Do dynamic global vegetation models capture the seasonality of carbon fluxes in the Amazon basin? A data‐model intercomparison

Abstract

To predict forest response to long-term climate change with high confidence requires that dynamic global vegetation models (DGVMs) be successfully tested against ecosystem response to short-term variations in environmental drivers, including regular seasonal patterns. Here, we used an integrated dataset from four forests in the Brasil flux network, spanning a range of dry-season intensities and lengths, to determine how well four state-of-the-art models (IBIS, ED2, JULES, and CLM3.5) simulated the seasonality of carbon exchanges in Amazonian tropical forests. We found that most DGVMs poorly represented the annual cycle of gross primary productivity (GPP), of photosynthetic capacity (Pc), and of other fluxes and pools. Models simulated consistent dry-season declines in GPP in the equatorial Amazon (Manaus K34, Santarem K67, and Caxiuanã CAX); a contrast to observed GPP increases. Model simulated dry-season GPP reductions were driven by an external environmental factor, 'soil water stress' and consequently by a constant or decreasing photosynthetic infrastructure (Pc), while observed dry-season GPP resulted from a combination of internal biological (leaf-flush and abscission and increased Pc) and environmental (incoming radiation) causes. Moreover, we found models generally overestimated observed seasonal net ecosystem exchange (NEE) and respiration (R ) at equatorial locations. In contrast, a southern Amazon forest (Jarú RJA) exhibited dry-season declines in GPP and R consistent with most DGVMs simulations. While water limitation was represented in models and the primary driver of seasonal photosynthesis in southern Amazonia, changes in internal biophysical processes, light-harvesting adaptations (e.g., variations in leaf area index (LAI) and increasing leaf-level assimilation rate related to leaf demography), and allocation lags between leaf and wood, dominated equatorial Amazon carbon flux dynamics and were deficient or absent from current model formulations. Correctly simulating flux seasonality at tropical forests requires a greater understanding and the incorporation of internal biophysical mechanisms in future model developments.

Novel Representation of Leaf Phenology Improves Simulation of Amazonian Evergreen Forest Photosynthesis in a Land Surface Model

“…Improved model representation of the region's carbon and water fluxes is important if we are to assess their ongoing and future changes (Fu et al, ; Peñuelas & Filella, ; Wright et al, ). However, most Earth system models (ESMs) still fail to accurately model the responses of Amazonian evergreen forest to seasonal and interannual climatic variations; this is in part due to the poor representation of tropical phenology in current ESMs (Restrepo‐Coupe et al, ; Wu et al, ).…”

The importance of tree demography and root water uptake for modelling the carbon and water cycles of Amazonia

“…CC BY 4.0 License. (Getirana et al, 2014;Guimberteau et al, 2014;Restrepo-Coupe et al, 2016), was unable to maintain LE and GPP fluxes during the dry season.…”

“…These uncertainties arise from both the atmospheric (Ahlström et al, 2012) and the land surface components (Booth et al, 2012;Sitch et al, 2015). In land models (dynamic global vegetation models, or DGVMs) large sources of uncertainty include the vegetation response to droughts (Restrepo-Coupe et al, 2016), and tree demographic processes (Fisher et al, 2010;Rödig et al, 2018). Most DGVMs simulate the effect of water shortage on plant functioning by lowering leaf gas exchange rates using a multiplicative 15 water stress factor that depends on soil moisture (Christoffersen et al, 2014) and by including atmospheric water stress from increased vapour pressure deficit in their parameterization of stomatal conductance.…”

“…Most DGVMs simulate the effect of water shortage on plant functioning by lowering leaf gas exchange rates using a multiplicative 15 water stress factor that depends on soil moisture (Christoffersen et al, 2014) and by including atmospheric water stress from increased vapour pressure deficit in their parameterization of stomatal conductance. With this simplification, models typically fail to capture tropical carbon and water flux seasonality (Poulter et al, 2009;Restrepo-Coupe et al, 2016), and vegetation response to drought (Joetzjer et al, 2014;Powell et al, 2013). A few global DGVMs have recently adopted a more explicit representation of the soil-plant-atmosphere water column [e.g., Bonan et al, 2014;Christoffersen et al, 2014], 20 but much research is still needed to fully model these processes.…”

Surface‐Atmosphere Coupling Scale, the Fate of Water, and Ecophysiological Function in a Brazilian Forest

“…Restrepo‐Coupe et al () noted an inverse relationship (or no relationship) between precipitation at wet sites (light limitation), while sites to the south and east exhibited a correlation between rainfall and photosynthesis (water limitation). Baker et al () reproduced this behavior generally with a model, but Restrepo‐Coupe et al () noted that many models are unable to do so. While quantitative understanding of spatiotemporal ecophysiology in response to seasonal and anomalous forcing is expanding, we do not consider the matter closed.…”