To effectively evaluate the carbon footprint of urban ecosystems and appropriately manage urban carbon budget, it is crucial to investigate how urbanization affects the spatial pattern of carbon source/sink of green space. This study employed net ecological productivity (NEP) as an indicator for carbon source/sink, evaluating by the integration of net primary productivity and soil respiration. Then we formed the carbon source/sink distribution of urban green spaces, and analyzed the impact of the urban-rural gradient on carbon source/sink levels across different types of green spaces in Hangzhou City. Following the landscape pattern analysis with the land use data using Fragstats, we applied multiple linear regression and stepwise analysis to pre-examined the effects of landscape, vegetation, and microclimatic factors on NEP. A generalized additive model was then employed to analyze the relationship between NEP and each factor. Furthermore, the different performances of these factors across the urban-rural gradient and greenspace types from the same statistic process were compared. The results showed that there were significant variations of NEP distribution and its influencing factors across green space types and urban-rural gradient. From 2019 to 2022, the whole carbon budget of main urban area in Hangzhou was carbon source, with annual average NEP -0.277 kg C m^-2 a^-1 carbon sink greenspace mostly located at the western part of Hangzhou, while the carbon source greenspace at the central and eastern part. NEP of the entire green space was positively correlated with patch area, tree coverage, and shrub coverage, but negatively correlated with shrub richness and temperature. Furthermore, NEP increased gradually with urban, sub-urban and rural gradient. Urban NEP was positively correlated with tree coverage, but negatively correlated with Shannon's diversity index and temperature. Sub-urban NEP displayed a positive correlation with tree richness and shrub coverage, but a negative correlation with patch density and temperature. Rural NEP showed a positive correlation with the aggregation index, tree coverage and shrub coverage with different driving factors. Aggregation index, tree cover, and shrub cover were all positively connected with park's NEP, while negatively correlated with temperature, shrub richness, and landscape division index. Farmland's NEP had a negative correlation with temperature and a positive correlation with the aggregation index and shrub coverage. The natural vegetation's NEP had a positive correlation with tree cover and a negative correlation with both temperature and Shannon's diversity index. Our findings highlighted the impact of urbanization on the carbon source/sink of green space, and provided theoretical and empirical support for differentiated managements of urban and rural carbon budget.