投稿中心

审稿中心

编辑中心

期刊出版

网站地图

友情链接

引用本文:郭正强,严平川,向宣好,鲍仲涛.藻类生长预测模型的比较研究——以洪湖水体为例.湖泊科学,2022,34(4):1140-1149. DOI:10.18307/2022.0408
Guo Zhengqiang,Yan Pingchuan,Xiang Xuanhao,Bao Zhongtao.Comparative study on algae growth prediction models-A case study of Lake Honghu. J. Lake Sci.2022,34(4):1140-1149. DOI:10.18307/2022.0408
【打印本页】   【HTML】   【下载PDF全文】   查看/发表评论  【EndNote】   【RefMan】   【BibTex】
←前一篇|后一篇→ 过刊浏览    高级检索
本文已被:浏览 321次   下载 260 本文二维码信息
码上扫一扫!
分享到: 微信 更多
藻类生长预测模型的比较研究——以洪湖水体为例
郭正强1, 严平川2, 向宣好2, 鲍仲涛2
1.长江大学资源与环境学院, 油气地球化学与环境湖北省重点实验室, 武汉 430100;2.湖北省荆州市水文水资源勘测局, 荆州 434000
摘要:
根据洪湖2014—2019年水质及藻类监测数据,运用综合营养状态指数法评价了丰、平、枯3个时期的营养状态.在此基础上运用逐步回归分析法确定影响藻类生长的显著因子,并根据不同水量不同营养状态细分9种情形对藻类生长做回归预测分析,同时运用BP神经网络模型对回归预测的结果进行比较验证.结果表明:洪湖丰、平水期以蓝藻门为主,枯水期以硅藻门为主;湖泊的营养状态处于中度富营养与轻度富营养之间.分析各时期藻种生物量与影响因子的相关性,发现丰水期控制因子有水温、CODMn和透明度;平水期和枯水期控制因子有水温、总氮、总磷.以2014—2018年数据逐步回归分析得出枯水期+中营养和枯水期+轻度富营养决定系数较低,其余7种时期决定系数均在0.5以上,说明逐步回归并不适用于所有时期.使用2014—2018年的数据进行神经网络训练和验证,2019年的数据进行预测,比较BP神经网络与逐步回归的均方根误差发现全年预测时BP神经网络效果更好;枯水期+中营养和枯水期+轻度富营养逐步回归效果较好,逐步回归的均方根误差仅为1600~4000;丰水期和平水期2种方法预测效果相当.合理地选择预测模型能为湖泊水华做出预警,控制显著变量可以达到防治水华污染的效果.
关键词:  富营养化  藻类藻相  逐步回归  BP神经网络  洪湖
DOI:10.18307/2022.0408
分类号:
基金项目:湿地生态与农业利用教育部工程研究中心项目(KF201917)资助.
Comparative study on algae growth prediction models-A case study of Lake Honghu
Guo Zhengqiang1, Yan Pingchuan2, Xiang Xuanhao2, Bao Zhongtao2
1.Hubei Key Laboratory of Petroleum Geochemistry and Environment, College of Resources and Environment, Yangtze University, Wuhan 430100, P. R. China;2.Hubei Jingzhou Survey Bureau of Hydrology Resources, Jingzhou 434000, P. R. China
Abstract:
According to the monitoring data of water quality and algae in Lake Honghu from 2014 to 2019, the comprehensive trophic level index method was used to evaluate the nutritional status in the three periods of wet season, normal season and dry season. On this basis, stepwise regression analysis was used to determine the significant factors affecting the growth of algae, and regression prediction analysis was conducted on the growth of algae in nine situations according to different water amounts and different nutritional status. At the same time, BP neural network model is used to compare the results of regression prediction. The results showed that Cyanophyta was the dominant species in wet season and normal season in Lake Honghu, and Bacillariophyta was dominant in the dry season. The nutritional status of the lake is between moderate and mild eutrophication. The correlation between the number of algal cells and water temperature, total nitrogen, total phosphorus, CODMn and other influencing factors in each period was analyzed. The control factors in the wet season were water temperature, CODMn and transparency. The control factors were water temperature, total nitrogen and total phosphorus in normal and dry seasons. The stepwise regression analysis based on the data from 2014 to 2018 showed that the determination coefficients of dry season + medium nutrition and dry season + mild eutrophication were low, and the determination coefficients of the other seven periods were all above 0.5, indicating that the stepwise regression was not suitable for all periods. The data from 2014 to 2018 were used for neural network training and verification. The data from 2019 were used for prediction. The root mean square error of BP neural network and stepwise regression was compared. It was found that the BP neural network was better in the whole year prediction. The root mean square error of stepwise regression of dry season + medium nutrition and dry season + mild eutrophication was 1600-4000. Stepwise regression analysis is better and the prediction effects of the two methods in wet season and water season are similar. Selecting reasonable prediction models can make early warnings of lake algal bloom and help control significant variables to achieve the effect of preventing and controlling algal bloom pollution.
Key words:  Eutrophication  algous facies  stepwise regression  BP neural network  Lake Honghu
分享按钮