广义可加模型在R中的快捷实现及蓝藻水华预测分析中的应用  

Quick implementing of generalized additive models using R and its application in blue-green algal bloom forecasting

作　　者：邓建明[1,2] 秦伯强[1] 王博雯[3] 

机构地区：[1]中国科学院南京地理与湖泊研究所、湖泊与环境国家重点实验室,南京210008 [2]中国科学院大学,北京100049 [3]南京师范大学生命科字学院,南京210046

出　　处：《生态学杂志》2015年第3期835-842,共8页Chinese Journal of Ecology

基　　金：中国科学院对外合作重点项目(GJHZ1214);国家自然科学基金重点项目(41230744);中国科学院南京地理与湖泊研究所一三五重点项目(NIGLAS2012135003)资助

摘　　要：广义可加模型(generalized additive models,GAM)适用于响应变量与解释变量之间的关系是非线性或非单调的数据分析,近年来在生态学中受到越来越多的关注。本文利用太湖湖泊生态系统研究站的监测数据,在RStudio集成编程环境下通过R对数据进行预处理、确定连接函数、对模型进行筛选以及评估等步骤,运用GAM分析了环境因子对微囊藻生物量的影响。结果表明,水温、总磷、化学需氧量以及电导率是影响太湖微囊藻生物量的4个关键环境因子。分析过程中,通过自编函数能有效地减少编程过程中的代码输入工作;R+RStudio是高效、快捷的编程环境。Generalized additive models （GAM） is apphcable to analyze nonlinear and non-mono- tonic relationships between the response variable and explanatory variables. Increased attention has been paid to GAM in ecology study recently. In this study, long-term monitoring data from Lake Taihu were used to analyze the effect of environmental factors on Microcystis biomass. Data screening, link function confirmation, model selection and assessment were carried out by R. The results indicated that water temperature, total phosphorus, chemical oxygen demand and electri- cal conductivity were the main environmental factors that affected Microcystis biomass in Lake Taihu. The use of self-coding functions can be effective in reducing code numbers. In addition, R ＋RStudio might be the optimal programming environment.

关 键 词：可加模型 R语言 RStudio 生物统计 回归分析 

分 类 号：X832[环境科学与工程—环境工程] X52