ESTIMATING PARTICULATE MATTER ${bf (PM_{10})}$ CONCENTRATION USING REMOTE SENSING TECHNIQUE AND METEOROLOGICAL PARAMETERS OVER TEHRAN

Determination of particulate matter (PM) levels, as one of the most important pollutants, requires dense monitoring stations network in megacities. Extending monitoring network, especially in regions with sparse monitoring sites, needs a signicant economic source and may be rejected due to feasibility considerations. During the last decade, remotely sensed atmospheric data are known as a cost- eective way with appropriate and comprehensive spatial and temporal coverage to estimate ground-based PM concentrations. In this regard, aerosol optical depth (AOD), which represents the amount of aerosol in the column of atmosphere, was used as an independent satellite derived product to predict PM values in monitoring stations. Also, meteorological variables can be used as auxiliary parameters to improve model performance during validation period. In this study, a statistical model was developed using AOD along with eective meteorological parameters to estimate ground level of PM10 (particulate matters with aerodynamic diameter less than 10m). AOD was extracted from 6 collections 6 of Moderate Resolution Imaging Spectroradiometer (MODIS) by 3 km spatial resolution over Tehran during Marchof 2009. Meteorological variables can specify vertical distributions in atmospheric November column and optical properties of PM, and they are capable to improve AOD and PMs relationship. So, to improve the model performance, model it is developed by meteorological parameters. The meteorological parameters were collected from synoptic stations in Tehran, every 3 hours, during the study period. The linear mixed eect model was tted into all independent variables to examine their in uence on PM10 concentrations. The results showed that the proposed model could explain concentration accurately with relative high correlation coecient of the variation of daily PM10 ( R2 = 0.77 ). Statistical model performance was acceptable during cross validation with 0.88 (R2 = 0.61 ). The model had the best performance during correlation coe cient of 0.78 autumn with root mean square error (RMSE) of 15:4g/m3, while the worst one occurred in summer with RMSE of 19:3g/m3.