Study On Integrated Prediction Of Early Rice Yield In Hunan Province

The prediction of early rice yield is one of the most important aspects of meteorological services for agricultural production.A software system and data structure and computer algorithm were designed by using software engineering knowledge.Single forecasting yield model was established by using data mining technique.The integrated forecasting model were established base on a single forecasting yield model by using weighting coefficient for improving forecasting accuracy and service ability of early rice yield and overcoming the deficiency of a single forecasting yield model in Hunan province. It tested using the integrated forecasting model. And the research results will help improve forecasting accuracy of early rice yield and service capacity in Hunan province. It enhances the scientific and technological content of early rice yield forecasting products. The main research results were:(1) In our research, the temperature suitability model, water suitability model,sunshine suitability model and climatic suitability model of early rice in different growth periods were established base on maximum temperature, optimum temperature, low temperature, water requirement and light requirement of early rice growth in Hunan province. And the dynamic forecasting model of early rice yield by ten days were established based on climatic suitability by using the relationship between the climatic suitability from seeding to forecast time and the early rice yield from1962to2009. The forecasting models are passed the significance level testing of0.05.(2) The pivotal meteorological factors of affecting yield of early rice were confirmed in Hunan province. And the dynamic forecasting model of early rice yield by ten days were established from seeding to forecasting time based on pivotal meteorological factors.Except the forecasting model in April30th is not passed the significant level testing of0.10, other one are passed the significant level testing of0.01.(3) The difference of meteorological elements in the same time between the forecasting year and historical any one year were calculated by using the Euclid’s distance and similarity coefficient. So that, a comprehensive diagnosis index was established. Meanwhile, the dynamic forecasting model of early rice yield by ten days were established by using the data from1962to2009based on historical meteorological influence index for bumper or poor harvest of crop yield in Hunan province.(4) The early rice was simulated daily by using localized model of ORYZA2000and climatic data from1962to2012in Hunan province. And the dynamic forecasting model of early rice yield by ten days were established based on crop growth simulation in Hunan province.(5) The sample percentage of consistency btween the results of single forecasting model amd the measured data were analysed. And the dynamic and integrated forecasting model of early rice yield by ten days were established by using weight based on climatic suitability, pivotal meteorological factor, historical yield meteorological effect and crop growth simulation in Hunan province. The testing results show that the integrated forecasting models were obviously improved and enhanced than that of the single model in some aspects, such as the sample percentage of consistency,forecasting accuracy and sample percentage of the error.