Abstract:Ambient exposure to fine particulate matter (PM2.5) is known to harm public health in China. Satellite remote sensing measurements of aerosol optical depth (AOD) were statistically associated with in-situ observations after 2013 to predict PM2.5 concentrations nationwide, while the lack of surface monitoring data before 2013 have created difficulties in historical PM2.5 exposure estimates. Hindcast approaches using statistical models or chemical transport models (CTMs) were developed to overcome this limitation, while those approaches still suffer from incomplete daily coverage due to missing AOD data or limited accuracy due to uncertainties of CTMs. Here we developed a new machine learning (ML) model with high-dimensional expansion (HD-expansion) of numerous predictors (including AOD and other satellite covariates, meteorological variables and CTM simulations). Through comprehensive characterization of the nonlinear effects of, and interactions among different predictors, the HD-expansion parameterized the association between PM2.5 and AOD as a nonlinear function of space and time covariates (e.g., planetary boundary layer height and relative humidity). In this way, the PM2.5-AOD association can vary spatiotemporally. We trained the model with data from 2013 to 2016 and evaluated its performance using annually-iterated cross-validation, which iteratively held out the in-situ observations for a whole calendar year (as testing data) to examine the predictions from a model trained by the rest of the observations. Our estimates were found to be in good agreement with in-situ observations, with correlation coefficients (R2) of 0.61, 0.68, and 0.75 for daily, monthly and annual averages, respectively. To interpolate the missing predictions due to incomplete AOD data, we incorporated a generalized additive model into the ML model. The two-stage estimates of PM2.5 sacrificed the prediction accuracy on a daily timescale (R2 = 0.55), but achieved complete spatiotemporal coverage and improved the accuracy of monthly (R2 = 0.71) and annual (R2 = 0.77) averages. The model was then used to predict daily PM2.5 concentrations during 2000-2016 across China and estimate long-term trends in PM2.5 for the period. We found that population-weighted concentrations of PM2.5 significantly increased, by 2.10 (95% confidence interval (CI): 1.74, 2.46) μg/m3/year during 2000-2007, and rapidly decreased by 4.51 (3.12, 5.90) μg/m3/year during 2013-2016. In this study, we produced AOD-based estimates of historical PM2.5 with complete spatiotemporal coverage, which were evidenced as accurate, particularly in middle and long term. The products could support large-scale epidemiological studies and risk assessments of ambient PM2.5 in China and can be accessed via the website (http://www.meicmodel.org/dataset-phd.html).

A machine learning approach to downscale EMEP4UK: analysis of UK ozone variability and trends

Future Local Ground‐level Ozone in the European area from Statistical Downscaling Projections Considering Climate and Emission Changes

A multiscale geographically weighted regression kriging method for spatial downscaling of satellite-based ozone datasets

Downscaling a global climate model to simulate climate change impacts on US regional and urban air quality

Ozone Concentration Estimation and Meteorological Impact Quantification in the Beijing‐Tianjin‐Hebei Region Based on Machine Learning Models

Investigating the Effect of Fine Particulate Matter (PM2.5) Emission Reduction on Surface-Level Ozone (O3) during Summer across the UK

Supplementary Material to "tracking Daily NOx Emissions from an Urban Agglomeration Based on TROPOMI NO2 and a Local Ensemble Transform Kalman Filter"

A Framework for Scalable Ambient Air Pollution Concentration Estimation

An Ensemble Learning Approach for Estimating High Spatiotemporal Resolution of Ground-Level Ozone in the Contiguous United States.

Ozone, nitrogen dioxide, and PM2.5 estimation from observation-model machine learning fusion over S. Korea: Influence of observation density, chemical transport model resolution, and geostationary remotely sensed AOD

High resolution mapping of nitrogen dioxide and particulate matter in Great Britain (2003-2021) with multi-stage data reconstruction and ensemble machine learning methods

A machine learning model for estimating daily maximum 8-hour average ozone concentrations using OMI and MODIS products

Predicting particulate matter, nitrogen dioxide, and ozone across Great Britain with high spatiotemporal resolution based on random forest models

Machine Learning-Based Approach Using Open Data to Estimate PM2.5 over Europe

Inferring Surface NO2 Over Western Europe: A Machine Learning Approach With Uncertainty Quantification

The use of satellite and ground based measurements for estimating and reducing uncertainties in the spatial distribution of emissions of nitrogen oxides

Estimation of surface-level NO2 and O3 concentrations using TROPOMI data and machine learning over East Asia

Urban ozone variability using automated machine learning: inference from different feature importance schemes

The impact of synoptic weather on UK surface ozone and implications for premature mortality

Spatiotemporal Continuous Estimates of PM2.5 Concentrations in China, 2000-2016: A Machine Learning Method with Inputs from Satellites, Chemical Transport Model, and Ground Observations.

Analysis of European Ozone Trends in the Period 1995–2014