Prediction Effect Analysis Of Different Models Of Wheat Grain Protein Content

Wheat nitrogen nutrition has an important influence on grain protein content. The timely obtained information of its growth status, especially nitrogen status, will be particularly important for fertilizer management and estimation of grain protein content during the production of wheat. The experiment was carried out on the National Precision Agricultural Research bases from2011to2013. Wheat was the mainly study object in the experiment. The prediction model of wheat grain protein content was built by combining the remote sensing data and agronomic parameter on the ground based on nitrogen nutrition. Aim to build the grain protein content model in wheat, the relationship between remote sensing parameter and nitrogen diagnosis index, and the relationship between nitrogen diagnosis index and grain protein content were investigated in this study. Meanwhile the relation between nitrogen nutrition we screened and the spatial variability of wheat grain protein content was studied based on many methods such as traditional statistics and geo-statistics.The contents and results are summarized as below:(1) Correlation coefficient was chosen as the indicator to reflect the intimate level of linear correlation between grain protein content and some nitrogen diagnosis indexes obtained from four developmental stages.The results showed that Nitrogen Nutrition Index (NNI) at flag growth stage of wheat was the best agricultural parameter for grain protein content detection, compared with other agricultural parameters at other stages. It can be used as an intermediate variable between remote sensing parameter and grain protein content. The model which NBI predicted NNI was built, with a R2value of0.783, a root-mean-square-error (RMSE) value of0.164. The grain protein content established prediction model, using NNI and remote sensing parameter, had obtained certain accuracy, with a R2value of0.728, a predictive RMSE value of0.959%. (2) Field campaign was conducted to obtain NNI at flag growth stage of wheat and hyperspectral parameter such as MTCI、NIR/G、RVI、SR and DVI. Using these data, Grey Relational Analysis (GRA) and Partial Least Square (PLS) was used to establish NNI prediction model based on relationships between hyperspectral parameter and NNI, and between NNI and grain protein content, with with a R2value of0.73, a RMSE value of0.174. The NNI prediction model was coupled with grain protein content established model, with with a R2value of0.58, a RMSE value of1.03%.(3) The spatial structure between wheat stress fluorescent parameter nitrogen balance index (NBI) and grain protein content were analyzed using geostatistical methods, and their semivariogram was developed. The outcome showed that NBI and the data of wheat grain protein content were approximately normally distributed; Variables had good regionalized variable characteristics and spatial structures under impersonal factors; Grain protein content was significantly correlated to NBI at each developmental stage. Large enough NBI existing for a long time was beneficial to forming higher grain protein content. Correspondingly, large enough NBI existing for a short time often led to forming lower grain protein content; The spatial distribution of combined NBI and grain protein content had similarity, so NBI could be used to predict grain protein content. These results provided a theoretical basis and technical approach for managing crop nitrogen and improving grain quality in a certain area.