Peat-ML2
A new global benchmark for global peatland carbon inventories

Peatlands are among the world’s most critical ecosystems for climate regulation, storing twice as much carbon in their soils as in the world’s forests. Despite their importance, there are no accurate, spatially explicit and internally consistent global-scale assessments of peatland extent, distribution, and carbon storage. The lack of reliable peat carbon maps allows substantial carbon emissions from undetected degradation of these cryptic ecosystems. In our project, we will address the societal need for global peatland inventories through the production of two data products, leveraged from our earlier efforts in this area: 1) a self-contained and curated dataset of peat soil cores giving information about peatland presence, depth, and carbon content (Peat-DBase); and 2) a machine learning framework (Peat-ML 2) that uses Peat-DBase as training data alongside covariates of peatland formation and presence such as remotely sensed vegetation indices and spatially distributed climate, geomorphological, and soil data. Peat-ML 2 will be a high resolution global-scale benchmark of peatland distribution and carbon stocks, including robust accuracy assessments, confidence intervals, and spatial applicability measures. Our product will enable a new era of cost-effective peat-focused nature-based climate solutions.

Contact: Joe Melton (Environment and Climate Change Canada, Scott Winton (University of California Santa Cruz, USA)

We encourage peatland scientists to work with us to improve global peatland maps. Please contact us via this form

Benefits for data contributors

• Citeable DOI
• Collaboration opportunities
• Training opportunities in digital peatland mapping

Snapshot of Peat-DBase showing present distribution of peat cores. Background map of peatland extent provided by Peat-ML (Melton et al. 2022)