Inflence of Meteorological Forecast Model Parameterization Schemes on PM2.5 Concentration Forecast Effect in Heavy Pollution Process

Based on the Nested Air Quality Prediction Model System (NAQPMS), this paper is oriented toward the parameterization of seven types of physical processes in the weather-driven model, Weather Research and Forecast Model (WRF). Fifty-one sets of different WRF model operating configurations are constructed through single disturbance and combined disturbance methods. The paper compares and analyzes the performance of NAQPMS for PM2.5 concentration forecasting during the heavy pollution period in Beijing from 16–21 December 2016, under different scheme configurations. The results show that during the heavy pollution period, the PM2.5 concentration forecast accuracy of the combined disturbance optimization scheme at the central station and the suburban station is significantly higher than the forecast results under the configuration of the baseline parameterization scheme. The combined disturbance optimization scheme can significantly improve the model’s forecast error for the end time of the heavy pollution process under the baseline scheme, and significantly reduce the forecast deviation that exists on 21 December 2016. Judging from the statistical indicators, the city center station has the highest forecast correlation under the combined optimization scheme, with a correlation coefficient>0.7; from the perspective of the root mean square error of the forecast, the combined optimization scheme has the smallest error. Furthermore, suburban stations have the highest forecast correlation under the combined optimization scheme, and the deviation from the observations is smaller than that of the central station. From the perspective of the spatial distribution of pollutants and meteorological elements, the combined disturbance optimization scheme can better reproduce the changes in meteorological elements during the pollution period than the baseline scheme. The forecasted wind speed is low and the relative humidity is high, which is conducive to the maintenance and accumulation of high concentrations of PM2.5 in Beijing on 21 December. The results of this paper show that the uncertainty of the parameterization scheme of the meteorological forecasting model is the key source of uncertainty for heavy pollution forecasting. Choosing a suitable parameterization scheme can reduce the simulation deviation of meteorological elements during the heavy pollution period and further increase the PM2.5 concentration forecast accuracy during the heavy pollution period.