Ensemble-based deep learning for estimating PM 2.5 over California with multisource big data including wildfire smoke

Estimating PM concentrations and their prediction uncertainties at a high spatiotemporal resolution is important for air pollution health effect studies. This is particularly challenging for California, which has high variability in natural (e.g, wildfires, dust) and anthropogenic emissions, meteorology, topography (e.g. desert surfaces, mountains, snow cover) and land use. Using ensemble-based deep learning with big data fused from multiple sources we developed a PM prediction model with uncertainty estimates at a high spatial (1 km × 1 km) and temporal (weekly) resolution for a 10-year time span (2008-2017). We leveraged autoencoder-based full residual deep networks to model complex nonlinear interrelationships among PM emission, transport and dispersion factors and other influential features. These included remote sensing data (MAIAC aerosol optical depth (AOD), normalized difference vegetation index, impervious surface), MERRA-2 GMI Replay Simulation (M2GMI) output, wildfire smoke plume dispersion, meteorology, land cover, traffic, elevation, and spatiotemporal trends (geo-coordinates, temporal basis functions, time index). As one of the primary predictors of interest with substantial missing data in California related to bright surfaces, cloud cover and other known interferences, missing MAIAC AOD observations were imputed and adjusted for relative humidity and vertical distribution. Wildfire smoke contribution to PM was also calculated through HYSPLIT dispersion modeling of smoke emissions derived from MODIS fire radiative power using the Fire Energetics and Emissions Research version 1.0 model. Ensemble deep learning to predict PM achieved an overall mean training RMSE of 1.54 μg/m (R : 0.94) and test RMSE of 2.29 μg/m (R : 0.87). The top predictors included M2GMI carbon monoxide mixing ratio in the bottom layer, temporal basis functions, spatial location, air temperature, MAIAC AOD, and PM sea salt mass concentration. In an independent test using three long-term AQS sites and one short-term non-AQS site, our model achieved a high correlation (>0.8) and a low RMSE (