Against the backdrop of global climate change, reducing CO₂ emissions from transportation is urgent. This study focuses on Wuhan, integrating multi-source data such as road monitoring, traffic flow, vehicle types, and speeds to build a spatiotemporal carbon emission model using the Random Forest algorithm. Results show the model achieves an R² of 0.74 and RMSE of 26.22 ppm, accurately simulating hourly road CO₂ variations. Emissions exhibit "peak concentration, directional asymmetry, and arterial dependency," with morning peaks averaging 492.46 ppm (peak 804.18 ppm), evening peaks at 488.79 ppm (peak 788.27 ppm), and nighttime lows averaging 477.31 ppm. Enclosed corridors like the East Lake Tunnel show significantly higher CO₂ levels (550–810 ppm), while radial arterials connecting urban cores and peripheries account for over half of total emissions. Inner and second ring roads act as emission hotspots, with concentrations 3.9%–4.5% higher than outer rings. Nighttime emissions drop by 4.7%–5.3%. Tunnels exhibit the highest average CO₂ (620 ppm), 28.8% above other road types, due to restricted exhaust dispersion and high traffic density. These findings highlight the need for optimized urban planning and traffic management.