Abstract:

On the basis of air quality research for the last five years in the city of Changsha，two air quality statistical forecasting methods were presented in details in this paper，and verification of the two methods was conducted for more than one year.It was revealed that air pollution in Changsha was mainly caused by PM10 and SO2，which exceeded the official permission value.The air pollution showed some degree of spatial distribution.However，changes in the air quality showed a trend towards a better situation over the last five years.The two methods can be used to determinate the forecasting ability for various contaminations.The results showed that the two methods have an accuracy of about 98% for the forecast of NO2；the methods have an obvious effect than the PM10 forecast for the forecast of SO2；and the two methods have similar forecasting results for the three kinds of contaminations.It was indicated that for the two air quality forecasting methods，the forecasting accuracy of NO2 is superior to that of SO2，and the forecasting accuracy of SO2 is superior to that of PM10.

Key words: air quality, forecasting methods, weather pattern similarity, stepwise regression