A Study On The Occurrence Prediction Of Dendrolimus Punctatus In Qianshan City And Comparison Of Insect Stages In Different Regions

This project was originated from the national forest public welfare industry scientific research professional master's degree(201404410),the Dendrolimus punctatus data were systematically and scientifically sorted out for Macheng city and Lichuan city of Hubei province,Wuyishan city and Pucheng city of Fujian province,Qianshan city and Fengyang county of Anhui province and then insect state period were compared and analyzed in different areas.Nine methods,such as periodic distance,stepwise regression,stationary time series,Markov Chain,BP artificial neural network,multi-stage-factor contingency table analysis,catastrophe prediction,analysis of variance periodic extrapolation,fuzzy mathematics comprehensive evaluation and so on 9 methods,were used to analyze Dendrolimus punctatus in Qianshan with the time span was large and complete data.The forecast of occurrence period,occurrence quantity,occurrence area and hazard index of Dendrolimus punctatus provided theoretical basis for prevention and treatment of this insect.The results of the comparison of the occurrence periods in different regions show that the peak time of Dendrolimus punctatus was related to latitude and altitude.The higher the latitude,the later the occurrence;The higher the altitude,the later the occurrence.Results of mathematical analysis on Dendrolimus punctatus in Qianshan city,anhui province:BP neural network method was used to predict the occurrence period and amount of Dendrolimus punctatus.lts historical coincidence rate reaches 100%.The historical coincidence rate of anova periodic extrapolation and fuzzy comprehensive evaluation method in predicting the occurrence amount reached 100%.BP neural network prediction method were ideal methods to predict the occurrence area and the historical coincidence rate of disaster analysis method for the occurrence area was over 90%.Both the periodic extrapolation method and the disaster analysis method had high accuracy in predicting the hazard index.Based on comprehensive analysis.BP neural network method,stationary time series method,analysis of variance periodic extrapolation and Markov chain method had the best historical coincidence rate and other five methods had better prediction results.The key was to select forecast factor which was closely related to forecast quantity and scientific classification.