عنوان مقاله [English]
The amount of dust and other air pollutants have increased during recent years. The PM10 is one of the most important variables that is used for monitoring and assessment of dust pollution. To predict PM10, different studies have used various statistical methods. In this study, two aims were pursuit: 1) Using of Spearman analysis to determine the relation between PM10 and weather variables such as temperature (maximum, average, minimum), relative Humid (maximum, average, minimum), rain, wind (speed and direction), and visibility, and 2) prediction of PM10 with using of Random Forest model on daily data (in a period study: 2008 to 2011). The results of Spearman analysis were shown that PM10 had most relation with visibility and minimum temperature and least relation with rain, -0.376, +0.349, and -0.077, respectively. In addition, Random Forest analysis was shown that for prediction of PM10, visibility and minimum temperature were very important. Fitting curve between observed and prediction data was shown a medium correlation with Y=0.1686x +183.49 and R2=47/0, sig=0.99. Final sequence of trees of random forest was shown that of all data, just maximum and minimum of relative humid and minimum of temperature were able to classification with 396 (>0.205 %), 389(>0.305 %), and 387 (>5.5 oC) data for each variable, respectively.