Main content area

High-Resolution Satellite-Derived PM2.5 from Optimal Estimation and Geographically Weighted Regression over North America

van Donkelaar, Aaron, Martin, Randall V., Spurr, Robert J. D., Burnett, Richard T.
Environmental Science & Technology 2015 v.49 no.17 pp. 10482-10491
aerosols, emissions, geophysics, particulates, satellites, statistical models, uncertainty, North America
We used a geographically weighted regression (GWR) statistical model to represent bias of fine particulate matter concentrations (PM₂.₅) derived from a 1 km optimal estimate (OE) aerosol optical depth (AOD) satellite retrieval that used AOD-to-PM₂.₅ relationships from a chemical transport model (CTM) for 2004–2008 over North America. This hybrid approach combined the geophysical understanding and global applicability intrinsic to the CTM relationships with the knowledge provided by observational constraints. Adjusting the OE PM₂.₅ estimates according to the GWR-predicted bias yielded significant improvement compared with unadjusted long-term mean values (R² = 0.82 versus R² = 0.62), even when a large fraction (70%) of sites were withheld for cross-validation (R² = 0.78) and developed seasonal skill (R² = 0.62–0.89). The effect of individual GWR predictors on OE PM₂.₅ estimates additionally provided insight into the sources of uncertainty for global satellite-derived PM₂.₅ estimates. These predictor-driven effects imply that local variability in surface elevation and urban emissions are important sources of uncertainty in geophysical calculations of the AOD-to-PM₂.₅ relationship used in satellite-derived PM₂.₅ estimates over North America, and potentially worldwide.