Modeling The Atmosphere of Sulfur Dioxide (SO2) and Particulate Matter (PM) Concentration at the Isparta Province by using Multi-Linear Regression

25302

In this study, totally six model equation are obtained for SO2 and PM by stepwise selection model in the multiple linear regression model by using daily data of meteorology and air quality observed for five years belonging to 2007-2012 winter periods (October-March). For the winter season 2011-2012, SO2 and PM predictions are made by separately using the obtained models. The prediction performances of the models are stated with the help of average error square root, fit index and correlation coefficient. Besides, cross correlation assessment which frequently used in stating model performances and selecting independent parameters which defines the relationship between pollutants and meteorological variables is made. When we look at the performance values belonging to SO2 models including 2011-2012 winter period statistic assessments, it is expected that square root mean error to be low, fit index and correlation coefficient to be high. When these values are analyzed, it is seen that values for square root mean error is 58,5, fit index is 0,90 and correlation coefficient 0,84.The best estimation for SO2 values in this period is achieved by using the third model equation. When we survey performance values belonging to PM models in 2011-2012 winter periods, it is seen that values for square root mean error is about 112, fit index is 0,89 and correlation coefficient is 0,70. The best estimation for PM models in this period is achieved by using the first and third pattern model equations.

95

AirPollution , Statistical Models , Meteorological Factors , Correlation , SO2 , PM

Journal Of Natural an‎d Applied Sciences

108

Journal Of Natural an‎d Applied Sciences