Objective  Utilization of digital images taken by unmanned aerial vehicle (UAV) to quickly and accurately monitor the nitrogen nutrition of winter wheat crops in the field for fertilization management was explored.

  Method  The digital images of field winter wheat at flagging and flowering stages obtained by camera on a UAV along with the agronomic information of the crop on the ground were collected. A correlation analysis on the data was conducted, and the variance inflation factors integrated for index selection. The selected indices sensitive to the variations of nitrogen nutrition indicators (NNIs) and free of co-linearity among themselves were employed to develop a mathematic model on nitrogen nutrition using the partial least square regression method and verified with the collected data for prediction accuracy and applicability.

  Result   The correlation coefficient and variance inflation factor allowed a precise index selection from the digital images. The indices for the winter wheat at flagging stage were thus determined to include b, g/b, (r-g-b)/(r+g), NDI, and WI, while those at flowering stage b, r/b, (r-g-b)/(r+g), and VARI. The coefficient on the model at the flowering stage was 0.008 8 higher, and the root mean square error 0.021 7 lower, than those at the flagging stage.

  Conclusion  Using the UAV digital images of the winter wheat at flagging and flowering stages, a visualized distribution map of the indices was constructed to enable a clear and accurate display of the nitrogen nutrition status of the crop in the field. The results at the flowering stage were considered slightly more sensitive than those obtained at the flagging stage in providing information on fertilization management.