基于贝叶斯网络的生态系统服务空间格局优化——以泾河流域为例

城市高速发展引起了区域土地利用格局的改变,不仅影响生态环境质量,而且还对生态系统服务的空间格局产生影响,在此基础上优化生态系统服务显得至关重要。在泾河流域2000—2020年净初级生产力（Net primary productivity, NPP）、农业生产力、土壤保持和产水服务空间评估的基础上,将贝叶斯网络和生态系统服务相结合,在关键变量子集和可视化的最优状态子集的基础上,评估了4种生态系统服务需要优化的区域,为区域经济和生态和谐发展提供参考。结果表明：（1） 水文评价模型（Soil and water assessment tool, SWAT）模型能较准确地模拟区域的径流量。通过模拟值和观测值对比分析,该模型具有较高的决定性系数（R²>0.6）和纳什效率系数（NSE>0.5）,可为进一步评估产水服务提供保障。（2） 2000—2020年泾河流域4种生态系统服务的时空差异性较为显著。在时间尺度上,4种生态系统服务均呈现波动中上升的趋势,在空间尺度上呈现较为稳定的变化趋势。（3） 通过对4种生态系统服务优化区域进行叠加分析,发现综合优化区域集中在彭阳县的中部和西南部以及环县的零星区域。研究结果对指导优化区域进行生态系统的可持续管理以及改善生态系统的退化状况具有重要意义。

Spatial pattern optimization of ecosystem services based on Bayesian networks: A case of the Jing River Basin

Changes in regional land use patterns due to rapid urbanization affect the quality of the ecological environment and the spatial pattern of ecosystem services. Therefore, optimizing ecosystem services is crucial. Using the spatial assessment of net primary production, crop production, soil conservation, and water yield services in the Jing River Basin, northwest China from 2000 to 2020 as the basis, this study combined Bayesian networks and ecosystem services and employed a subset of key variables and a visualized subset of optimal states to determine the areas necessitating the optimization of the aforementioned ecosystem services for regional economic and ecological harmonious development. Results showed that (1) the SWAT model can accurately simulate the regional runoff and generate a high coefficient of determination (R²>0.6) and a Nash-Sutcliffe efficiency coefficient (NSE>0.5) when comparing simulated and observed values, thus providing a guarantee for the further assessment of water yield services. (2) The spatial and temporal variations of the four ecosystem services in the Jing River Basin from 2000 to 2020 were substantial. All four ecosystem services showed an upward fluctuation trend on the temporal scale and a stable trend on the spatial scale. (3) Overlay analysis of the optimized areas for the four ecosystem services revealed that the integrated optimized areas were concentrated in the central and southwestern parts of Pengyang County, Ningxia Province and the scattered areas of Huan County, Gansu Province. This study is important for guiding region optimization for the sustainable management and reducing the degradation of ecosystems.