<p>Air pollution is a significant global issue affecting health, climate, and ecosystems, with urban areas especially impacted by emissions from mobile sources. A multitude of factors influences the intensity and spatial distribution of these emissions. Accurately estimating air pollutant emissions spatially is crucial for implementing effective reduction measures. However, previous studies did not focus on classifying emissions and fuel consumption (FC) from mobile sources independently while considering spatial trajectory characteristics. This approach is essential for accurately understanding environmental characteristics and their impacts on mobile source emissions in different locations, as well as for managing local air quality (AQ). The current research focuses on classifying FC and emissions of individual taxis. To achieve this, an approach utilizing Hidden Markov Models (HMM) is employed in binary and multiclass modes. The study area encompasses a segment of the Beijing metropolis in China. Results indicate that HMM performs better in binary classification compared to multi-class classification (three and four classes). For test data, the classification accuracy of HMM in binary mode for FC, nitrogen oxides (NOx), hydrocarbons (HC), and carbon monoxide (CO) stands at 0.80, 0.84, 0.88, and 0.84, respectively. Following model validation, a map detailing FC and emissions from mobile sources under scrutiny is generated. This map serves to identify areas with varying levels of FC and emissions, aiding in visualizing the risk posed by air pollutants from the tested mobile sources. By modeling individual mobile sources, this research sets a foundation for larger-scale studies, potentially offering valuable insights into combating air pollution more effectively.</p>