The soy was taken as the investigative object and the red edge spectral reflectivity was simulated by inverted Gaussian model. The red edge swing (Dλred) was gained after differential coefficient original spectrum and SWeibull model to estimate chlorophyll has been set up. The correlative coefficient was 0.781 2. At the same time the multivariate un-linear regression equation between TK content and absorption depth parameters were obtained by the method of continuum removal with their correlative coefficient of 0.762 5. It is proved that the possibility of forecast plants' nutrition status by the least spectral data within given wave band and the theory also provided a foundational consult to exploit on-line detecting nutrition and variable fertilization in the fields. 