<p>Current research on road freight carbon emissions (RF-CO₂) lacks fine-scale analysis within cities and often overlooks the temporal heterogeneity of driving factors. To address these gaps, the main purpose of this study is to develop a high-resolution RF-CO₂ inventory and systematically investigate its spatiotemporal dynamics and determinants. To achieve this, we employ a bottom-up framework combined with massive heavy-duty truck (HDT) trajectory data to construct the emission inventory, and apply the Geographically and Temporally Weighted Regression (GTWR) model to capture the spatiotemporal heterogeneity of factor influences. The key findings reveal significant spatial clustering, with less than 16% of subdistricts generating 60% of total emissions. Moreover, the GTWR model effectively captures hourly variations and spatial non-stationarity, demonstrating that coefficients of key variables such as primary road density and economic level change substantially over time and space. The main contributions of this study lie in providing a nuanced understanding of RF-CO₂ drivers and enabling the proposal of targeted, spatiotemporally explicit carbon-reduction strategies for sustainable freight transportation.</p>