中国农业气象 ›› 2016, Vol. 37 ›› Issue (04): 408-414.doi: 10.3969/j.issn.1000-6362.2016.04.004

• 论文 • 上一篇    下一篇

利用BP神经网络模型对太湖水体叶绿素a含量的估算

王雪莲,宋玉芝,孔繁璠,王宇佳   

  1. 1.南京信息工程大学应用气象学院江苏省农业气象重点实验室,南京 210044;2.南京信息工程大学江苏省大气环境与装备协调创新中心,南京 210044
  • 收稿日期:2015-12-28 发布日期:2016-08-10
  • 作者简介:王雪莲(1990-),硕士生,主要研究方向为环境工程。E-mail:wangxuelian_415@163.com
  • 基金资助:

    国家自然科学基金(41471446)

Estimation of Chlorophyll-a Concentration in Taihu Lake by Using Back Propagation (BP) Neural Network Forecast Model

WANG Xue-lian, SONG Yu-zhi, KONG Fan-fan, WANG Yu-jia   

  1. 1.Jiangsu key Laboratory of Agricultural Meteorology, School of Applied Meteorology, Nanjing University of Information Science & Technology, Nanjing 210044,China; 2.Jiangsu Collaborative Innovation Center of Atmospheric Environment and EquipmentTechnology (CICAEET),College of Environmental Science & Engineering, Nanjing University of Information Science &Technology, Nanjing 210044
  • Received:2015-12-28 Published:2016-08-10

摘要:

利用太湖2001-2006年常规水质监测资料和气象资料进行主成分分析,确定影响太湖水体叶绿素a 含量的主要因子。在此基础上构建BP神经网络模型,利用模型对太湖湖心区水体叶绿素a含量进行估算,并对模型进行敏感度分析;将所建模型运用于太湖梅梁湾、贡湖湾、竺山湾以及东太湖4个湖区叶绿素a含量的估算,以验证其适用性。结果表明:基于主成分分析的BP神经网络模型估算的湖心区叶绿素a含量与实测值的拟合度良好,对已建立的BP神经网络预测模型进行敏感度分析表明,气温和溶解氧与浮游植物叶绿素a含量密切相关;该模型对太湖其它4个湖区水体叶绿素a含量的估算结果与实测值拟合度良好, 表明其适用性也较好,因此,可以运用于对太湖水体叶绿素a含量的估算及预测。

关键词: 叶绿素a, 主成分分析, BP神经网络模型, 敏感度分析

Abstract:

To estimate chlorophyll-a concentration in the centre area of Taihu Lake, back propagation (BP) neural network forecast model was constructed based on principal component analysis according to conventional water quality monitoring data and meteorological data in Taihu from 2001 to 2006 and the sensitivity analysis of model was performed. The results showed that in the centre area of Taihu Lake, estimated value of chlorophyll-a concentration according to BP neural network forecast model had a better fit with the measured data of chlorophyll-a concentration. Through sensitivity analysis of established estimation model, it was found that temperature and dissolved oxygen were highly related with the chlorophyll-a concentrations. At the same time, chlorophyll-a concentrations in different areas of Taihu Lake (Meiliang Bay, Gonghu Bay, Zhushan Bay and East Taihu) were estimated by using BP neural network forecast model, close agreement was observed between estimated and the measured data of chlorophyll-a concentration. In general, BP neural network forecast model could be used to estimate and predict the chlorophyll-a concentration of the whole lake in Taihu.

Key words: Chlorophyll-a, Principal component analysis, BP neural network, Sensitivity analysis