Research

Wetlands as a solution to climate change

Wetlands across the globe can have very different forms, but are all characterized by extended periods with water at or near the surface. These conditions result under natural conditions often in a specific vegetation, but apart from their nature value, wetlands serve many other ecosystem services such as flood protection and biodiversity, and can be important for food production, e.g. in rice cultivation. Additionally, wetlands play a crucial role in climate change mitigation and adaptation by acting as carbon sinks and regulating local climates.

The wet conditions in wetlands generate an oxygen-poor environment that in many places has led to long-term accumulation of organic carbon, which places wetlands amongst the ecosystems with highest carbon densities¹.

However, in the last centuries, global wetland area has decreased by over 20%² and many wetlands have been degraded³, mainly as a result of land use-imposed drainage and peat mining, impeding their function as carbon stores. This makes protection and restoration of wetlands a promising strategy for avoiding or reducing emissions of carbon dioxide (CO₂) through sequestration of organic matter.

However, while rewetting potentially reduces the emissions of CO₂, it may also enhance emissions of methane (CH₄) and nitrous oxide (N₂O), two greenhouse gases that are preferentially produced under oxygen-poor conditions. Similarly, short-term water table management in cultivated wetlands can lead to unwanted emissions. To support decisionmaking on management and restoration of wetlands, quantitative information on wetlands’ net greenhouse gas balances is urgently needed.

Estimating wetland greenhouse gas budgets

To estimate greenhouse gas emissions from wetlands, it is crucial to understand the link between biogeochemical processes and local hydrology. Greenhouse gas flux measurements can help to quantify the balance for an individual wetland, but the number of sites with such measurements is limited. Extrapolation of these measurements in space and time is best achieved with accurate understanding of the underlying processes. To support process understanding, we focus on measurements of the belowground conditions and their impacts on biogeochemical processes.

Understanding the changes in wetland hydrology and ecosystem processes on seasonal-decadal timescales is greatly supported by an increasing availability of Earth Observation data. We will develop specific wetland products that can help to quantify the complex landscape-scale variations and management impacts. We will exploit Machine Learning to derive relevant ecosystem properties from high-resolution imagery, with a focus on developing methods to assess the near-surface water conditions to support understanding of wetland biogeochemistry.

Given the strong impact of wetlands on the global budgets of carbon dioxide⁴, methane⁵ and nitrous oxide⁶, an accurate representation of wetland processes in global-scale models and assessments is of crucial importance. For simulation of local-scale biogeochemical processes, mechanistic models require detailed information and parameterization and are hence of limited use for large-scale assessment of greenhouse gas budgets. On the other hand, the generic nature of state-of-the-art global-scale models makes these unsuited for policy advice: They operate at a spatial resolution that does not allow for a representation of landscape-scale variations in topography, land use and wetland management. A detailed representation of the landscape in our models will support the understanding of lateral water movement and hotspots of greenhouse gas exchange.

Model development in the Global Wetland Center will leverage recent developments in Artificial Intelligence and Computer science. The integration of mechanistic modelling and data-driven alternatives will result in process-guided Machine Learning⁷ and will enable integration of the wealth of Earth Observation data. The use of differentiable programming⁸ will enable valuable developments for parameter estimation and uncertainty analysis.

In addition, the use of GPU systems will empower parallel computing, thereby speeding up the computations and enabling global simulations at a much higher spatial resolution than hitherto possible. In this way, high-resolution Earth Observation data can support the representation of landscape-scale variability in large-scale models.

Tools to support policymaking for wetland management

With the tight integration of Earth Observation data in the modelling of hydrology and biogeochemistry, and the combination of mechanistic and data-driven models, the Global Wetland Center will provide methodological developments that can serve as examples for other applications in the Environmental sciences.

Moreover, wetlands can take a key role in policymaking: With the United Nation’s focus on ecosystem restoration, the European Union’s strategy on rewetting of organic agricultural soils, or the focus on climate-smart irrigation strategies in rice, wetlands move into the spotlight of policymakers.

The novel approaches for modelling wetlands’ greenhouse gas budgets developed by the Global Wetland Center can guide wetland management as contribution to the global agenda towards climate neutrality. Through close collaboration with the UNEP-DHI centre on Water and Environment and other stakeholders, we will ensure that our Earth Observation-based products and results from hydrological and biogeochemical modelling will benefit policymakers and nature developers. 

References

1. Lal, R. Carbon sequestration. Philosophical Transactions of the Royal Society B: Biological Sciences 363, 815–830 (2008).
2. Fluet-Chouinard, E. et al. Extensive global wetland loss over the past three centuries. Nature 614, 281–286 (2023).
3. Courouble, M. et al. Global Wetland Outlook: Special Edition 2021. www.global-wetland-outlook.ramsar.org/ (2021).
4. Friedlingstein, P. et al. Global Carbon Budget 2023. Earth Syst Sci Data 15, 5301–5369 (2023).
5. Saunois, M. et al. Global Methane Budget 2000–2020. Earth Syst Sci Data Preprint (2024) doi:10.5194/essd-2024-115.
6. Tian, H. et al. Global nitrous oxide budget (1980–2020). Earth Syst Sci Data 16, 2543–2604 (2024).
7. Shen, C. et al. Differentiable modelling to unify machine learning and physical models for geosciences. Nat Rev Earth Environ 4, 552–567 (2023).
8. Gelbrecht, M., White, A., Bathiany, S. & Boers, N. Differentiable programming for Earth system modeling. Geosci Model Dev 16, 3123–3135 (2023).