Daily PM2.5 estimates at a 1 km2 resolution, derived from NASA’s MODIS satellite instrument and the MAIAC algorithm, were estimated for 2015-2018 and linked to California ZIP codes using spatial interpolation weighting factors. Within- and between-ZIP code spatial and temporal variations were calculated for all cities with at least three ZIP codes, then aggregated and analyzed using time-series and GIS analyses. Results: Variation of PM2.5 concentrations within ZIP codes is negligible, with much greater variation observed between ZIP codes in the same city. Within and between ZIP code variation did show strong seasonal influences with greater variation observed during the cooler months of the year, particularly in Northern California which is heavily influenced by local emissions sources. In all cases, temporal variability was much greater than spatial variability at the sub-urban spatial resolutions included in this study. Conclusions: For much of the year, there is little additional information gained by using spatial resolutions more refined than ZIP code level estimates in California during the study period. However, during specific seasons there may be additional value in assigning exposures based on resolutions down to 1 km2. As air quality research moves forward, there may be diminishing returns from further improvements in spatial resolution for PM2.5 and efforts may be better allocated to other areas of exposure assessment.