Ensemble prediction of air quality using the WRF/CMAQ model system for health effect studies in China

摘要

Accurate exposure estimates are required for health effect analyses ofsevere air pollution in China. Chemical transport models (CTMs) are widelyused to provide spatial distribution, chemical composition, particle sizefractions, and source origins of air pollutants. The accuracy of air qualitypredictions in China is greatly affected by the uncertainties of emissioninventories. The Community Multiscale Air Quality (CMAQ) model withmeteorological inputs from the Weather Research and Forecasting (WRF) modelwere used in this study to simulate air pollutants in China in 2013. Foursimulations were conducted with four different anthropogenic emissioninventories, including the Multi-resolution Emission Inventory for China(MEIC), the Emission Inventory for China by School of Environment atTsinghua University (SOE), the Emissions Database for Global AtmosphericResearch (EDGAR), and the Regional Emission inventory in Asia version 2 (REAS2). Model performance of each simulation was evaluated againstavailable observation data from 422 sites in 60 cities across China. Modelpredictions of O and PM generally meet the model performancecriteria, but performance differences exist in different regions, fordifferent pollutants, and among inventories. Ensemble predictions werecalculated by linearly combining the results from different inventories tominimize the sum of the squared errors between the ensemble results and theobservations in all cities. The ensemble concentrations show improvedagreement with observations in most cities. The mean fractional bias (MFB)and mean fractional errors (MFEs) of the ensemble annual PM in the 60cities are −0.11 and 0.24, respectively, which are better than the MFB(−0.25 to −0.16) and MFE (0.26–0.31) of individual simulations. Theensemble annual daily maximum 1 h O (O-1h) concentrations arealso improved, with mean normalized bias (MNB) of 0.03 and mean normalizederrors (MNE) of 0.14, compared to MNB of 0.06–0.19 and MNE of 0.16–0.22 of the individual predictions. The ensemble predictions agree betterwith observations with daily, monthly, and annual averaging times in allregions of China for both PM and O-1h. The study demonstratesthat ensemble predictions from combining predictions from individual emissioninventories can improve the accuracy of predicted temporal and spatialdistributions of air pollutants. This study is the first ensemble modelstudy in China using multiple emission inventories, and the results arepublicly available for future health effect studies.