Study On Monitoring Of Wheat Grain Protein Content Based On UAV Hyperspectral

Grain protein content is an important index to evaluate winter wheat quality.Rapid and accurate detection of grain protein content in wheat is beneficial to its evaluation and grading management.Hyperspectral remote sensing technology is a nondestructive and rapid monitoring method,it could monitor the grain protein content of winter wheat and obtained the current wheat grain protein content information quickly and accurately in the process of wheat growth.Four different winter wheat varieties were selected in this study,which were Gaoyou 5766(strong gluten),Jimai 23(strong gluten),Jimai 22(medium gluten)and Taimai 198(medium gluten).Based on UAV platform we obtained high-resolution hyperspectral images of wheat in critical period,and link leaf and plant nitrogen content,nitrogen accumulation,SPAD value,establish the monitoring model based on spectral parameters-agronomy parameters-wheat grain protein content.The independent data test of the related model was carried out to verify the accuracy and stability of the model.The specific results of this study are as follows.1 Monitoring Model of Leaf nitrogen content in Winter Wheat.Leaf nitrogen content of winter wheat monitored by the model based on normalized vegetation index NDVI:y=-4.19+20.86x-14.2x~2(x=NDVI(800,670),R~2=0.72)has the best effect.By independent data test,sample determination coefficient(R~2)is 0.72,RMSE and MRE are 0.18%and 4.76%.So the model can provide technical reference for accurate monitoring of leaves nitrogen content in winter wheat.2 Monitoring model of leaf nitrogen accumulation in winter wheat.Leaf nitrogen accumulation of winter wheat monitored by the model based on normalized vegetation index RVI:y=103.65-8.83x(x=RVI(810,560),R~2=0.66)has the best effect.By independent data test,sample determination coefficient(R~2)is 0.84,RMSE and MRE are 12.67%and 15.25%.So the model can provide technical reference for accurate monitoring of leaves nitrogen accumulation in winter wheat.3 Monitoring model of plant nitrogen content in winter wheat.Plant nitrogen content of winter wheat monitored by the model based on normalized vegetation index GNDVI:y=-1.87+9.593x-4.213x~2(x=GNDVI(750,550),R~2=0.83)has the best effect.By independent data test,sample determination coefficient(R~2)is 0.85,RMSE and MRE are 9.64%and 10.59%.So the model can provide technical reference for accurate monitoring of plant nitrogen content in winter wheat.4 Monitoring model of plant nitrogen accumulation in winter wheat.Plant nitrogen accumulation of winter wheat monitored by the model based on normalized vegetation index RVI:y=451.03-94.33x+6.19x~2(RVI(810,560),R~2=0.72)has the best effect.By independent data test,sample determination coefficient(R~2)is 0.81,RMSE and MRE are 21.62%and 14.73%.So the model can provide technical reference for accurate monitoring of plant nitrogen accumulation in winter wheat.5 monitoring model of SPAD value of winter wheatSPAD value of winter wheat monitored by the model based on normalized vegetation index NDVI:y=y=-113.67+437.27x-272.10x~2(x=NDVI(800,670),R~2=0.82)has the best effect.By independent data test,sample determination coefficient(R~2)is 0.85,RMSE and MRE are 3.15%and 5.37%.So the model can provide technical reference for accurate monitoring of SPAD value in winter wheat.6 Monitoring model of grain protein content of winter wheatA multi-factor model(y=10.52-0.06×SPAD+2.45×PNC+0.02×LNA,R~2=0.95)was established based on leaf nitrogen accumulation,plant nitrogen content,SPAD value and grain protein content in wheat filling stage has the best effect on grain protein content of winter wheat.By independent data test,sample determination coefficient(R~2)is 0.94,RMSE and MRE are 0.30%and 1.60%.The prediction model of SPAD is y=113.67+437.27x-272.10x~2(x=NDVI(810,670),R~2=0.82),the prediction model of plant nitrogen content is y=-1.87+9.593x-4.213x~2(x=GNDVI(750,550),R~2=0.83),and the prediction model of leaf nitrogen accumulation is:y=103.65-8.83x(x=RVI(810,560),R~2=0.66).