Felix Beer, Universität Greifswald

Peatlands are permanent wetlands that provide manifold important ecosystem services. They stock twice as much carbon as all forest globally and are important for regulating water flows in freshwater networks. But peatlands are under threat on a global scale by degradation through unsustainable land use. Drainage turns peatlands from a carbon sink to a carbon source, now contributing with more than 5 percent to global greenhouse gas emissions.

In the Cerrado biome, the world’s largest and most biodiverse tropical savannah, peatlands are part of ‘veredas’, i.e. stream valleys that drain from the central Brazilian plateaus towards some of Brazil’s main rivers as a complex headwater network. Industrial cash crop production has already destroyed 50% (1 million km²) of Cerrado’s natural vegetation, driven by market demands from G20 countries, of which the German demand counts for 16%. Severe changes in the carbon and water cycle and losses of biodiversity with strong negative impacts on livelihoods result. Although headwater wetlands may cover up to 30% of the Cerrado and locally store large amounts of carbon, their distribution, extent, condition, and carbon stocks remain poorly known.

By combining ecological field work with latest remote sensing analyses this research project aims at shedding light at the role of peatlands and stream valley wetlands in the carbon and hydrological cycle at federal state levels in Cerrado. Time series of Sentinel-1 and -2 will be combined in a Convolutional Neural Network (aka Deep Learning) to use the natural seasonal dynamics and spatial patterns for mapping of wetlands and peatlands. Ground truthing is going to provide the necessary validation and ecological data for a better understanding of the ecosystems in the field. Their degree of degradation and potentials of conservation will be assessed subsequently to evaluate the threats that current land use changes and climate change are posing to these important ecosystems.