Open AccessGeomorphological controls over carbon distribution in permafrost soils: the case of the Narsajuaq river valley, Nunavik (Canada)
Author(s) -
Serge Gag,
Michel Allard
Publication year - 2020
Publication title -
arctic science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.672
H-Index - 4
ISSN - 2368-7460
DOI - 10.1139/as-2019-0026
Subject(s) - permafrost , landform , soil carbon , soil water , physical geography , environmental science , carbon fibers , carbon cycle , hydrology (agriculture) , total organic carbon , soil science , geology , geomorphology , environmental chemistry , geography , ecology , oceanography , ecosystem , materials science , geotechnical engineering , chemistry , biology , composite number , composite material
Soils in the northern circumpolar region play a central role in the global carbon cycle because the release of carbon through permafrost thaw and geomorphological disturbances can potentially cause a feedback on climate. However, large uncertainties in estimates of permafrost carbon stocks remain, mainly because of wide gaps in the spatial coverage of soil carbon sampling sites and the large mapping polygons used to upscale data. By combining mapping of landforms and knowledge of surficial geology to upscale soil carbon content measurements, we provide an assessment of soil total carbon content in the region of the Narsajuaq river valley (Nunavik, Canada) to generate the first high-resolution soil carbon estimate confirmed by field measurements in Nunavik. We estimate that the Narsajuaq river valley and the surrounding uplands have a weighted average of 3.4 kg C m −2 (0–100 cm), with 73% of the total carbon content stored in the top 30 cm. The results also indicate that the valley is a carbon hotspot in the region, containing 76% of the total carbon content (0–100 cm) of the study area. Although soil carbon estimates will always require field sampling, the geomorphological mapping approach can significantly improve carbon content estimates and provide better inputs for models.