基于Logistic回归的中稻推荐施肥模型  

Model on Middle-season Rice Recommended Fertilization Based on Logistic Regression

作　　者：苏荣瑞[1] 金卫斌[1] 艾天成[1] 吴以学 王向平 

机构地区：[1]长江大学农学院,湖北荆州434025 [2]洪湖市农业局,湖北洪湖433200

出　　处：《湖北农业科学》2009年第7期1594-1598,共5页Hubei Agricultural Sciences

基　　金：湖北省自然科学基金创新群体项目(2006ABC013)

摘　　要：以湖北省洪湖市测土配方施肥项目2006、2007年中稻肥料效应试验结果为数据源构建Logistic逐步回归模型,得到目标产量≥6 750 kg·hm-2、≥7 200 kg.hm-2和≥7 500 kg·hm-2时的Logistic模型,模型中引入有效磷、速效钾养分等级和氮肥用量为自变量,3个回归模型的显著性和各参数效应均满足了显著性检验(P<0.05)。由概率预报模型可计算出不同土壤养分等级和不同氮肥用量下的中稻产量≥6 750kg.hm-2、≥7 200 kg·hm-2和≥7 500 kg·hm-2的概率分布,可为当地施肥决策提供科学依据。In this research forecasting models based on logistic regression were made for middle-season rice recommended fertilization at three levels of goal yield ≥6 750 kg.hm^-2, ≥7 200 kg-hm^-2 and ≥7 500 kg-hm^-2, in accordance with middleseason rice fertilization response experiments carried out from 2006 to 2007 in the Honghu city of Hubei province. The categorized level of soil available P, available K and nitrogenous fertilizer amount were introduced as independent variable into this forecasting model, which all pass the significance examination （P〈0.05）. Through the forecasting probability model, the probability distribution for local middle-season rice recommended fertilization could be created.

关 键 词：LOGISTIC回归 推荐施肥 中稻 

分 类 号：S344.1[农业科学—作物栽培与耕作技术] O212.1[农业科学—农艺学]