A new approach combining a simplified FLEXPART model and a Bayesian-RAT method for forecasting PM₁₀ and PM_2.5

Area of Science:

• Atmospheric Science
• Environmental Modeling
• Air Quality Forecasting

Background:

• Particulate matter (PM) poses significant health risks, necessitating accurate regional-scale concentration predictions.
• Existing models often require improvement for precise forecasting of PM10 and PM2.5.

Purpose of the Study:

• To evaluate a simplified Lagrangian particle dispersion (LPD) modeling system combined with Bayesian-RAT for enhanced PM concentration prediction.
• To assess the LPD system's performance against observational data and compare it with other established models.

Main Methods:

• Utilized a simplified Lagrangian particle dispersion (LPD) system, integrating the Bayesian and multiplicative ratio correction optimization (Bayesian-RAT) method.
• Compared LPD model predictions with 95 observation stations in Xingtai, China, during December 2017.
• Benchmarked LPD against WRF-Chem and Camx models for PM10 and PM2.5 hourly concentrations.

Main Results:

• The LPD forecast system showed significantly improved performance over pre-calibration results (e.g., R=0.64 vs. 0.48 for PM10).
• LPD demonstrated higher accuracy than WRF-Chem and Camx, reducing RMSE for PM10 by 36.51% and 42.21%, respectively.
• PM2.5 forecast accuracy was also superior, with RMSE reductions of 26.44% and 18.47% compared to WRF-Chem and Camx.

Conclusions:

• The LPD forecast system offers substantial advantages for predicting PM concentrations at a regional scale.
• The validated LPD system shows potential for improving forecasts of other atmospheric pollutants as well.