Abstract:In order to improve the management ability of wheat nitrogen accurately and efficiently, real-time, convenient and accurate variable rate fertilization was realized. Based on the nitrogen nutrition diagnosis of winter wheat by UAV remote sensing and the fertilization model of spectral diagnosis, the recommended fertilization models for key growth stages of winter wheat, such as green-turning stage, jointing stage and heading stage were established.The results showed that the DVI obtained from UAV images correlated best with plant nitrogen concentration(R2=0.869 8), and the critical DVI values of nitrogen diagnosis at turning green stage, jointing stage and heading stage of winter wheat were 0.594, 0.784 and 0.807, respectively. The highest and best yield of winter wheat was 11 364.3 kg/ha, the best yield was 11 230.9 kg/ha, and the total nitrogen application amount was 272.9 kg/ha, respectively.The optimum economic fertilization amount was 264.1 kg/ha. The recommended fertilizer application models based on spectral diagnosis were y=-39.69 x+23.58 in the green-back stage, y=-129.97 x+101.95 at jointing stage and y=-159.79 x+128.93 at heading stage. The establishment of diagnostic indicators laid a foundation for the model of fertilizer top dressing based on spectral diagnosis. The critical DVI value did provide the basis for the diagnosis of whether nitrogen fertilizer was added in the key growth period of winter wheat. The establishment of the effect curve of nitrogen fertilizer and the selection of the application model of spectral diagnosis realized the quantitative recommendation of top dressing.