Studies On The Long-term Forecast Of PM₁₀ Concentratio In Xi'an City

The use of Xi'an City in January 2001, December 2005 air monitoring daily PM₁₀ concentration values in the spatial and temporal distribution of PM₁₀ profiling on the basis of the screening programme were used by orthogonal modelling and the main elements what is the function of a programme cycle used forecasting models, Xi'an's PM₁₀ pollution in the medium to long term forecasting research.Xi'an PM₁₀ concentrations time profiling shows : 2001 -2005, the overall trend of yearly decline in PM₁₀ concentrations, but in 2002 and 2004 PM₁₀ concentrations rise;Winter and spring PM₁₀ pollution more serious, less polluting summer and autumn;Heating quarter PM₁₀ concentrations than non-heating quarter higher. 5 monitoring points PM₁₀ pollution measured in the order of 2001 -2002 followed by Hamlet, Textiles City, Xingqing district, Pressure switches plants, Caotan, 2003-2005, followed by Textiles City, Xingqing district, Pressure switches plants, Hamlet, Cao Tan.Orthogonal screening programme using the couple-score criterion to identify screening guidelines since the variables properly extract data series advantage of the cycle as the equation since variables. The modelling component elements of the programme will be the main analysis and Fangcha over combining principles from linear variables between the total elimination of guaranteed access to the model cycle extended up factor is the function of the remarkable nature of the adoption cycle will spur formation and function are known to reduce the function of selection are then used to screen out weight and the main factor to reinstate modelling.Fitting and forecasting tests showed that forecasting results are satisfactory, the two forecasting equation modelling programme of PM₁₀ concentrations measured changes in the trends and forecasting trends are very consistent, not only can a good fit the trend of changes in the concentration of air pollutants, but also to peak, valley value can more accurately forecasting out. Orthogonal screening programme which the average absolute prediction error for 0.14%, to 0.876 correlation coefficient R, symbols consistent withthe rate of 100%. Modelling of the main components of the programme forecast to 0.25% average absolute error, the correlation coefficient R for 0.742,consistent with the rate of 90%. By contrast, orthogonal screening programme established forecasting models over the main components modelling programme established forecasting models to determine this recommended Xi'an PMio pollution-forecasting models.