Quantification of nitrogen status in rice canopy by low-altitude hyperspectral remote sensing

Abstract. Rice is a major grain crop all over the world, and the yield of rice is affected by the amount of nitrogen fertilizer applied in rice cultivation. Based on low-altitude multispectral imaging remote sensing technology, nitrogen content of rice canopy could be detected timely and effectively, a fine nutrient management program could be developed and rice yield could be improved. In this research, the rice of Chunyou 84 was used as the object of the research, and SPAD values were used as the nitrogen evaluation index. The low-altitude remote sensing experiment was carried out by an IMEC hyperspectral snapshot camera with 25 bands（600-1000 nm）on a low-altitude remote sensing simulation platform. Then low-altitude remote sensing analytical models of SPAD values for rice canopy were established respectively based on the full spectral information and vegetation indices of the multispectral images. The results showed that good predictive results could be obtained for the SPAD of rice canopy by the calibration of 69 canopy samples and the prediction of 30 canopy samples. The PLS, BPNN and LS-SVM models on the full spectral information could obtain high correlation. When the prediction models of SPAD for rice canopy were set up based on vegetation indice of the spectral images, The linear model on NDVI-like with the optimized wavebands of 693 and 849 nm not only provided higher correlation but also simplified the model. The above results indicated that low-altitude hyperspectral imaging remote sensing could be used to achieve the quantification of nitrogen status in rice canopy.