Estimated seasonal patterns of PM2.5 concentrations in Novi Sad using previously developed LUR models for winter and summer

For the first time in Serbia, two Land Use Regression (LUR) models were developed to estimate seasonal PM2.5 concentrations during winter and summer, specifically for Novi Sad. Based on these models, two prediction maps were generated to map and visualize the distribution of estimated seasonal variation of PM2.5 during heating and non-heating season on unsampled locations in Novi Sad. The estimation efficiency of winter and summer models was 55% and 40.3%, using two explanatory variables: total road length within a 50 m buffer (for both models) and total suburban area within a 3.000 m buffer (winter model). According to predictive map for winter, higher PM2.5 concentrations are observed around Novi Sad's urban core and suburban areas, where a significant portion of the population resides in residential houses and buildings. The estimated seasonal PM2.5 concentrations in Novi Sad ranged from 26.37 to 65.09 μg/m³. The lowest concentrations were found in the urban core, likely due to district heating, in contrast to suburban areas with mixed heating systems. Traffic intensity, as reflected by the 50 m road buffer, caused noticeable fluctuations, particularly in the areas like Klisa, Sajlovo, Shaingai and parts of Petrovaradin and Alibegovac. In summer, PM2.5 concentrations ranged from 12.43 to 16.09 μg/m³, with traffic patterns influencing concentrations, especially in the areas like Detelinara, Telep, and Petrovaradin. The predictors used as proxies for traffic intensity and domestic heating areas provided a solid foundation for developing relevant PM2.5 prediction surfaces