The accurate assessment of cotton nitrogen (N) content over a large area using an unmanned aerial vehicle (UAV) and a hyperspectral meter has practical significance for the precise management of cotton N fertilizer. In this study, we tested the feasibility of the use of a UAV equipped with a hyperspectral spectrometer for monitoring cotton leaf nitrogen content (LNC) by analyzing spectral reflectance (SR) data collected by the UAV flying at altitudes of 60, 80, and 100 m. The experiments performed included two cotton varieties and six N treatments, with applications ranging from 0 to 480 kg ha⁻¹. The results showed the following: (i) With the increase in UAV flight altitude, SR at 500–550 nm increases. In the near-infrared range, SR decreases with the increase in UAV flight altitude. The unique characteristics of vegetation comprise a decrease in the “green peak”, a “red valley” increase, and a redshift appearing in the “red edge” position. (ii) We completed the unsupervised classification of images and found that after classification, the SR was significantly correlated to the cotton LNC in both the visible and near-infrared regions. Before classification, the relationship between spectral data and LNC was not significant. (iii) Fusion modeling showed improved performance when UAV data were collected at three different heights. The model established by multiple linear regression (MLR) had the best performance of those tested in this study, where the model-adjusted the coefficient of determination (R²), root-mean-square error (RMSE), and mean absolute error (MAE) reached 0.96, 1.12, and 1.57, respectively. This was followed by support vector regression (SVR), for which the adjusted_R², RMSE, and MAE reached 0.71, 1.48, and 1.08, respectively. The worst performance was found for principal component regression (PCR), for which the adjusted_R², RMSE, and MAE reached 0.59, 1.74, and 1.36, respectively. Therefore, we can conclude that taking UAV hyperspectral images at multiple heights results in a more comprehensive reflection of canopy information and, thus, has greater potential for monitoring cotton LNC.

Rapid and accurate prediction of crop nitrogen content is of great significance for guiding precise fertilization. In this study, an unmanned aerial vehicle (UAV) digital camera was used to collect cotton canopy RGB images at 20 m height, and two cotton varieties and six nitrogen gradients were used to predict nitrogen content in the cotton canopy. After image-preprocessing, 46 hand features were extracted, and deep features were extracted by convolutional neural network (CNN). Partial least squares and Pearson were used for feature dimensionality reduction, respectively. Linear regression, support vector machine, and one-dimensional CNN regression models were constructed with manual features as input, and the deep features were used as inputs to construct a two-dimensional CNN regression model to achieve accurate prediction of cotton canopy nitrogen. It was verified that the manual feature and deep feature models constructed from UAV RGB images had good prediction effects. R² = 0.80 and RMSE = 1.67 g kg⁻¹ of the Xinluzao 45 optimal model, and R² = 0.42 and RMSE = 3.13 g kg⁻¹ of the Xinluzao 53 optimal model. The results show that the UAV RGB image and machine learning technology can be used to predict the nitrogen content of large-scale cotton, but due to insufficient data samples, the accuracy and stability of the prediction model still need to be improved.

For crop growth monitoring and agricultural management, it is important to use hyperspectral remote sensing techniques to estimate canopy nitrogen content in a timely and accurate manner. The traditional nadir method has limited ability to assess the nitrogen trophic state of cotton shoots, which is not conducive to high-precision nitrogen inversion, whereas the multi-angle remote sensing monitoring method can effectively extract the canopy’s physicochemical information. However, multi-angle spectral information is affected by a variety of factors, which frequently causes shifts in the band associated with nitrogen uptake, and lowers the estimation accuracy. The capacity of the spectral index to estimate aerial nitrogen concentration (ANC) in cotton was therefore investigated in this work under various observation zenith angles (VZAs), and the Relief−F method was employed to select the best spectral band with weight for ANC that is insensitive to VZA. Therefore, in this study, the ability of the spectral index to estimate ANC in cotton was explored under different VZAs, and the Relief-F algorithm was used to optimize the optimal spectral band with weight for ANC that is insensitive to VZA. The angle insensitive nitrogen index (AINI) for various VZAs was calculated using the expression (R₅₃₀ − R₇₀₄)/(R₁₄₁₂ + R₇₀₄). The results show that the correlation between the spectral index and the ANC chosen in this study is stronger than the correlation between off-nadir observations, and the correlation coefficients between Photochemical Reflectance Index (PRI), AINI, and ANC are highest when VZA is −20° and −50° (r = 0.866 and 0.893, respectively). Compared with the traditional vegetation index, AINI had the best correlation with ANC under different VZAs (r > 0.84), and the performance of ANC in the backscatter direction was estimated to be better than that in the forward-scatter direction. At the same time, the ANC estimation model of the optimal indices AINI and PRI was combined with the machine learning method to achieve better accuracy, and the prediction accuracy of the random forest (RF) model was R² = 0.98 and RMSE = 0.590. This study shows that the AINI index can estimate cotton ANC under different VZAs. Simultaneously, the backscattered direction is revealed to be more conducive to cotton ANC estimation. The findings encourage the use of multi-angle observations in crop nutrient estimation, which will also help to improve the use of ground-based and satellite sensors.

