Study On Hyperspectral Characteristics Of Indica Hybrid Rice Canopy And Its Relationship With Agronomic Parameters Under Different Nitrogen Application Levels

The acquisition of crop growth is the basis and premise of the implementation of precision agriculture.Hyperspectral remote sensing promotes the rapid development of precision agriculture via the advantages of obtaining target information timely,accurately and non-destructively.In order to obtain the growth status of rice accurately and quickly,this study took indica hybrid rice as the research object,and conducted the split-plot experiment of two factors concerning cultivar and nitrogen application level.Depending on the canopy hyperspectral reflectance and agronomic parameters[aboveground dry biomass(ADB),leaf area index(LAI),leaf chlorophyll relative content(SPAD),leaf nitrogen content(LNC),leaf nitrogen accumulation(LNA)and aerial part nitrogen accumulation(APNA)]in the main growth period,the canopy hyperspectral characteristics of indica hybrid rice under different nitrogen application levels and different growth periods were demonstrated;the correlation between hyperspectral parameters and agronomic parameters was analyzed;the agronomic parameter inversion models were constructed with hyperspectral reflectance,vegetation index and hyperspectral characteristic parameters as independent variables;and the inversion models were evaluated by the determination coefficient(r²),root mean square error(RMSE)and relative error(RE).The main conclusions are as follows:(1)The canopy hyperspectral characteristics of indica hybrid rice under different nitrogen levels were studied.As for the original hyperspectral characteristics,the change trend of canopy original hyperspectral reflectance of Qyou 6 and Yixiangyou 2115 was the same.With the increase of nitrogen application level,the reflectance at 401?680 nm did not change remarkably,but the reflectance at 750?2500 nm increased gradually.As for the first derivative hyperspectral characteristics,with the increase of nitrogen application level,the peak at 525 nm showed a gradually increasing trend,the trough at 570 nm showed a gradually decreasing trend,and the change trend of qyou 6 and Yixiangyou 2115 was the same.Under different nitrogen application levels,the first derivative hyperspectral characteristics of Qyou 6 and Yixiangyou2115 at 680?780 nm indicated obvious“multi peak”phenomenon,and the maximum peak was near 740 nm.(2)The canopy hyperspectral characteristics of indica hybrid rice at different growth stages were explored.As for the original hyperspectral characteristics,with the growth process moving forward,from jointing stage to heading stage,the hyperspectral reflectance of Qyou 6was lower at 401?680 nm band,and there was no considerable difference,while the hyperspectral reflectance of maturity stage was considerably higher;at 750?1350 nm,the hyperspectral reflectance increased from jointing stage to booting stage,but decreased from booting stage to maturity stage.With the development of growth period,as for Yixiangyou 2115,the reflectance near 540 nm for booting and heading stage was remarkably higher than that at jointing stage,and the hyperspectral reflectance at maturity stage was considerably higher than that at the first three stages;at 750?1350 nm,the change trend of hyperspectral reflectance of Yixiangyou 2115was the same as that of Qyou 6,but the hyperspectral reflectance at all growth stages was higher than that of Qyou 6.As for the first derivative hyperspectral characteristics,for Qyou 6,with the development of growth period,the peak value at 480?550 nm showed an upward trend from jointing stage to booting stage,and a downward trend from booting stage to heading stage;the first derivative hyperspectral reflectance of maturity stage was remarkably higher than that of the first three periods;the trough value of the lowest trough at 550?600 nm showed a downward trend from jointing stage to booting stage,whereas a gradual upward trend from booting stage to maturity stage.For Yixiangyou2115,with the development of growth period,the peak value at 480?550 nm showed an upward trend from jointing stage to booting stage,and a downward trend from booting stage to maturity stage,and the peak value of maturity stage remained at a high level;the change trend for the trough value of the lowest trough at 550?600 nm was the same as that of Qyou 6.(3)The hyperspectral inversion models for agronomic parameters of indica hybrid rice were constructed.The inversion model based on the vegetation index is better than the inversion model based on the sensitive wavelength;the inversion model based on the first derivative is better than the model based on the original hyperspectral reflectance;the estimation ability of the multiple regression model is generally better than that of the unary regression model;The best indicator to characterize the nitrogen nutrition status is the leaf nitrogen content.The hyperspectral parameters and modeling methods selected in this study are not suitable for inversion of LAI of indica hybrid rice.The optimal inversion model for ADB of indica hybrid rice is a 4-variable regression model,and the model expression is ADB=-1.38E+04+6.44E+03R?RVI_(702,526)+3.99E+05R?DVI_(1292,1217)-1.20E+06D?DVI_(728,722)+1.24E+04A3;r²,RMSE and RE of the test set is 0.909,1.398 t/hm² and 13.809%,respectively.The optimal inversion model of Indica hybrid rice SPAD is a 9-variable regression model,and the model expression is SPAD=8.72E+02-7.52E+02R?RVI_(772,773)+1.50E+03R?NDVI_(773,772)+7.32D?RVI_(773,721)+5.27D?NDVI_(773,721)-1.17E-01?o+1.01E+01A3-8.71A4-3.03E-01A5-1.64A6;r²,RMSE and RE of the test set is 0.694,1.708 and 4.110%,respectively?The optimal inversion model for the LNC of indica hybrid rice is the 11-variable regression model,and the model expression is LNC=-3.07E+01-4.29E+02D₇₅₁+3.28E+01R?RVI_(578,524)+6.79E+02R?DVI(751,750)-8.79E+01R?NDVI_(578,524)-3.83E+01SDb-2.99E+02Dr+6.06SDr+7.45E+01?g-9.53E-01SDg-1.38E+01?r-4.01E-01A5;r²,RMSE and RE of the test set is 0.911,0.298%and 13.540%,respectively.The optimal inversion model for the LNA of indica hybrid rice is the 7-variable regression model,and the model expression is LNA=2.44E+03+2.86E+04D₇₅₅-2.75E+04Dy-6.46E+02SDy-1.29E+04Dr-4.37?g+5.18E+02?r-4.51E+01A6;r²,RMSE and RE of the test set is 0.817,14.579 kg/hm² and 25.964%,respectively.The optimal inversion model for indica hybrid rice APNA is the 15-variable regression model,and the model expression is APNA=5.15E+03-3.18E+02R₆₈₀+2.71E+05D₆₇₅+1.31E+03R?DVI_(1080,752)+1.02E+05Db-3.03E+04Dy+4.78E-01?y-1.04E+04Dr-8.73E+01SDr-4.02E+02?g-9.30?g-1.99E+01SDg-3.14E+02?r+1.24A1-1.47E+02A4-2.40E+01A5;r²,RMSE and RE of the test set is 0.618,26.052 kg/hm² and 19.445%,respectively.