文献类型：期刊文献

中文题名：基于气象因子的森林病虫鼠害发生率预测模型研究

英文题名：Prediction of Incidence Rate of Forest Diseases, Mice and Pests Using Meteorological Factor

作者：张乃静[1] 鞠洪波[1] 纪平[1]

第一作者：张乃静

机构：[1]中国林业科学研究院资源信息研究所

年份：2012

期号：4

起止页码：96-101

中文期刊名：林业资源管理

外文期刊名：Forest Resources Management

收录：北大核心:【北大核心2011】；

基金：国家科技基础条件平台建设项目(2005DKA32200);Tropical Forest Fire Monitoring and Management System Basedon Satellite Remote Sensing Data in China[ITTO PD 228/03 Rev.3(F)]

语种：中文

中文关键词：气象因子;森林病虫鼠害;预测模型

外文关键词：meteorological factor, forest diseases, mice and pests, prediction model

分类号：S443;S764.5

摘要：基于气象因子,使用一元线性回归、多元线性回归、一元非线性回归以及BP神经网络4种不同的回归模型对森林病虫鼠害发生率进行预测,结果表明:对于线性模型,多元线性回归模型的判定系数和均方根误差均优于一元线性回归模型;对于非线性模型,BP神经网络模型的判定系数和均方根误差均优于一元非线性回归模型;按优劣排序为BP神经网络模型、一元非线性回归模型、多元线性回归模型和一元线性回归模型。气象因子与森林病虫鼠害发生率的关系并非单纯的线性关系,非线性的预测模型可以更好地解释森林病虫鼠害的发生程度。
Based on some meteorological factors, prediction was made on the incidence rate of forest dis- eases, mice and pests by simple linear regression （SLR）, multiple linear regressions （MLR） , simple non - linear regression （SNLR）and BP neural network （BP）. The results show that the determination coefficient （ R2 ） and root mean square error（RMSE） of MLR are better than that of SLR,the R2 and RMSE of BP are better than that of SNLR. According to the quality of models, BP model is the best of all, and SNLR is bet- ter than MLR and SLR. The conclusion is that the relations between meteorological factors and incidence rate of forest diseases, mice and pests are not linearity, non - linear model could effectively explain the in- cidence rate of forest diseases, mice and pests.