Changes in land surface properties, often associated with changing vegetation and soil characteristics, have dramatic effects on the climate simulated by general circulation models (GCMs). Thus, during the last several years efforts are aimed to evaluating land surface parameterizations. Currently, there is a diversity of soil-vegetation-atmosphere schemes. The objectives of these models are mainly in the improvement of bottom boundary condition of the atmospheric models by accurately evaluating the ground-surface impact on the atmosphere. The surface schemes ranges from those that simulate only the exchange of momentum, energy and water vapor between vegetated surface and atmosphere, as the "Simplified Simple Biosphere Model" – SSiB, even more sophisticated models, as, for example, the "Integrated Biosphere Simulator Model" – IBIS, which includes, in addition to processes listed above, the terrestrial carbon and nutrient cycling, and vegetation dynamics. The SSiB is coupled to the global and regional meteorological models of the Brazilian Center for Weather Forecasting and Climate Studies/Brazilian National Institute for Space Research (CPTEC/INPE). It includes a complex treatment of the albedo, the energy balance of the surface and the soil moisture. Recently, the IBIS-2.6 model of dynamical vegetation was also coupled to the GCM/CPTEC to make it able to diagnose the loss of biomass and its impact on storage of energy and water. To test the surface modelxs performance, different simulations are been proposed. The goal of this work is evaluating the energy balance simulated by SSiB and IBIS models against two sets of measurements. The data were collected at a tropical rainforest in the Brazilian Amazon and in the caatinga in Northeast Brazil.

The dataset for Amazonian forest used in this work was collected at the Reserva Biológica do Jaru (62º22'W; 10º45'S) forest site, located in the state of Rondônia, Brazil. Micrometeorological and hydrological data were collected during 2001, as part of the Large-Scale Biosphere-Atmosphere Experiment in Amazonia – LBA.

The caatinga is a type of vegetation adapted to the semi-arid conditions, with varied physiognomy, and covers the northeast portion of Brazil. It occupies an extension of about 800,000 km2 that corresponds to 70% of the region. Micrometeorological and hydrological data were collected from July 2004 to June 2005, in a site located in the Agricultural Research Center of the Semi-Arid Tropic – CPATSA (9°03' S; 40°19' W; 350 m, Petrolina, Pernambuco, Brazil.

In this study, the surface schemes were used in the so-called “off-line mode”. In this version, the models were not coupled with the atmospheric model and were driven using prescribed atmospheric forcing. The results of the energy balance simulated by both models and for each vegetation type were compared.

The results showed that the net radiation was well simulated by SSiB model for both caatinga and forest sites, while the latent heat fluxes were overestimated for both sites. In contrast, the sensible heat flux was underestimated for forest and overestimated for caatinga, mainly during the night period; notwithstanding, the values for forest were nearer to the observed ones.

With the IBIS model, the simulations for the radiation balance were close to the observed data for forest site, in contrast to the obtained for caatinga site. The latent heat flux and the sensible heat flux were underestimated by the model for caatinga. For forest site, the energy partition was not well simulated; thus, the SSiB model provides better results. It is highlighted that the simulations with the IBIS model were made considering the parameters previously calibrated for both forest and caatinga. Thus, these results are preliminary and other analyses are still necessary to better understand the biophysical processes involved. ion balance were close to the observed data for forest site, in contrast to the obtained for caatinga site. The latent heat flux and the sensible heat flux were underestimated by the model for caatinga. For forest site, the energy partition was not well simulated; thus, the SSiB model provides better results. It is highlighted that the simulations with the IBIS model were made considering the parameters previously calibrated for both forest and caatinga. Thus, these results are preliminary and other analyses are still necessary to better understand the biophysical processes involved.