Using Meteorological Factors In Different Areas And In Previous Years To Establish Annual Occurrence Dynamics Prediction Model Of Pests In Tobacco Fields In Chenzhou

Taking the historical information of Spodoptera litura（Fabricius）and Myzus persicae（Sulzer）in tobacco field in Chenzhou City of Hunan Province for example,the feasibility that establish the annual occurrence dynamics forecasting model of pests by using the meteorological factors in different ares and in previous years is discussed.Collecting and collating the historical meteorological data（such as temperature,humidity,rainfall,sunshine duration,etc.）of Changsha from 2000 until 2013,Chenzhou,Guangzhou and Nanchang from 2000 until 2015 and the historical data of Spodoptera litura（Fabricius）and Myzus persicae（Sulzer）in Chenzhou from 2000 until 2015（such as the annual occurrence quantity of adult and larvae,etc.）,stepwise regression analysis and Chi-square analysis are used to establish effective multi-factor forecasting models.The forecasting models are established and the impact factors are selected based on the data of16-year-long Spodoptera litura（Fabricius）and Myzus persicae（Sulzer）in Chenzhou city,170 meteorological factors of corresponding years in Chenzhou and Changsha,and 104meteorological factors of corresponding years in Guangzhou and Nanchang.According to the regression models（equations）are satisfied with three basic conditions that the significant level of totality or a certain factor P≦0.05,the variance inflation factors VIF≦5,Chi-square test（Karl.Pearson）of the theoretical predicted value and the observed value P ²_（χ）≧0.05,we obtained the results as follows:First,taking Spodoptera litura（Fabricius）for example,62 significant impact factors were obtained and 32 effective models were established.The results of Chi-square test showed that all of the back-substitution predicted results of 32 models were satisfied withχ²<χ²_0.995,in other words,there is no significant difference between the back-substitution predicted values and the observed values.We obtained six groups of Chi-square cumulative value;the first group was satisfied withχ²<χ²_0.995,the second group was satisfied withχ²<χ²_0.950,the fifth group was satisfied withχ²<χ²_0.975,the sixth group was satisfied withχ²<χ²_0.950,the ninth group was satisfied withχ²<χ²_0.250.250 and the tenth group was satisfied with χ2<χ2_0.500.Second,taking Myzus persicae（Sulzer）for example,65 significant impact factors were obtained and 32 effective models were established.The results of Chi-square test showed that all of the back-substitution predicted results of 32 models were satisfied withχ²<χ²_0.995,in other words,there is no significant difference between the back-substitution predicted values and the observed values.We obtained six groups of Chi-square cumulative value;the third group was satisfied withχ²<χ²_0.950,the fourth group was satisfied withχ²<χ²_0.975,the seventh group was satisfied withχ²<χ²_0.975,the eighth group was satisfied withχ²<χ²_0.995,the eleventh group was satisfied withχ²<χ²_0.950.950 and the twelfth group was satisfied withχ²<χ²_0.900.It is accurate and feasible to predict the annual occurrence dynamics of pests in the same place by reasonably using the meteorological factors in different ares and in previous years.Compared with the predictive methods using only current year and local meteorological factors,the main advantages of this approach is that it can be released in advance long-term predicted time and a prediction model can also be established even without local historical meteorological data as long as there is an accumulation of historical pests information that has a sufficient duration in a system.The above two points are the real problems that people are always seeking to solve,therefore,the results of this study have strong practicality and generalizability.