China’s residential energy consumption has been rising due to the country’s fast urbanization and economic development. As a result, the precise measurement of residential carbon dioxide emissions (CE) is crucial for reducing greenhouse gas emissions and can serve as a foundation for the adoption of carbon reduction laws in urban areas. The purpose of this work is to develop a method that integrates fine-grained features of land use on a grid and nighttime light intensity (NLI) to estimate urban residential CE. First, the population and nighttime lighting data are used to depict the fine-grained land use features; second, three different unmixing models are built to obtain the NLIs of various land use types as well as the lighting values of residential areas; and third, the residential CE was estimated using the method of integrating NLI with the fine-grained land use features of the grid, and a comparative test was conducted. The study’s findings indicate that (1) there is a strong positive linear association between the total amount of lights in residential areas and the residential CE of Guangzhou inhabitants, with a fitted R2 of 0.9318 at a 95% confidence probability. (2) From 2014 to 2022, Guangzhou residents’ residential CE clearly displayed a growing tendency and a more pronounced clustering effect. (3) Residential CE can be estimated more accurately and precisely reflect the differences between different locations when it is based on fine-grained land use features and NLI. On the other hand, some residential regions’ residential CE may be overestimated in the contrasting spatial visualization results, which are less likely to accurately reflect the variability of high CE locations (CE more than 2131t). The research findings can provide a solid database for future investigations, assisting the departments in developing more detailed environmental management and differentiation plans.