Study On Hyperspectral Characteristics And Estimating Models About Physiological Parameters Of Rape

Agriculture is a country’s basic industry,belonging to the primary industry,is to provide support for national economic construction and development of basic industries.The average yield of rapeseed in domestic vegetable oil accounted for more than half.At the same time,rapeseed oil production efficiency is high,so in China’s vegetable oil industry has a pivotal position.Therefore,rapid,accurate,large-scale,real-time access to rape-related physiological and biochemical parameters,detection of rape growth situation is particularly important.In this study,rapeseed(Brassica napus L.)was obtained from field trials and laboratory tests.The hyperspectral remote sensing images of Brassica napus and the physiological and biochemical parameters of rape were obtained from different scales.The optimal physiological and biochemical parameters prediction model of rapeseed was constructed by computer and mathematical statistics method.The method of inversion and verification of hyperspectral remote sensing images on different scales was used to lay the foundation for the development of fine agriculture in rape and promote the process of agricultural digital informatization..The main conclusions are as follows:(1)From the overall trend point of view,different growth stages of rape canopy hyperspectral characteristics are more consistent.At the same time,with the increase of canopy chlorophyll content and nitrogen content,the high spectral reflectance of rape canopy decreased in the visible range,and the high reflectance platform in the near infrared band was increased.With the increase of canopy chlorophyll content and nitrogen content,the red edge feature also appeared "red shift" phenomenon.From the seedling stage to the flowering stage,the "redshift" phenomenon of the red edge feature is obvious,and the red edge feature gradually shows the phenomenon of "blue shift".It can be seen that the physiological and biochemical parameters of rape canopy and the hyperspectral reflectivity exist between this inextricably linked.(2)A series of spectral characteristic parameters and vegetation index,which can reflect the physiological and biochemical parameters of vegetation,were extracted from the canopy spectral data of rape.The results showed that most of the spectral characteristics and vegetation indices were related to chlorophyll(SPAD)content and leaf nitrogen(P <0.05).The results showed that most of the spectral parameters and vegetation index were related to chlorophyll(SPAD)content and leaf nitrogen content(SPAD)content and leaf nitrogen content prediction model of rape were not only higher than that of rapeseed at different growth stages,but also avoided the single growth period prediction model.Physiological and biochemical parameters of the drawbacks.(3)In the study of chlorophyll content(SPAD)content and leaf nitrogen content in rape rape during the whole growth period,we found that by using the random forest regression algorithm,we chose the spectral characteristic parameters and vegetation index with independent influence as the independent variables,The prediction model is better than traditional regression algorithm and multiple stepwise regression algorithm.Based on the previous studies,the improvement of chlorophyll(SPAD)content and the prediction model of leaf nitrogen content in rapeseed was improved.(4)The results showed that the inversion results were basically consistent with the distribution of rapeseed(Brassica napus L.),and the results showed that the inversion results were basically consistent with the distribution of rapeseed.