Crop nitrogen is an efficient index for estimating crop yield. Using hyperspectral information to monitor nitrogen in cotton information in real time can help guide cotton cultivation. In this study, we used drip-irrigation cotton in Xinjiang as the research object and employed various nitrogen treatments to explore the correlation between hyperspectral vegetation indexes and leaf nitrogen concentration (LNC) and the canopy nitrogen density (CND) of cotton in different growth periods and interannual. We employed 30 published hyperspectral vegetation indexes obtained through spectral monitoring in 2019 and 2020 to screen for hyperspectral vegetation indexes highly correlated with the nitrogen in cotton indexes. Based on the same group of hyperspectral vegetation indexes, interannual and multi-temporal nitrogen estimation models of cotton were established using three modeling methods: simple multiple linear regression (MLR), partial least-squares regression (PLSR), and support vector regression (SVR). The results showed the following: (1) The correlations between LNC and CND and vegetation index in individual growth periods of cotton were lower than those for the entire growth period. The correlations between hyperspectral vegetation indexes and cotton LNC, CND, leaf area index (LAI), and aboveground biomass (AGB), were significantly different between years and varieties. The relatively stable indexes between vegetation and LNC were TCARI, PRI, CCRI, and SRI-2, and the absolute values of correlation were 0.251~0.387, 0.239~0.422, 0.245~0.387, and 0.357~0.533. In addition, the correlation between CIred-edge and REIlinear and group indicators (CND, AGB, and LAI) was more stable. (2) In the models established by MLR, PLSR, and SVR, the R² value from the SVR method was higher in the estimation model based on the entire growth period data and LNC and CND. (3) Using the same group of selected hyperspectral vegetation indexes to estimate nitrogen in cotton in different growth stages, the accuracy of the estimation model of canopy nitrogen density (CND) was higher than that of the estimation model for leaf nitrogen concentration. The canopy nitrogen density most stable model was established by MLR at the flowering and boll stages and the full-boll stage with R² = 0.532~0.665. This study explored the application potential of hyperspectral vegetation indexes to the nitrogen of drip-irrigated cotton, and the results provide a theoretical basis for hyperspectral monitoring for crop nutrients and canopy structure.

Cotton is the main economic crop in China and is important owing to its use as an industrial raw material and a cash crop. This experiment was conducted in the main cotton-producing area of Xinjiang, China. A hyperspectrometer was used to monitor the canopy spectral reflectance of cotton at different stages of growth. The results showed that the leaf area index (LAI) increased with the increase in the amount of nitrogen fertilizer added during the early full boll stage and decreased with the increase in nitrogen fertilization in the full and late boll stages. Insufficient or excessive fertilization led to a decrease in the LAI. The visible light band indicated that the canopy spectral reflectance decreased, and the amount of fertilizer increased in all the growth stages. The near-infrared band revealed that the canopy spectral reflectance increased with the amount of nitrogen applied during the bud stage, early boll stage, and the most vigorous period of boll growth. During the flowering period, the spectral reflectance followed the order N3 > N4 > N2 > N1 > N0. During the entire growth period of cotton, the values of the cotton LAI predicted using the ratio vegetation index (RVI) model were found to best fit the measured values. The LAI monitoring models of cotton in each growth stage were different. The TVI model is the best in the bud and early boll stages. The NDVI model is the best in the flowering stage, and the DVI model is the best in the full boll stage. This study provides a basis to accurately monitor the LAI in each growth period of cotton.