Seasonal leaf dynamics for tropical evergreen forests in a process-based global ecosystem model
Abstract. The influence of seasonal phenology on canopy photosynthesis in tropical evergreen forests remains poorly understood, and its representation in global ecosystem models is highly simplified, typically with no seasonal variation of canopy leaf properties taken into account. Including seasonal variation in leaf age and photosynthetic capacity could improve the correspondence of global vegetation model outputs with the wet–dry season CO2 patterns measured at flux tower sites in these forests. We introduced a leaf litterfall dynamics scheme in the global terrestrial ecosystem model ORCHIDEE based on seasonal variations in net primary production (NPP), resulting in higher leaf turnover in periods of high productivity. The modifications in the leaf litterfall scheme induce seasonal variation in leaf age distribution and photosynthetic capacity. We evaluated the results of the modification against seasonal patterns of three long-term in-situ leaf litterfall datasets of evergreen tropical forests in Panama, French Guiana and Brazil. In addition, we evaluated the impact of the model improvements on simulated latent heat (LE) and gross primary productivity (GPP) fluxes for the flux tower sites Guyaflux (French Guiana) and Tapajós (km 67, Brazil). The results show that the introduced seasonal leaf litterfall corresponds well with field inventory leaf litter data and times with its seasonality. Although the simulated litterfall improved substantially by the model modifications, the impact on the modelled fluxes remained limited. The seasonal pattern of GPP improved clearly for the Guyaflux site, but no significant improvement was obtained for the Tapajós site. The seasonal pattern of the modelled latent heat fluxes was hardly changed and remained consistent with the observed fluxes. We conclude that we introduced a realistic and generic litterfall dynamics scheme, but that other processes need to be improved in the model to achieve better simulations of GPP seasonal patterns for tropical evergreen forests.