Publicações

Global dynamic vegetation models are needed to valuate the synergistic effects of changes in climate and fire activity in Amazonia. Coupled land-surface and climate models, for example, are being improved to evaluate the vulnerability of Amazon rainforest to more frequent and severe droughts, either through a direct effect on tree mortality or indirectly via increased occurrence of vegetation fires. In this context, we are working on improving the HadGEM2-ES (UK Met Office Hadley Centres Global Environmental Model version 2 – Earth System Model) to better represent land-atmosphere interactions over South American biomes, including fires. To this end, a fire dynamics sub-model is being developed for evaluating fire occurrence and impacts considering major natural and anthropogenic factors in the region. Current fire equations are based on methods already tested in global dynamic vegetation models, and lead to reasonable representation of major spatial and temporal features of the global fire occurrence. At large scale there is correct representation of time and location for most of the burned area reported in datasets based on remote sensing. However, important under- and over-estimation of model results occur at smaller scales, in part explained by the simplicity of equations and current parametrization based on spatial and temporal averages of reference fire data. We are now working to enhance the precision of the fire equations and to improve the representation of the effects of fire on vegetation and atmospheric composition, including better representation of sources of ignition, changes in vegetation mortality rates, and emissions due to fire. These improvements will then contribute for evaluating the signal and strength of potential fire-climate feedbacks in the region.