以海河流域为研究区域，通过 InVEST 模型对 1990—2020 年碳储量状况进行分阶段定量反演，并结合 Markov-PLUS 模型从数量与空间角度进行未来模拟，最终明确流域碳储量发展趋势及关键驱动因素。结果表明：1990— 2020 年海河流域碳储量共减少 1.30 亿t，耕地碳储量损失最大；碳储量高值主要集中在西部与北部山地地区，低值 分布于东南部平原地区。人口密度在各阶段解释力占比均超过 20%，且呈现增加趋势；年均降水对碳储量空间分 异影响最小，最低占 0.21%。各因子相互作用对碳储量空间分异解释力显著提高，人口密度与其他因子交互作用 对碳储量变化的解释力最强。Markov-PLUS-InVEST 预测结果显示，在城市快速发展情景下，2030 年碳储量与 2020 年基本持平，但生态保护情景增加 0.88 亿t。相关结论可为海河流域进行土地类型间的生态调控及未来规划 发展提供理论支撑。

The change of regional land use affects carbon emission and carbon sequestration processes, which in turn causes changes in the carbon cycle and carbon storage in terrestrial ecosystems. Based on the China land cover dataset, the carbon storage and sequestration module of the integrated valuation of ecosystem services and trade-offs model was used to estimate the carbon storage and changes in terrestrial ecosystems from 1990 to 2020 in the Haihe River basin. The patch-generating land use simulation model was combined to predict future land use and carbon storage. The integrated valuation of ecosystem services and trade-offs model could quantify regional carbon storage based on regional carbon density. The carbon storage and sequestration module mainly depended on land cover types and a basic carbon pool model. The basic carbon pool model divided carbon storage into four types: aboveground biomass carbon, belowground biomass carbon, soil carbon, and dead organic matter carbon. The patch-generating land use simulation model is a cellular automata model based on raster data that could simulate land use changes at the patch scale. It integrated the land expansion analysis strategy and a cellular automata model based on multiple random patch seeds, which could be used to explore the driving factors of land expansion and predict the patch-level evolution of land use landscapes. Additionally, geodetector was utilized to quantitatively explain the driving mechanisms of carbon storage in the Haihe River basin. Geodetector is a method for detecting spatial variations of geographic features and revealing their driving factors. This method allows for direct correlation analysis between the dependent variable and the independent variables without considering collinearity among factors.The results showed that:(1)the carbon storage decreased by a total of 4.98% from 1990 to 2020, with the year 2003 as the turning point. Carbon storage exhibited a fluctuating declining trend, followed by a decrease in the magnitude of fluctuations, fluctuating around 2.05 billion tons.(2)The spatial distribution of carbon density in the basin exhibited significant heterogeneity. High carbon density values were mainly concentrated in the eastern and northern forested areas of the basin, while low values were primarily distributed in cities and the Bohai Bay region.(3)In 2030, apart from the scenario of rapid urban development, other scenarios show varying degrees of carbon stock increase. The scenario with the highest increase is the ecological conservation scenario, which has a carbon stock of 0.77 million tons higher than the scenario of rapid urban development. This indicates that ecological improvement is beneficial for carbon sequestration in regional terrestrial ecosystems. In the scenario of rapid urban development, the expansion of impervious surfaces encroached upon cropland, leading to a significant reduction in carbon storage in the eastern plain area of the Haihe River basin.(4)Natural factors had a higher explanatory power than socio-economic factors, and the interaction between population density, DEM, and other climatic factors has the strongest explanatory power for changes in carbon stock. The results will provide certain theoretical support for land regulation and future low-carbon development in the Haihe River basin and also serve as a reference for better implementation of the carbon peaking and carbon neutrality goals.