Publicações

Deforestation of tropical forests for logging and agriculture, associated to slash-and-burn practices, is a main contributor of CO2 emissions, both immediate due to biomass burning and future due to the elimination of a potential CO2 sink. Moreover, climate change in the Amazon during the 21st century is projected to decrease the resilience of the Amazon forest and possibly replace large parts of tropical forests with savannas, adding to the direct anthropogenic deforestation. The potential feedbacks between LUCC (Land-Use and Land-Cover Change) and climate change are thought to further increase the loss of tropical forests and increase the rate of CO2 emissions, through mechanisms such as land and soil degradation and the increase in wildfire occurrence and severity. However, current understanding of the processes of fires (including ignition, spread and consequences) in tropical forests and feedbacks with climate are poorly understood and need further research. As the processes of LUCC and associated fires occur at local scales, linking them to large-scale atmospheric processes requires a means of up-scaling higher resolutions processes to lower resolutions. Our approach is to couple models which operate at various spatial and temporal scales: a Global Climate Model (GCM) (and possibly a Regional Climate Model; RCM), Dynamic Global Vegetation Model (DGVM) and local-scale LUCC model and fire behavior model. The climate model(s) resolve larger scale atmospheric processes and forcings, which are felt by the surface DGVM and fed-back to the climate. Higher-resolution processes such as deforestation, land use management and associated (as well as natural) fires are resolved at the local level. Anthropogenic LUCC (such as deforestation and pasture burning) is modeled using Geographical Information Systems (GIS) and spatially-explicit models. Modifications to the land surface are imposed on the land model, which has a potential feedback on the land surface properties through processes resolved by the atmospheric and land models. Fire behavior is modeled at the regional scale (~100m) to represent the detailed landscape using semi-empirical or empirical fire spread models. Improved parametrizations from our fully-coupled experiments, for use in atmospheric and land surface models, will allow for a better understanding and evaluation of CO2 emissions from LUCC and feed-backs with climate change.