The Simplified Spatial Dispersion Model for PM2.5 Particles in Užice, Serbia

Abstract

Užice, town in a river valley, surrounded by hills, with low wind is air polluted, during heating season. The prevailing wind direction is northwest (169‰), the most frequently in spring (192‰) and least in autumn (96‰). Atmospheric calms dominate (591‰), limiting pollutant dispersion. This study investigates the impact of heavy fuel oil combustion at the central heating plant on PM2.5 concentrations under northwest wind. Simplified Gaussian dispersion equations were developed to model pollutant transport. The model demonstrated good agreement for daily PM2.5 concentrations during December, with deviations within 10%, though lower accuracy occurred under unstable March and April atmosphere. The validation model yielded as follows: Mean Absolute Error 1.26 μg/m3, Root Mean Square Error 9.71 μg/m3, Coefficient of Determination R2 0.92. Fractional Bias is 1.18, Fraction of Prediction within Factor 2 is 0.26 and Normalized Mean Square Error 0.0036. The overall results confirm that the model holds stability, predictive value and provides highly reliable baseline for further analysis. A Geographic Information System with population density and PM2.5 particles use for health risks evaluation. Within 500 m radius (population density: 20,840 inhabitants/km2), PM2.5 exposure was 17.4 ug/m3. Within 300 m, population density is 22,791 inhabitants/km2 and PM2.5 reaches 18.9 ug/m3.