Combination of weather factors and Area of High Variation to estimation PM2.5 concentration

Abstract

Particulate matter, also referred to as fine PM2.5, is divided into natural and man-mad. That is one of the important indicators of air pollution. It is harmful for human organs, when someone inhales the substance. Four image characteristics and three weather factors are considered to estimat PM2.5. In this paper, providing Area of High Variation(AoHV) a method to calculate the corresponding pixel features in the mage and add the three weather features, import that into the SVR model for calculation; then, that could get the value of PM2.5 be estimated. The AoHV method based on the imagery and weather information provided by the National Monitoring Station. Furthermore, compared the AoHV method with Chen’s method. The results of experimental prove adding three estimations of weather factors, including RH, temperature and wind is better than Chen’s method that only added RH. However, using the AoHV method proposed in this paper, the best estimation result of PM2.5 that $\text{R}^{2}$ value reaches 0.944, and the RMSE value reaches 4.131. There is a certain degree of improvement in estimated results.