Occurrence Degree Forecasting Of The Second Generation Of Corn Borer In Shandong Based On Big Data

As an important part of the big data, agricultural big data is the application and development of big data concept, technology and method in the field of agriculture. Depending on the data that have collected for many years, which included crop pests, crop growth status, weather conditions, surrounding environment, soil characteristics, agricultural management measures. Depending on the concept, technical analysis and predictive function, of big data, making the data into the action of monitoring and early warning for the occurrence of pests is the application of the idea and concept of big data, it should be the trend of future development of pest monitoring and forecasting.Pyrausta nubilalis Hübner(Order: Lepidoptera), a pest of the family Pyralidae, make great harm on corn production. Asian corn borer Ostrinia furnacalis(Guenée) and European corn borer Ostrinia nubilalis(Hübner) are the two species that occurred in maize field.The Asian corn borer is the dominant species, which occurs 2-3 generations in the maize fields, and the second generation is the main damaging generation. The larvae feed on corn leaf, stem and ear, causing maize crop off the wind, early blight and grain reduction. According to the database of meteorological factors when the second-generation corn borer out broke in Shandong Province,and based on SPSS analysis method and R language, we established multiple linear regression model, principal component regression model and principal component polynomial regression model to predict the occurrence degree of corn borer. The results are as follows:(1) After correlation analysis of meteorological factors of Shandong Province in the year 2003 to 2013, such as the average annual temperature, precipitation, sunshine duration, wind speed and other meteorological data, with multiple linear regression, we analysis and predict the occurrence degree of corn borer in late June, early, middle, and late July, and in early August. After examination, in late June, early July, late July and early August, the accuracy of the model fitness was 91%, 77%, 69%, and 68% respectively,(2) Depending on R language as the platform for data analysis, the principal component analysis method was adopted. Nine principal components in the 16 meteorological factors were identified. The 9 principal components were regarded as the independent variables. Based on the principal component, we established principal component regression, which are based on the second generation of the corn borer occurrence degree in Shandong Province, the multiple R-squared is 0.437.(3) Using R language as the platform for data analysis as well,we established polynomial regression, which are based on the second generation of the corn borer occurrence degree in Shandong Province, the multiple R-squared is 0.5012. Results of two kinds models show that: the impact factors that selected from nonlinear principal component analysis method are more accurate, comprehensive and stable than that from correlation analysis, which the final 9 principal component reached to 99.36% of contribution rate. After verification, principal component regression explain ability is weak, referential meaning is not well; principal component polynomial regression is better than principal component regression, it can explain in a certain extent, and is effective in forecasting of corn borer in agricultural yield.