期刊
ECOLOGICAL INFORMATICS
卷 10, 期 -, 页码 37-48出版社
ELSEVIER
DOI: 10.1016/j.ecoinf.2012.03.007
关键词
Missing data; Multiple imputation; Logistic regression; Cyanobacteria bloom
类别
资金
- National Major Science and Technology Programs [2008ZX07102-001]
Blue-green algae (BGA) bloom is a typical phenomenon in eutrophied lakes. However, up to now, no environmental mechanism has been commonly accepted. Systematic and complete data sets of BGA blooms and environmental factors without any missing data are rare, which seriously affected previous studies. In this study, a bootstrapping based multiple imputation algorithm (EMB) was first applied to reconstruct a complete data set from the available data set with missing data, hence forming a basis for quantitatively relating BGA bloom to contributing factors. Then, the probability of RCA bloom outbreak was simulated using a binomial (or binary) logistic regression model, which is an effective tool for recognizing key contributing factors. The results suggest that 1) the outbreak frequency or probability of BGA bloom tends to first increase and then decrease with a turning point between June and September each year: 2) air temperature, relative humidity, and precipitation were significant positive factors correlated with outbreak frequency, whereas wind speed and the number of sunshine hours were negative factors: 3) water temperature had a strong positive effect on the probability of BGA bloom outbreak, whereas other water quality factors, such as concentrations of organics and nutrients, were not so significant. However, water quality factors, such as NO3-N, SD, pH, NH4-N. COD and DO, still need to be concerned, which had a potential to aggravate the outbreak of BGA bloom in Dianchi Lake, if they were out of control. (c) 2012 Elsevier B.V. All rights reserved.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据