【摘要】 近年来,农作物长势监测和产量预测研究大多是通过建立复杂的生长模型来实现的,而这往往不具有较强的推广性。本研究建立了一种基于植被指数和产量统计数据的玉米长势监测及产量预测方法。以玉米为研究对象,利用MODIS09A1数据建立其2000-2018年的增强型植被指数(EVI)时间序列,并将该序列作为径向基(RBF)神经网络的输入参数,下一阶段的EVI值或玉米产量作为网络的输出参数,完成玉米的长势监测及产量预测。该方法被成功应用到黑龙江省哈尔滨市宾县的玉米研究中,对玉米EVI值的预测精度达到了90.0%以上,产量预测相较于传统的线性回归模型也有明显提高,预测精度达到了98.6%。依赖植被指数和产量统计数据的长势监测及产量预测方法有较大的应用推广前景。更多还原

【Abstract】 In recent years, the research of maize growth monitoring and yield prediction is mostly achieved by building complex growth models, but these methods are not easy to be popularized. To overcome the problem, a method of maize growth monitoring and yield prediction was established based on vegetation index and statistical yield data. Taking maize as the research object, the enhanced vegetation index(EVI) time sequence from 2000 to 2018 of the maize was established by using MODIS09 A1 data. Moreover, this time sequence was taken as the input parameter of radial basis function(RBF) neural network, and the values of EVI or maize yield in the next stage were taken as the output parameters. This method has been successfully applied to the research on maize in Binxian County, Harbin City, Heilongjiang province, with the prediction accuracy of over 90% for the value of EVI. Compared with the results based on conventional linear regression forecasting method, the prediction accuracy of yield based on the proposed method in this study was significantly improved. In conclusion, the method of growth monitoring and yield prediction based on vegetation index and statistical yield data has a great application prospect.更多还原