As an emerging technique, machine learning algorithms can capture the complex nonlinear relationships with interactions. Compared to the traditional statistical method, machine learning is more reliable for the air pollution analysis. PM2.5 and O3 are the two secondary ambient pollutants which can exert adverse health effects on the human beings. In this study, we plan to combine machine learning techniques (BP neural network and convolutional neural network), ground observation, satellite data to generate high resolution PM2.5/O3 products for the Pearl River Delta region.
The student needs to learn various machine learning techniques
By completing this project, students would have basic understand of air quality forecasting and its uncertainty.