Aboveground biomass analysis of an alpine meadow based on unmanned aerial vehicle hyperspectral images in the Haibei pilot area

The yield and quality of natural pastures play important roles in livestock production. However, traditional multispectral remote sensing has limitations in aboveground biomass (AGB) estimation of natural grasslands because of fewer discontinuous spectral channels. With the rapid development of unmanned aerial vehicle (UAV) imaging via hyperspectral remote sensing technology and its widespread application in AGB monitoring of grasslands in recent years, there is a possibility of achieving high-precision estimation of natural grassland AGB. Herein, the alpine meadow in the Haibei pilot area was studied using UAV imaging hyperspectral remote sensing and ground measurement data. The alpine natural grassland estimation model was developed using the LASSO approach and RF machine learning algorithm, and the growth conditions of grassland in the Haibei pilot area were evaluated. The main results showed that 1) in the growing season of May to September, the AGB of the alpine meadow was highest in August with an average value up to 3025.70 kg·ha⁻¹; meanwhile, the sensitive band of AGB was located in the green (503, 510, 513, 536, 566, and 573 nm), red (733 nm), and near infrared (803, 850, 875, and 879 nm) regions. 2) The RF approach was applied to estimate the AGB in alpine meadows, with an R² of 0.81 and root mean square error (RMSE) of 489.36 kg·ha⁻¹ for the validation set and an R² of 0.95 and RMSE of 248.70 kg·ha⁻¹ for the training set of the optimal RF estimation model. 3) UAV hyperspectral remote sensing technology can realize high-precision monitoring of the AGB of alpine meadows, and the research results provide an important reference for the rapid monitoring of AGB in alpine meadows.