Spatializing Net Ecosystem Exchange in the Brazilian Amazon biome using the JULES model and vegetation properties

Abstract. The large extension and diversity of the Brazilian Amazon biome hampers the assessment of the regional-scale carbon budget based solely on local observations. Considering the shortage of observations, this study aims to examine the carbon fluxes throughout the Brazilian Amazon biome using a process-based model (JULES, Joint UK land environment simulator). A sensitivity analysis detected five critical model parameters for the Amazon tropical broadleaf evergreen forest, optimized using carbon flux and meteorological data from four forest sites. The simulations with the new parametrization were compared with JULES default parameter values and with simulations of the Vegetation Photosynthesis and Respiration Model (VPRM). Net ecosystem exchange (NEE) and gross primary production (GPP) estimates were improved at all sites, reaching a Root Mean Squared Error (RMSE) about 30 % lower in comparison to the default version. The optimized parameter values varied among the four sites, indicating that a single parameterization for the whole Amazonia may not be adequate. JULES model parameters were spatialized for the Brazilian Amazonia, based on canopy height and leaf area index gridded data. Applying JULES with spatially dependent parameterization for the year 2021 resulted in a carbon sink of −1.34 Pg C yr−1. Regional differences were observed in the carbon fluxes, with a carbon source of 0.75 kg C m−2 yr−1 in the southwest and north, likely explained by increased ecosystem respiration in older and taller forests.