使用自组织特征映射神经网络对湖泊水生态功能分区

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

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 hm^(2),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.

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 selforganzing 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 hm^(2),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.