Aplication of Dynamic Vegetation Model IBIS to the conditions of terra firme forest in Central Amazônia

It is critical for Dynamic Global Vegetation Models (the acronym DGVMs in English) to correctly represent the diurnal cycle, seasonal and interannual variations of mass and energy exchange between the ecosystem and the atmosphere, soil hydrology, phenology, and evolution of carbon storage in the ecosystem. This study attempts to validate IBIS DGVM for a terra firme tropical rainforest ecosystem, located in the Biological Reserve of Cuieiras, 50 Km North of Manaus. Nine years of data from the K34 flux tower, as well as data from biomass inventory and neutron probes, were employed to force the model and to validate the simulations of net radiation, latent heat, sensible heat, and carbon dioxide fluxes. To better represent soil water content, an adjustment was made to soil model physical parameters porosity, field capacity, wilting point and saturated hydraulic conductivity. Simulations were also carried out employing future climate change scenarios for the year 2099, by superimposing temperature increases and reduction of rainfall as projected by the climate model HADCM3. Also the IPCM4 model trends indicating a temperature rise and increased rainfall were employed. Comparing model simulations with observed data, results indicate that the model water balance components were well represented. Around 50% of the precipitation returned to the atmosphere by evapotranspiration and approximately 49% by drainage or runoff. The simulated soil water content for each specific layer was not well represented, overestimating some and underestimating others, but the integration of the five layers, down to 4.8 m depth, were well represented in the model, with a coefficient of determination R2 = 0.80. The model underestimated the net radiation due to the albedo and surface temperature overestimated (R ² = 0.99), but represented relatively well the latent heat flux (R ² =0.74) and the sensible heat flux R ² = 0.55. The CO2 fluxes (GPP, Reco and NEE) showed the expected magnitude, but not the seasonal variation (R ² = 0.54 for NEE, for example). However, uncertainties in the validation data were not assessed for analyses of possible seasonal bias. The model overestimated LAI and fine litter storage in approximately 50% and 100%, respectively, but underestimated the biomass carbon storage of about 30 tC / ha. The model also did not reproduce well the magnitude of biomass increase observed in permanent plots of forest inventories. In the simulations of future scenarios the energy balance and soil water content were very sensitive to temperature increase and changes in rainfall. The CO2 fluxes, carbon storage and LAI were sensitive to temperature and CO2 increases, but with a predominance of increased CO2 concentration. The tests also showed the importance of considering the rainfall regime as faithful as possible, especially the interannual climate variability on water balance, energy, and forest carbon. The interannual variability has a role in limiting the increase of carbon storage in the scenario of increased CO2. However, the results indicate the need to improve the parameterization of plant functional type representative of the tropical rainforest and the inclusion of a subroutine of severe events.