Joint Seminar: 3D geospatial mapping of Arctic permafrost carbon
Considerable uncertainties persist in accurately mapping Arctic permafrost carbon on a circumpolar scale, increasing at greater soil depth. Quantifying the permafrost soil carbon pool has shown to be particularly challenging, due to the large spatial and vertical variability of soil properties in the Arctic environment and the scarce availability of soil reference data. Although the northern circumpolar permafrost region makes up only 15% of the global soil area, it contains up to half of the global soil organic carbon pool and twice as much carbon as currently in the atmosphere. At the same time, the Arctic is warming two to four times faster compared to the global average, leading enhanced greenhouse gas emissions from the permafrost. This makes permafrost carbon a potentially strong climate feedback that could further amplify global warming.
Here, we aim to provide updated circumpolar soil carbon maps using geospatial modelling, comparing two digital soil mapping (DSM) approaches – 3D versus 2.5D – in terms of predictions of SOC density over depth and prediction accuracy. We predicted SOC density up to 3 m depth using random forest models based on covariates from a digital elevation model and climate data. However, calibrating and evaluating our model required detailed and accurate soil reference data. Most existing permafrost carbon datasets do not serve that purpose, as they lack in structure, resolution and additional metadata. Therefore, we developed a new harmonized, standardized, structured, extensive, analysis ready and FAIR soil profile reference database for the northern circumpolar permafrost region, called the Circum-Arctic Soil Permafrost Region database (CASPeR).
To our knowledge, CASPeR is already the largest structured general purpose permafrost soil database for the Arctic permafrost region, containing nearly 1,000 profiles and approximately 14,000 samples. CASPeR will also be used for validation as part of the permafrost downstream task in the AWI 3D-ABC Helmholtz Foundation Model initiative to develop an AI model to quantify the current state of global terrestrial carbon stocks. Results from our modelling indicate that the two approaches perform similarly, with 3D DSM showing no significant improvement over 2.5D DSM. Evaluation also showed that our model explains about half the variation in the upper 1 m of soil, with decreasing accuracy in deeper soil layers. We attribute this to the limited predictive power of the covariates currently included, and the lack of reference data in deeper soil layers. Nevertheless, 3D DSM offers advantages, with evidence that depth is an important predictor of SOC. Future studies should investigate additional covariates and harmonize more reference data, including for deeper soil layers.
Datum
21.05.2025
Uhrzeit
13:30–14:30 h
Ort
- Bundesstr. 53, room 022/023
- Seminar Room 022/023, Ground Floor, Bundesstrasse 53, 20146 Hamburg, Hamburg