Demirarslan, Kazım OnurYener, İsmet2022-12-162022-12-162022Demirarslan, K. O., & Yener, İ. (2022). Investigation of total suspended particulate matter dispersion from quarries in Artvin, Turkey, using AERMOD and its relationship with topography. Air Quality, Atmosphere & Health, 15(12), 2313–2327. ‌https://doi.org/10.1007/s11869-022-01253-5https://hdl.handle.net/11494/4297As the number of quarries, one of the sectors providing raw material to the construction industry, increases daily, their adverse effects on ecosystem health rise, with a particularly negative impact on human health due to the air pollution they create. Air quality monitoring networks throughout the country, which are difficult to set up because of some economic and political issues, are necessary for precise air quality measurements. Therefore, more accurate and special modeling programs are gaining importance. The present study aims (i) to model total suspended particulate matter (TSP) emissions from quarries in Artvin, Turkey, using AERMOD, and (ii) to determine the contribution of topography to that distribution using multilinear regression (MLR) and multilayer perceptron neural network (MLP) models. TSP data from seven quarries in Artvin were used for this purpose. According to the dispersion model results, the hourly maximum TSP values were 9.87 and 75.17 µg/m3. First, Pearson correlation coefficients between TSP and topographic variables were calculated to determine the relations between topography and TSP, and then MLR and MLP models were developed. While 25 to 54% of the variance in TSP was explained by MLR, MLP explained 65 to 87% of that. MLP models with a higher adjusted coefficient of determination (R2adj) and lower root mean square error (RMSE), mean absolute error (MAE), Akaike information criterion (AIC), and Bayesian information criterion (BIC) outperformed MLR. The most influential variables on TSP dispersion were elevation, distance to the source, the rate of forest area, and easting (longitude).eninfo:eu-repo/semantics/closedAccessModelingMultilayer perceptronMultilinear regressionTopographyTotal suspended particulate matterArticle15122313232710.1007/s11869-022-01253-5N/AQ2