The role of the Amazon forest as a carbon sink has been extensively studied. However, recent results show that this sink is declining possibly due to climate variability, driving an increase in tree mortality, with feedbacks between growth and death of trees. Even though forest inventory plots have been the main source of information to monitor and assess impacts of global changes on ecosystems, there are uncertainties on the stability of extrapolations from local to regional scales. One way to improve the accuracy of these estimates is to use remote sensing data to extent the measurement area and analyze forest dynamics at several spatial scales. The main goal of my project is to assess up-scaling techniques of local-scale dynamics and biomass loss estimates from inventory plots to multiple-scales based on remote sensing data. To achieve this aim, there is a need for a more precise understanding on what remote sensing data can detect in higher spatial scales taking into account the processes in plot scale. During the internship in University of Leeds, advised by Dr. Oliver Phillips, I propose to (1) acquire a database of long-term field data from RAINFOR network sites over Amazonia and calculate field-based metrics related to tree mortality and biomass loss; (2) develop and validate a method for automatic tree mortality detection using high-resolution multi-date images; (3) assess the satellite-based tree mortality detection and dead biomass estimates as a function of factors related to the trees (size of trees, position in the canopy, mode of death) and environment (Amazon regions, precipitation, and water deficit); and (4) assess the satellite tree mortality in function of spatio-temporal scales of the detection.
News published in Agência FAPESP Newsletter about the scholarship: