秋末石榴冻害气象预测预警模型的构建  

Construction of Meteorological Forecast and Early Warning Model for Pomegranate Freezing Injury in Late Autumn

作　　者：李全景 安广池[2] 段宝敏 崔云鹏 王倩 胡园春 LI Quanjing;AN Guangchi;DUAN Eaomin;CUI Yunpeng;WANG Qian;HU Yuanchun(Zaozhuang Meteorological Bureau,Zaozhuang Shandong 277800;Zaozhuang Agricultural Technology Promotion Center,Zaozhuang Shandong 277800;Yicheng District Meteorological Bureau,Yicheng Shandong 277300)

机构地区：[1]枣庄市气象局,山东枣庄277800 [2]枣庄市农业技术推广中心,山东枣庄277800 [3]枣庄市峰城区气象局,山东峰城277300

出　　处：《山东林业科技》2023年第2期34-38,共5页Journal of Shandong Forestry Science and Technology

基　　金：“十三五”山东重大气象工程项目“山东现代农业气象服务保障工程”(鲁发改农经[2017]97号);山东省气象局气象科研技术研究项目(2017sdqxm13);山东省气象局气象科研技术研究项目(2019sdqxm15);枣庄市科技发展计划项目(2020ns04);国家科技部星火计划项目(2012GA740012);山东省农业科技成果转化资金项目(鲁科农字[2012]56号);山东省农业良种工程重点课题(鲁科农字[2012]213号)。

摘　　要：冻害问题一直是制约石榴发展的重要因素,分为初春冻害与秋末冻害两种典型的冻害模式。Logistic回归为概率型非线性回归模型,是研究观察结果与其影响因素之间关系的一种多变量分析方法。本文对山东枣庄35年的气象资料进行统计分析,研究了石榴秋末冻害与最低气温、降温幅度、<0℃持续时间、稳定通过11℃日期这4个气象因子之间的关系,采用Logistic回归建立了各气象因素与石榴秋末冻害的相关预测模型。结果表明:对于1981—2015这35年中发生冻害的16年进行模型判别,只有其中2年被模型预测为无冻害,正确率为87.5%;对于不发生冻害的19年中,利用上述模型判断,其中有3年预测为发生冻害,则对于不发生冻害预判正确率为84.2%,总体判断正确率为85.97%。对2016—2019年预测结果正确率86.7%,为枣庄地区石榴秋末冻害预测与防治提供了重要依据。Freezing damage has always been an important factor in restricting the development of pomegranate and it is divided into two typical modes,including freezing damage in early spring and freezing damage in late autumn.Logistic regression which is a probabilistic nonlinear regression model,is a multivariate analysis method to study the relationship between observation results and its influencing factors.This paper conducts a statistical analysis on the meteorological data of Zaozhuang City,Shandong Province for 35 years,and studies the relationship between the freezing damage of pomegranate in late autumn and four meteorological factors o£minimum temperature,cooling range,duration of temperature below 0℃,and duration of temperature at or above 11℃stably.In this paper,logistic regression is used to establish the correlation prediction model of various meteorological factors and the freezing damage of pomegranate in late autumn.16 years of freezing damage occurring during 35 years from 1981 to 2015 are distinguished by the model.Only 2 of them are predicted to be free of freezing damage by this model,and the prediction accuracy rate is 87.5%..Other 19 years without freezing damage are also judged by this model.Among them,3 years are predicted as freezing damage by this model,and the prediction accuracy rate of no freezing damage is 84.2%.The overall prediction accuracy rate is 85.97%.The accuracy rate of 2016-2019 forecast results is 86.7%.Logistic regression model provides an important basis for the prediction and control of freezing damage of pomegranate in late autumn in Zaozhuang area.

关 键 词：石榴 秋末冻害 气象预测 LOGISTIC回归模型 

分 类 号：S665.4[农业科学—果树学]