Abstract
Bud phenology identifies the growing period of trees and determines the pattern of mass and energy exchanges between forest and atmosphere over time and space. Canopy color metrics derived from phenocams have been widely used to investigate tree phenology. However, it remains unclear which color-based index better tracks the seasonal variations of tree phenology in evergreen forest ecosystems. Herein, we compared four color metrics (red chromatic coordinate (RCC), green chromatic coordinate (GCC), vegetation contrast index (VCI) and excess green index (ExG)) derived from phenocam images with bud phenological phases recorded in black spruce [Picea mariana (Mill.) B.S·P] during 2017–2020 at a boreal forest site in Quebec, Canada. Canopy redness (RCC) and greenness (GCC, ExG, and VCI) showed a bimodal and bell-shaped seasonal pattern, respectively. The phases of bud burst and bud set lasted from end-May to end-June and from mid-July to end-September, respectively. The neural network model indicated that GCC had the best predictive ability in capturing the sequential phases of bud phenology. Bud phenological phases predicted by GCC showed the highest correlation with actual bud phenological phases among four indices, with R2 above 0.9 and RMSE lower than 0.5. Overall, color indices performed better when representing bud burst than bud set. Our findings improve the efficiency and confidence of the phenocam greenness index to characterize the growing season of evergreen forests.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call