Abstract: In 2020, China explicitly put forward the goals of “carbon peaking” by 2030 and “carbon neutrality” by 2060. Urban green space is the only direct and natural carbon sink in cities, so how to estimate the carbon storage of urban green space is very important for achieving the carbon peaking and carbon neutrality goals. Taking the urban green space within the Fifth Ring Road of Haidian District, Beijing as an example, this research, with GF-2 remote sensing data as information source, selects 139 sample sites of park green space, protective green space, affiliated green space and regional green space by stratification for carbon storage estimation research. It is found that different types of sample sites are significantly different in both carbon storage value and normalized difference vegetation index (NDVI). In view of this, this research further builds a fitting model through regression analysis for NDVI and carbon storage of the four types of green spaces mentioned above, and selects another 40 test samples and, with artificially identified carbon storage data, examines the rationality of the regression model, with a view to building a perfect urban green space carbon storage estimation system. The estimation results show that the total amount of carbon storage of urban green space within the Fifth Ring Road of Haidian District is about 41,400 tons, and carbon sink capacity is different for different green space types, which can be expressed as follows: park green space > affiliated green space > regional green space > protective green space. This research is of great significance to guide the estimation of carbon storage of urban green space and realize carbon neutrality in various cities across the country.