Nitrogen Nutrition Monitoring And Yield Prediction Of Winter Wheat By Hyperspectral Remote Sensing

People pay more and more attention on precision agriculture as they pursue the effective, low consumption and low pollution field management and the safe agricultural products. The remote sensing can obtain more information about crop growth quickly and precisely, therefore, it can guide field management. The hyperspectral remote sensing is high spectral resolution, thus it contains plentiful information, to monitor crop growth precisely. In this study a series of field experiments with two wheat varieties and five nitrogen statuses were carried out in two years. Based on analysis of canopy spectral reflectance, assay of agronomic parameters and physic-chemical index, this study built the hyperspectral models which can predict the nitrogen status and yield precisely. These models can guide the field management.The graphs of the reflectance of the wheat on different N treatment were different, but had the same trend as same as the plant’reflectance graph. It has a peak near550nm wave-band, after700nm it soared quickly, near750nm the reflectance reached the highest and steady. At this reflection platform, the reflectance increased as the use amount of nitrogen.The result of the relation between spectral parameters and nitrogen content of two winter wheat leaf showed that:(1) Grain filling stage was more appropriate stage for diagnosis wheat leaf nitrogen content by hyperspectral remote sensing.(2) The spectral reflectance of680nm waveband R680can predict leaf nitrogen content at jointing, heading and grain filling stage of winter wheat growth period. The best leaf nitrogen content regression equation of XiNong979based on spectral reflectance of680nm waveband at jointing stage was Y=27.54-280.247X+1456.245X2; at heading stages was Y=8.632X-0.24which based on spectral reflectance of green peak; at grain filling stage was Y=25.83X1.012which based on vegetation index of (SDr-SDb)/(SDr+SDb). The best leaf nitrogen content regression equation of JinMai47were the same as XiNong979, they were Y=45.13-138.15X+796.33X2、 Y=14.38X-0.57and Y=12.35X0.65respectively.The result of the relation between spectral parameters and yield of two winter wheat showed that:(1) Grain filling stage was more appropriate stage for forecast yield of two winter wheat by hyperspectral remote sensing. SDr, SDr/SDy and (SDr-SDy)/(SDr+SDy) were the better spectral parameters for yield forecast.(2) The vegetation index of (SDr-SDy)/(SDr+SDy) was the best spectral parameters of XiNong979for yield forecast at heading and grain filling stages, The best yield regression equation based on vegetation index of (SDr-SDy)/(SDr+SDy) were Y=10709.92X+2495.54and Y=9825.32X+4282.19respectively. The best yield regression equation based on Rg for JinMai47at heading stage was Y=6158.19-21376.19X, at grain filling stage it was Y=3864.46X-2613.37which based on SDr/SDy.