基于聚类分析和主成分分析的长江下游稻田氮磷监测指标筛选  

Optimization of Monitoring Indexes of Nitrogen and Phosphorus in Rice Fields in the Lower Reaches of the Yangtze River Based on Cluster Analysis and Principal Component Analysis

作　　者：胡冰涛 陈玉东 HU Bing-tao;CHEN Yu-dong(Nanjing Institute of Environmental Sciences,Ministry of Ecology and Environment,Nanjing 210042,China)

机构地区：[1]生态环境部南京环境科学研究所,江苏南京210042

出　　处：《生态与农村环境学报》2020年第11期1495-1504,共10页Journal of Ecology and Rural Environment

基　　金：国家重点研发计划(2017YFD0801300)。

摘　　要：于2018年对上海市青浦区现代农业园内全水稻生育期内稻田田面水以及地下30和60 cm深度淋溶水进行采样,获取了田面水和淋溶水氮磷指标数据。采用主成分分析法(principal component analysis,PCA)和聚类分析法(cluster analysis,CA)筛选出该区域水稻田田面水和淋溶水中氮磷监测指标,并分别建立该区域水稻田氮磷监测指标的最小数据集(minimum data set,MDS),比较基于聚类分析的最小数据集水质综合得分(water quality index-CA,WQI-CA)、基于主成分分析的最小数据集水质综合得分(water quality index-PCA,WQI-PCA)和全量数据集水质综合得分(water quality index-total data set,WQI-TDS)。结果表明:(1)应用主成分分析和聚类分析均可从稻田田面水以及30和60 cm深度淋溶水的8项氮磷监测指标中筛选出3~4项组成监测指标最小数据集;(2)基于不同数据集的稻田水质综合得分结果差异明显,水质综合得分变化范围和均值表现为WQI-PCA较WQI-CA更接近于WQI-TDS,并且WQI-PCA与WQI-TDS的Nash有效系数和相关度均高于WQI-CA,这表明基于主成分分析的最小数据集(MDS-PCA)较基于聚类分析的最小数据集(MDS-CA)更适合替代全量数据集作为稻田水质污染的监测与评价指标。Pollution from nutrients in rice production can affect water quality.To generate a nitrogen and phosphorus index for the surface water and leaching water from rice paddy fields in the lower reaches of the Yangtze River,samples were collected in the modern agricultural park of the Qingpu District,Shanghai in 2018 over the whole period of paddy growth.Leaching water was sampled at depths of 30 and 60 cm.Principal component analysis(PCA)and cluster analysis(CA)were used to optimize the monitoring indexes and to establish the minimum data set(MDS)for the monitoring indicators of nitrogen and phosphorus in rice paddy fields in this region.The comprehensive scores of water quality index(WQI)of the MDS and the comprehensive scores of water quality index of the total data set(WQI-TDS)based on different analysis methods were compared.The results show that:(1)a minimum data set composed of 3-4 indicators could be selected from eight monitoring indexes of the surface water and leaching water at depths of 30 and 60 cm in paddy fields by principal component analysis and cluster analysis.(2)The comprehensive scores of water quality in paddy fields based on different data sets were significantly different.The range and mean value of comprehensive scores of water quality show that WQI-PCA was closer to WQI-TDS than WQI-CA,and the Nash effective coefficient and correlation of WQI-PCA and WQI-TDS were higher than WQI-CA.This indicates that MDS-PCA is more suitable than MDS-CA to replace WQI-TDS as the monitoring and evaluation index for rice paddy water pollution.

关 键 词：田面水 淋溶水 主成分分析 聚类分析 最小数据集 

分 类 号：X832[环境科学与工程—环境工程]