Date of Award

Fall 2004

Project Type


Program or Major

Earth and Environmental Science

Degree Name

Doctor of Philosophy

First Advisor

George Hurtt


Fires are major disturbances for ecosystems in Amazonia. Because of their short time scale and strong links to biogeochemical cycles, fires significantly affect fluxes and stocks of carbon and nutrients, change air composition, and modify ecosystems structure and functioning. Fires are strongly related to land-use, land-cover and climate conditions. Because the increasing development of the region, these factors have been changing and leading to different patterns of fire activity. Thus it is very important to understand the dynamics of fires in the region and to develop models that can project their potential changes. To contribute in all these subjects, we used fieldwork, remote sensing and modeling studies. These studies are presented here organized in four chapters. In the first, we review several studies representing different methods for fire research with emphasis in their synergy. We reinforced the benefits of high-resolution fire information for large domains based on multiple methods, and discuss the challenges and perspectives for producing such datasets. The second chapter presents a new fire model able to reproduce the large-scale patterns of fire activity and suitable for coupling to ecosystem models for the region. Applying this model to scenarios of development, we found that that substantially increased and altered fire patterns are likely to result from future land-use practices without new efforts to mitigate fire activity. The third chapter presents a new method to enhance the interpretation of satellite fire data based on passive ground-based analyses. Using that method we found that the total accuracy for two widely used fire products is very high and dominated by accurate non-fire detection; fire-detection accuracy is lower, and errors of commission were less than errors of omission. In the fourth chapter, detailed relations between fire activity and changes in land use and cover were determined using recent land-cover transitions data. The relations we found confirmed results from independent studies where fire activity was positively correlated with fragmentation and losses of forest area. In addition, we found indications that the removal of forest biomass and the age of forest conversion are dominant factors.