Improving the prediction of vegetation carbon, phosphorus, and C:P ratio of alpine grassland in the Three-River Headwater Region by combining the field and laboratory reflectance spectroscopy

Abstract

Vegetation carbon (C), phosphorus (P), and C:P ratio of alpine grassland is vital for understanding forage quality and yield. Visible and near-infrared reflectance (VNIR) spectroscopy is an effective approach to analyzing the vegetation growing status in low-altitude region. However, the feasibility of applying VNIR spectroscopy for vegetation biochemical parameters monitoring in high altitude regions, such as the alpine grassland above 4500 m, remains unclear, and little is known about enhanced approaches to improve the estimation accuracy using comprehensive spectral information. Here, we explored the feasibility of using VNIR to estimate these stoichiometries in the Three-River Headwater Region (TRHR) by using field and laboratory spectra, and the possibility of using complementary spectral information to improve estimation accuracy. The R² of only using field spectra was less than 0.45 for C, P, and C:P. The laboratory spectra improved the estimation accuracy to 0.81, 0.66, and 0.64 respectively, and these accuracies were further improved to 0.87, 0.79,0.83 by complementary information from field and laboratory spectra. The results demonstrated the potential of VNIR to estimate accurate nutrient indicators of alpine grassland. The correlation between the field and laboratory spectra was changing and weak, and it varied from positive to negative in the ‘Red-edge’ region, implying that their spectral information was complementary. Improved estimation accuracy of nutrient indicators using complementary spectral information contributes to monitoring accurately the vegetation health in the TRHR and promotes our understanding of the forage quality and the growing status of alpine grassland.