Mapping direct carbon emissions at high-resolution in urban environments could help in the development of measures to mitigate carbon emissions through optimizing the layout of inner structures. It requires the use of a mapping method combining the bottom-up and top-bottom calculations with uncertainty evaluations. This study developed a method for urban scale analyses of carbon emissions, including a theoretical framework of uncertainty distribution and transmission. Using Jinjiang City, China, as a case study, we applied this method to calculate the amount of carbon emissions in grids distributed across a city. This information was used to analyze emission uncertainties and its sources. The calculated emissions were allocated through the accurate spatial identification of three emission sectors and proxy data. Two different population spatialization methods were constructed in order to create 30 m and 500 m resolution grid maps. We designed four different Monte-Carlo simulation scenarios to analyze the uncertainties of the two maps. The results showed that the method developed here was suitable for delineating carbon emissions at the urban-scale. The 30 m resolution map showed that residential emissions were widely distributed, whereas industrial emissions were more concentrated, with the opposite trend being detected in the 500 m resolution map. Calculations of carbon inventory and spatial proxy had more impacts on the 30 m resolution map than on the 500 m resolution map. During the process of spatial superposition, the uncertainties from different sectors showed a nonlinear relationship, which was represented by smaller total uncertainties compared with the sum of uncertainties from the three emission sectors. In conclusion, this study provides important baseline data that could be used to optimize urban form by promoting low-carbon city construction.