High-resolution digital mapping of soil organic carbon in permafrost terrain using machine learning: a case study in a sub-Arctic peatland environment

Siewert, Matthias B.

doi:https://doi.org/10.5194/bg-15-1663-2018

Articles | Volume 15, issue 6

https://doi.org/10.5194/bg-15-1663-2018

© Author(s) 2018. This work is distributed under
the Creative Commons Attribution 4.0 License.

Special issue:

Changing Permafrost in the Arctic and its Global Effects in...

https://doi.org/10.5194/bg-15-1663-2018

© Author(s) 2018. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 15, issue 6

Research article

|

21 Mar 2018

Research article |

| 21 Mar 2018

High-resolution digital mapping of soil organic carbon in permafrost terrain using machine learning: a case study in a sub-Arctic peatland environment

Matthias B. Siewert

Supplement

https://doi.org/10.5194/bg-15-1663-2018-supplement

Data sets

High-resolution land cover classification for Stordalen, Abisko region, northern Sweden M. B. Siewert https://doi.org/10.1594/PANGAEA.886298

High-resolution maps of soil organic carbon for Abisko, northern Sweden M. B. Siewert https://doi.org/10.1594/PANGAEA.886296

Soil organic carbon storage and organic layer depth for the Stordalen catchment, Abisko, Sweden M. B. Siewert https://doi.org/10.1594/PANGAEA.887418

Download

Article (5197 KB)
Full-text XML

Short summary

Large amounts of soil organic carbon are stored in the circumpolar permafrost region. This article aims to improve how we map this carbon. Typically the amount of soil organic carbon is estimated using soil or land cover maps. Here the amount of carbon is modeled using machine learning. This is done at a very fine spatial resolution of 1 × 1 m. This reveals a lot of small-scale landscape variability and underlines the importance of permafrost-related landforms vulnerable to a warming climate.