Mapping change probabilities for local improvement of change detection in the Brazilian Cerrado

The Brazilian cerrado biome is under pressure, as an increasing part of it is converted to agriculture. In some parts, however, we see also that regeneration takes place. Because the cerrado is typically an open vegetation, it can be difficult to see the changes from cerrado to agriculture, as well as the difference between agriculture and regeneration, especially when agriculture includes perennial crops or crops with a longer growth cycle, such as cassava and sugarcane.

Rather than improving the overall classification of cerrado areas, this topic aims at first identifying the areas where the classification and/or the detected change are most uncertain and then focussing on decreasing the uncertainty in these areas, instead of looking for an overall classification improvement.

INPE is monitoring the cerrado biome with remote sensing and this research will contribute to that task. Time series remote sensing data from the INPE data cube can be used.

The topic can be focused on regeneration or degradation of the biome and is open to several approaches, so it can accommodate several MSc students. Good remote sensing knowledge and programming (e.g. Python or R) are a prerequisite. Knowledge of or interest in the processes of land use and land cover change in the cerrado are a plus.