Xi'an Of So <sub> 2 </ Sub> Pollution Meteorological Conditions And The Concentration Of Forecasting Methods

In this paper,the SO2 density of one-day average is acted as forestingobject. Meteorological data(temperature,wind, barometric pressure and soon)and SO2 density data are used from 1995 to 1999 day by dny. The data don'ttake part in building foresting models in 1999, which only to testing and verifyingthe effect of foresting models. According to analyzing the climate and the fact ofheating period in Xi'an, four forecasting models are built in heating period,spring, summer and autumn.Statistical forecasting methods not considering correlation andinorthogonality among forecasting factors are applied at present, so the result ofregression is instability and more error is brought. In the light of theseshortcomings, one-unit regression and empirical orthogonal function(EOF) arecombined with stepwise regression analysis method, thus a new forecastingmethod in building forecasting model is attained, which can select forecastfactors of significance linear correlation with SO2 density,turn forecastingfactors into orthogonal variables and build forecasting model. The model fittingand forecasting show that these models not only can fit the changing tendencyof SO, density, but also can forecast SO, density qulte accurae, such as, the' grade accuracy is 72.5 pprcent in heating period and l00 percellt in othePeriod o In condast with stepwise regression analysis method during the fOrecastexPerimellt, the result of the new forecasting method is more accot.In a word, the new forecasting method has good proSPect in aPPlicationaspect o...