Particulate matters (PM) with an aerodynamic diameter less than 10 microns will cause serious damages to human health. Moreover, their presence can have a critical impact on climate change, global warming, and earth radiance budget. Therefore, obtaining precise information about their concentration and spatial distribution is crucial for public health and environmental studies. High concentration of PM10 can be named as a major environmental and public health problems especially for industrial and populated cities around the world. Thus, policymakers and environmental organizations have decided to establish pollution station to measure various pollutants including PM10. Obviously, it is not possible to establish many pollution stations based on economic justifications so, an only limited number of these instruments are located in every city. However these instruments can measure and record the PM10 concentration with high precision, they only provide sparse point observations. In this case, remote sensing data can be utilized to fill this gap and solve the existing discontinuity problem. Generally, two kinds of remote sensing data which have a good representation of existing pollutant in the atmosphere can be used for this purpose. Aerosol optical depth (AOD) and aerosol contributions to apparent reflectance (ACR) are two of these data. ACR images can be simply calculated from each satellite image consisting of Red and SWIR (2.1 µm) bands. This could be achieved by estimating the surface reflectance (SR) of the Red band from the top of atmosphere reflectance (TOAR) of the SWIR band. Then, the difference of SR and TOAR of the Red band can be a representation of the amount of atmosphere reflectance related to existing pollutants. In this study, we have used ACR images instead of AOD data to estimate PM10 concentrations and produce PM10 pollution maps for Tehran city in Iran based on three reasons. First, they have better spatial resolution and second, they are spatially continuous in contrast to AOD data which include much gaps in the study area due to dark target limitations for AOD value retrieval. Lastly, an aerosol robotic network (AERONET) station is not located in this area which is required to evaluate the precision of retrieved AOD values. MODIS level-1B images named MOD02HKM for 8 days in 2017 with corresponding ground measurements of PM10 concentrations from 14 pollution stations have been utilized in this area. Four different regression model including linear, exponential, logarithmic, and power regressions are employed to estimate PM10 concentrations and produce pollution map. Three criteria of R square, the correlation between estimated and observed (measured) PM10 concentrations, and root mean square error (RMSE) are employed to investigate the performance of four regression models. Based on the R square criterion, the linear regression model with 0.5912 performs better than exponential, logarithmic and power regressions with 0.5826, 0.5808, and 0.5782 R square values respectively. Since we have observed different performance from four regression model based on three evaluation criterion, we have applied a ranking method based on the evaluation criterion to determine the best regression model. Based on the ranking, we recognize that the exponential regression model performs better than linear, logarithmic and power regressions.