Comment on egusphere-2025-397 - No compliance with the policy of the journal

Résumé fourni par la source

Abstract. Uncertainty in the dynamics of Amazon rainforest poses a critical challenge for accurately modeling the global carbon cycle. Current dynamic global vegetation models (DGVMs), which use one or two plant functional types for tropical rainforests, fail to capture observed biomass and mortality gradients in this region, raising concerns about their ability to predict forest responses to global change drivers. Here we assess the importance of spatially varying parameters to resolve ecosystems spatial heterogeneity in the ORCHIDEE (ORganizing Carbon and Hydrology in Dynamic EcosystEms) DGVM. Using satellite observations of gross primary productivity (GPP), tree aboveground biomass (AGB) and biomass mortality rates, we optimized two key parameters: the alpha self-thinning (α), which controls tree mortality induced by light competition, and the nitrogen use efficiency of photosynthesis (η), which regulates GPP. The model incorporating spatially optimized α and η parameters successfully reproduces the spatial variability of AGB (R2=0.82), GPP (R2=0.79), and biomass mortality rates (R2=0.73) when compared to remote sensing observations in intact Amazon rainforests, whereas the model using spatially constant parameters has R2 values lower than 0.04 for all observations. Furthermore, the relationships between the optimized parameters and ecosystem traits, as well as climate variables were evaluated using random forest regression. We found that wood density emerges as the most important determinant of α, which are in line with existing theory, while water deficit conditions significantly impact η. This study presents an efficient and accurate approach to enhancing the simulation of Amazonian carbon pools and fluxes in DGVMs by assimilating existing observational data, offering valuable insights for future model development and parameterization.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.