Winter wheat LAI estimation using unmanned aerial vehicle RGB-imaging

Leaf area index (LAI) is an important agronomic parameter used in evaluating crop growth characteristics. The accurate estimation of LAI based on remote sensing technology is critical for precision agriculture. The current cost-effective unmanned aerial vehicle (UAV) of agricultural remote sensing monitoring system, which was established based on a multi-rotor UAV with a digital camera mounted on its platform, has led to significant achievements in agricultural research. However, there has been little research on retrieving crop LAI based on UAV digital imagery. To demonstrate the feasibility of using UAV digital imagery to estimate winter wheat LAI, we used this cost-effective UAV system to monitor agricultural operation in the study area. Then many UAV digital images (also known as RGB images) used as the study data source recorded at three critical growth stages — booting, anthesis and filling stages of winter wheat. We calculated ten characteristic parameters from the RGB images based on digital imaging conversion principle. Furthermore, we systematically analyzed the relationship between LAI at the three growth stages of the two winter wheat varieties with the four nitrogen levels and characteristic parameters of RGB images. It was indicated that among the ten characteristic parameters, R/(R+G+B) and UAV-based VARIRGB (visible atmospherically resistant index based on UAV RGB image, which was calculated in this paper based on DN in the red, green and blue channels of UAV digital images and the calculation principle of VARI) regularly changed with LAI of winter wheat. The change occurred regularly and simultaneously for the three growth stages. It showed that different nitrogen levels in winter wheat not only influenced LAI, but also influenced some characteristic parameters of digital images. Meanwhile, the study also indicated that R/(R+G+B) and UAV-based VARIRGB were more significantly correlated with LAI under different conditions, including variety, nitrogen level and growth stage among the ten characteristic parameters. Then two comprehensive evaluation of LAI inversion models between LAI and R/(R+G+B) and UAV-based VARIRGB were established. The evaluation demonstrated that UAV-based VARIRGB was the best parameter which optimally retrieved LAI of winter wheat. LAI estimated by the exponential model of UAV-based VARIRGB strongly matched with measured LAI, with R² = 0.71, RMSE = 0.8 and at 0.01 significance level. Therefore, the results showed that the application of UAV digital imagery in retrieving winter wheat LAI was feasible. The study also enriched the achievements and experience of using cost-effective UAV remote sensing monitoring system in precision agriculture.