This study aimed to clarify the spatiotemporal heterogeneity of carbon stocks in black soil regions and the influencing factors，and reveal how land use change affected carbon storage，so as to enrich carbon stock estimation methods and provide reference for formulating appropriate carbon sequestration policies.

Harbin City，a representative black soil region，was the study area. The land use data and multi-source spatial carbon density data in 2000，2010，and 2020 were employed to reveal the change pattern of carbon stock using time series analysis. The center of gravity migration model，exploratory spatial data analysis，and coldspot/hotspot spatial statistical method were used to characterize spatial differentiation. The geographical detector model was applied to identify the intensity and interaction of influencing factors for carbon stock in the study area.

From 2000 to 2020，cropland and forestland were the dominant land use types in the study area，with pronounced changes occurred，and the areas of land conversion were from cropland and forestland （only second to land conversion from water area），mainly into construction land and unused land. These land use changes shaped the spatiotemporal distribution of carbon stocks，which decreased overall from 6.92×10⁸ t in 2000 to 6.60×10⁸ t in 2020. The center of gravity of carbon stock under land use change shifted 548.61 m southeastward in the study area，with the most pronounced migration distance between 2010 and 2020，reaching 451.36 m. Carbon stock hotspot areas based on land use changes were concentrated and became increasingly expanded，while coldspot areas remained relatively scattered and shrank. Carbon stock under land use change in the study area was commonly influenced by natural environmental and socioeconomic factors，with elevation，land use intensity，distance from county center，and slope exhibiting strong explanatory power. Factor interactions were predominantly characterized by dual-factor enhancement，and interactions between natural and socioeconomic factors explained change in carbon stock more effectively than interactions among factors within a single category.

Between 2000 and 2020，carbon stock under land use change in the study area generally declines，and carbon stock loss is primarily due to the conversion of high-carbon-sink land use types （water areas and forestlands） into low-carbon-sink land use types （unused land，cropland，and construction land），indicating that human-driven land use change has become a key factor influencing carbon stock. Furthermore，carbon stock change shows clear spatial clustering，underscoring the scale-dependent impact of human activities on spatial reorganization of carbon pools. The interaction between natural constraints and human disturbances is a dual-engine dri-ving model for carbon stock change，with elevation，distance from county center，slope，and land use intensity as critical influencing factors of spatial differentiation. Accordingly，it is recommended that cropland retirement should be prioritized in high-risk areas identified by the slope-distance coupling model.