【摘要】 土地利用变化是驱动生态系统服务的重要因素，系统了解其背后的影响机制对生态系统的有效管理和缓解人地矛盾至关重要。本研究以滇中城市群(CYUA)为例，首先使用卷积神经网络-长短期记忆神经网络-元胞自动机(CNN-LSTM-CA)模型和多目标线性规划(MOP)模型模拟了2035年城市群多情景下的土地利用变化空间格局，并结合InVEST模型评估不同情景下5种生态系统服务功能，在此基础上构建生态系统服务贡献指数(ESCI)和耦合协调度(CCD)模型分析土地利用变化对生态系统的影响。结果表明：CNN-LSTM-CA模型能有效提取时空领域特征，整体模拟精度达到0.924 2,与其他模型相比，该模型具有更高的模拟精度和更真实的土地利用分布格局。在生态保护优先情景(EPs)下，除产水量和碳储量服务下降外，其余生态系统服务均呈上升趋势；在经济发展优先情景(EDs)下，碳储量、土壤保持、生境质量和水质净化服务大幅降低；在自然发展情景(NDs)下，生态环境质量整体呈下降趋势。土地利用变化显著影响生态系统服务，建设用地快速扩张以及林地和草地减少对生态系统服务产生了显著的负向影响，研究区土地利用变化与生态系统服务的耦合协调程度整体较高。研究结果有助于揭示土地利用变化对区域生态系统服务时空变化的影响机制，可为城市群国土空间规划和决策提供有效支撑。更多还原

【Abstract】 Land use change plays a pivotal role in driving ecosystem services(ES), and a systematic understanding of the underlying impact mechanisms is of great significance in promoting sustainable development in harmony with both people and the land. The central Yunnan urban agglomeration(CYUA) was selected as an example. Firstly, the convolutional neural network(CNN)-long short-term memory neural network(LSTM)-cellular automaton(CA)(CNN-LSTM-CA) model and multi-objective linear programming(MOP) model were employed to simulate the spatial patterns of land use change in the urban agglomeration under multiple scenarios in 2035. Five ES functions were assessed under these scenarios through integration with the integrated valuation of ecosystem services and tradeoffs(InVEST) model. Building upon this assessment, the ecosystem services contribution index(ESCI) and coupling coordination degree(CCD) model were constructed to quantify the impacts of land use change on ES. Furthermore, the self-organized feature mapping(SOFM) network was applied to identify ES bundles under varying scenarios. The results were summarized as follows. The CNN-LSTM-CA model can effectively extract the spatio-temporal neighbourhood features, with an overall simulation accuracy of 0.924 2, which surpassed other models in simulation accuracy and realism of land use distribution pattern. Under the ecological protection scenario(EPs), all ES exhibited an upward trend, with the exception of a decline in water yield and carbon storage services. In the economic development scenario(EDs), significant decreases were observed in carbon storage, soil retention, habitat quality, and water purification services. The natural development scenario(NDs) demonstrated an overall decline in ecological environment quality. Land use change significantly affected ES, particularly through the rapid expansion of built-up land and the reduction of forest land and grass land had resulting in a notable negative impact. The CCD between land use change and ES in the study area was generally favorable. These results contributed to a comprehensive understanding of the impact mechanisms of land use change on the spatio-temporal changes of ES in the region, offering effective support for territorial spatial planning and decision-making of urban agglomerations.更多还原