Monitoring Of Winter Wheat Grain Protein Content Based On Hyperspectral Remote Sensing

With the development of society,technology,and the improvement of economy,attention to food quality as grain yield has increased concern in recent years.Grain protein content(GPC)is often be treated as one of the main factors that affect winter wheat quality.This study around a scientific problem that remote sensing monitoring of winter wheat grain protein content expand analysis from the two aspects.One,considering that the formation of wheat grain protein is the formation process of a accumulated over a long period,starting from a simple empirical model that estimation model of grain protein content is based on multi-temporal spectral parameters.The other one,starting from grain nitrogen translocation,grain protein content is calculated by the ratio of grain nitrogen accumulation(GNA)and grain yield(GY).The conclusion mainly are as following:(1)Using grey correlation analysis and partial least squares regression,the estimating model of grain protein content based on crops spectral characteristic of multiple growth period had a higher precision.The results showed that the GPC model had better reliability in the later three growth stages(flag leaf,flowering,and filling stage).Comparing with the best precision in single growth stage(flowering stage),training coefficient of determination of the later three growth stages was improved and root mean square error was decreased obviously.(2)The grain yield research was used the cumulative value of typical vegetation index.We found that it has the best correlation between normalized difference vegetation index(744,784)and winter wheat grain yield during sensitive bands chosen about per vegetation index.The precision of nonlinear model were higher than the linetype.Compared with model in flowering,coefficient of determination in the later three growth stages was increased and root mean square error was reduced obviously.Therefore,the accuracy in several stages was higher than single stages.The prediction model of GY by integrated normalized difference vegetation index in several stages had a certain feasibility.(3)The research of grain nitrogen accumulation was based on nitrogen translocation and band depth analysis method.The grain nitrogen accumulation was composed of nitrogen translocation amount stored in plant before flowering and nitrogen assimilation amount after flowering.Based on nitrogen translocation,it turned NTA inversion into leaf nitrogen accumulation.Then FD742 was the best index to inversion leaf nitrogen accumulation at flowering.Combined with band depth analysis and partial least squares regression,the correlation between characteristics indices and NAA in flowering and filling stages were analyzed.The results showed that R2 between characteristic indices and NAA in different stage were all greater than 0.7 that was significant correlation.(4)The ratio of grain nitrogen accumulation and grain yield multiplied a conversion factor to obtain grain protein content model.Considering the values of coefficient of determination and root mean square error comprehensively,the results showed that characteristic index band depth normalized to band area used to inverse NAA has the highest accuracy during filling stage in coupling GPC model.Besides,it had a higher consistency between the predicted and the measured values in both modeling and validation.