Comparative evaluation of backpropagation neural network and genetic algorithm-backpropagation neural network models for PM2.5 concentration prediction based on aerosol optical depth, meteorological factors, and air pollutants

Abstract

Fine particles with an aerodynamic diameter ≤2.5 μm are called PM 2.5 , and accurate prediction of PM 2.5 concentration can help prevent the harmful effects of heavy pollution on humans. At present, the distribution of ground-based PM 2.5 monitoring stations in China's cities is relatively sparse. Hence, the aerosol optical depth (AOD) obtained from satellite remote sensing provides an effective means for large-scale routine PM 2.5 monitoring. In this study, the multi-angle implementation of atmospheric correction AOD (1 km resolution) product from the moderate resolution imaging spectroradiometer (MODIS), meteorological factors, and ground measurements of daily PM 2.5 concentrations from 2016 to 2020 for Dalian were input into a backpropagation neural network (BPNN) to predict PM 2.5 concentrations. To improve the prediction accuracy and stability, a genetic algorithm (GA)-optimized BPNN was further established based on the BPNN to achieve a comparative PM 2.5 concentration prediction. Results showed that the BPNN and GA-BPNN achieved the PM 2.5 concentration by integrating AOD, meteorological factors, and air pollutants with the model test set R 2 of 0.77 and 0.83 and root mean square error (RMSE) of 11.83 and 9.80 μg m −3 , respectively. GA-BPNN decreased the SD RMSE from 0.44 to 0.23 μg m −3 compared with BPNN and improved the model stability. The spatial distribution of annual averaged estimated PM 2.5 was predicted using GA-BPNN in Dalian from 2016 to 2020. The spatial distribution of PM 2.5 concentrations was generally consistent over 5 years, and the PM 2.5 concentrations exhibited an overall decreasing trend. The BPNN before and after optimization achieved longer-term interannual PM 2.5 daily concentration prediction, and the GA-BPNN had a better prediction effect for extreme values, handled complex fuzzy mapping relationships, and had lower computational complexity. Hence, GA-BPNN was found to be more suitable for practical applications with more advantages for PM 2.5 concentration prediction than BPNN.