Approach to estimation of winter wheat nitrogen by using remote sensing technology on diverse scale and its application

This paper focused on estimating winter wheat nitrogen content from hyper-spectral airborne images and ground measured spectral data, and research methods for variable-rate fertilization based on remote sensing images by combining the nitrogen estimation method and Lukina's diverse fertilization model. In order to reach the target, firstly, moment-matching and reflectance conversion were employed to correct the image's radiation difference, then the OMIS image was geo-corrected using a geo-referenced aerial photograph image. The Inversed-Gaussian model was used to simulate winter wheat red edge of the processed OMIS image spectra and ground measured spectra, and the red edge parameters, such as red edge position, red edge wide were derived. The correlation between red edge parameters and measured nitrogen content was calculated, and the optimal red edge parameter whose correlation was the most significant and RMSE was the smallest was selected to estimate wheat nitrogen content. The results show that the correlation coefficient between winter wheat nitrogen content and the area difference between simulated spectral curve and actual reflectance curve is the most significant. Lastly, the plant nitrogen estimating method was integrated to Lukina's variable-rate fertilization model, and the variable-rate fertilization prescription map was created based on the OMIS hyper-spectral image. The optimal nitrogen estimating method in this paper improved the nitrogen estimating accuracy in Lukina model, and the technology of obtaining large area information overcame the shortcoming of point sampling technology, which made variable-rate fertilization technology more practical and easier to be generalized.