基于气象条件的赤松毛虫越冬死亡率预报模型  

Forecast Model of Dendrolimus spectabilis Butler Overwintering Mortality Based on Meteorological Conditions

作　　者：张淑杰[1] 张玉书[1] 肖燕 吴海山 武晋雯[1] 纪瑞鹏[1] 刘庆婺[1] 

机构地区：[1]中国气象局沈阳大气环境研究所,沈阳110016 [2]辽宁省林业有害生物防治检疫局,沈阳110001 [3]阜新蒙古族自治县森林病虫害预测预报站,阜新123107

出　　处：《中国农业气象》2013年第1期94-99,共6页Chinese Journal of Agrometeorology

基　　金：中国气象局公益性行业科研专项(GYHY200906028)

摘　　要：利用辽宁省阜新县1976-2012年赤松毛虫越冬死亡率数据及同期气象站的温度、降水量、风速、相对湿度、日照时数等资料,采用相关分析、逐步回归分析、主成分分析与逐步回归相结合等方法,确定影响赤松毛虫越冬死亡率的关键气象因子,并建立赤松毛虫越冬死亡率的预报模型,两种方法建立的模型都通过了0.05水平的显著性检验。结果表明,冬季温度是影响赤松毛虫越冬死亡率的主要因子,其次是日照时数和温湿度的配合情况。仅采用逐步回归分析方法(方法Ⅰ)所建模型得到的拟合值与观测值的相关系数为0.75,D2(越冬死亡率模拟值与实际值相对差值的平方)均值为0.58,而采用主成分与逐步回归相结合方法(方法Ⅱ)建立模型得到的拟合值与观测值的相关系数为0.76,D2均值为0.62。利用两种方法对2009-2012年松毛虫越冬死亡率进行模拟预测,其中2009,2010和2012年方法Ⅰ预测准确率在70%以上,方法Ⅱ的预测准确率在80%以上,表明方法Ⅱ所建模型的拟合及预报准确率均高于方法Ⅰ,但由于两种模型的适用条件不同,在各气象因子数据容易获取的情况下,应选用方法Ⅱ,反之则可选用方法Ⅰ进行预报。Based on overwintering mortality data of the Dendrolimus spectabilis Butler and its relationship with the synchronous temperature, precipitation, wind speed, relative humidity, and sunshine hours in Fuxin county, Liaoning province from 1976 to 2012 ,the main meteorological factors influenced the overwintering mortality of the Dendrolimus spectabilis Butler were determined, and the prediction model of the overwintering mortality of Dendrolimus spectabilis Butler was established, which had passed significance and verification test, by using correlation analysis, stepwise regression analysis, the combination of principal component analysis and stepwise regression. The results showed that temperature was the primary factor for overwintering mortality of Dendrolimus spectabilis Butler, followed by sunshine hours and combined effect of temperature and humidity. The correlation coefficient of stepwise regression method （ method I ） was 0. 75 between fitted values and observed data, and average D2 （ square of relative difference between overwintering mortality simulation value and actual data） was 0.58. The correlation coefficient was 0.76 and the average D2 was 0. 62 of combination of principal component analysis and stepwise regression （ method Ⅱ ）. Compared the simulation data of these two methods from 2009 to 2012 ,the accuracy of method i was above 70% and method Ⅱ was above 80% in 2009,2010 ,and 2012. The results indicated that the forecast accuracy of method Ⅱ was higher than that of method I , but they could be used in different conditions. If it was easy to get the data of meteorological factors, method lI was the better choice, otherwise method I was the alternative.

关 键 词：赤松毛虫 越冬死亡率 影响因素 气象条件 预报模型 

分 类 号：S763.3[农业科学—森林保护学] S716.1[农业科学—林学]