使用自组织特征映射神经网络对湖泊水生态功能分区

以白洋淀淀区为研究案例，在水文、气象、水化学、水生态以及人类活动干扰多要素基础上，耦合生态系统服务空间分布，形成水生态分区指标体系框架，通过自组织特征映射（self-organizing feature map，SOFM）神经网络，并将案例区划分为核心湿地保护区、湿地生态缓冲区、入淀河流缓冲区和生态屏障区4类水生态功能区域，面积分别为9763.81、9538.59、5953.15和5417.53 hm2，分别占白洋淀淀区面积的31.83%、31.10%、19.41%、17.66%．分区结果体现了一定的层次结构与空间特征差异，可为未来科学识别不同区域压力源、淀区精准修复以及差别化水质管理提供科学的数据支撑． 

Eco-functional regionalization of Baiyangdian Lake water system according to self-organizing feature map of neural networks

Under constant pressure for improvement in water quality and restoration of water ecology, rational eco-functional regionalization plays an important role in resources utilization and management promotion of lakes．An index framework for Baiyangdian Lake was developed here aiming at eco-functional regionalization using self-organzing feature map(SOFM), taking into account hydrological, meteorological, chemical indictors, human disturbances, and evaluation of ecosystem services．Consequently, Baiyangdian Lake was divided into 4 functional zones: core protection zone, ecological buffer zone, ecological buffer zone of upstream rivers, ecological barrier zones, each area measured 9763.81, 9538.59, 5953.15, and 5417.53 hm2, accounting for 31.83%, 31.10%, 19.41% and 17.66% of the total lake area respectively．This eco-functional regionalization reflected fully lake spatial heterogeneity．This work will support identification of pressure sources in different regions and enable efficient administration of the lake water quality．