Estimation of ground-level PM2.5 concentrations with MERRA-2 aerosol diagnostics over the northwestern region of Turkey

  • Version
  • Download 775
  • File Size 2.05 MB
  • Create Date March 21, 2022
  • Download

Gizem Tuna Tuygun*, Serdar GÜNDOĞDU, Tolga Elbir

Atmospheric aerosols with an aerodynamic diameter of less than 2.5 micrometers (PM2.5) are the most widely studied component of air pollution. Due to the lack of ground-level monitoring stations, PM2.5 concentrations are not widely monitored  in Turkey. The Version 2 Modern-Era Retrospective analysis for Research and Applications (MERRA-2) provides spatially and temporally continuous a variety of aerosol diagnostic products from MERRA-2, including surface mass concentrations of dust (DUSMASS25), sea salt (SSSMASS25), black carbon (BCSMASS), organic carbon (OCSMASS), and sulfate (SO4MASS). With these parameters, ground-level PM2.5 concentrations were estimated using the Extreme Gradient Boosting (XGBoost) model at 48 air quality monitoring stations in the Marmara region, the most populated and industrialized area in Turkey. Several meteorological parameters from MERRA-2 were also used as inputs to the estimation model. Temporal variables (i.e., Year, Month, Day) were incorporated as covariates. The integration of these parameters directly affects the prediction accuracy in the model. Ten-fold cross-validation (10-CV) was used to evaluate model performance. Finally, ten-fold prediction results were combined and compared with PM2.5 concentrations from the ground-level monitoring station. The correlation coefficient (R values) and root mean square error (RMSE) were calculated as model performance indicators. Modeling results achieved overall daily cross-validation (CV) R, RMSE, and MAE values of 0.86, 9.52, and 5.85 µg/m3, respectively.

Keywords: MERRA-2, Aerosol diagnostic, PM2.5 estimation, XGBoost, Turkey

Proscience vol. 8

Pp 30-38

DOI: 10.14644/dust2021.004