摘要
富营养化是湖泊沼泽化进程加快的重要因素之一,而湖泊富营养化的预防与治理已经成为世界性的难题。湖泊富营养化评价是湖泊治理的基础,它可以为治理工作提供科学的依据。该文以乌梁素海为例,采用主成分——SOM人工神经网络耦合模型进行富营养化评价。先建立富营养化评价指标体系,然后用主成分分析剔除存在相关性、信息重叠的指标,再将利用主成分分析得到的具有代表性的主成分指标代替原来的评价指标,输入到自组织特征映射网络模型中,最后对富营养化状况进行聚类分析。所得结果与实际相吻合。该方法能根据实测资料对湖泊富营养化状况客观地分类并计算出评价权值,避免了选取评价指标时的主观随意性。
This paper concerns eutrophication assessment of lakes, taking Wuliangsuhai Lake as an example and using a new model （PAC-SOM combined model） to evaluate its eutrophic status. The model relates to the index system of eutrophication estimate established at first and then the principal component analysis to eliminate the indexes having the relativities and overlap information. The representative indexes obtained were imported into SOM model and finally clustering analysis was done. It is concluded that the combined method can more objectively present classification and assessment weights of eutrophication assessment of lakes.
出处
《环境科学与技术》
CAS
CSCD
北大核心
2009年第3期173-177,共5页
Environmental Science & Technology
基金
国家自然科学基金项目资助(50569002、50669004)
内蒙古自然基金项目资助(200711020604)
关键词
富营养化
主成分分析
自组织特征映射网络
聚类分析
eutrophication
principal component analysis
self-organizing feature map
clustering analysis
