How to spatially assess the vulnerability of soil organic carbon in a thawing permafrost at the scale of the Arctic region?

Details

Supervisors

Opfergelt, Sophie ; Bogaert, Patrick

Faculty

Faculté des bioingénieurs

Degree label

Master [120] : bioingénieur en sciences et technologies de l'environnement, à finalité spécialisée

Abstract

enBecause of climate warming, the Arctic region’s permafrost (permanently frozen part of the ground) is thawing and the active layer’s (seasonally thawing part of the ground, above the permafrost) environmental conditions are changing. As the permafrost contains between 30 and 50% of the Earth’s land surface carbon stock, a huge amount of previously trapped carbon could become exposed to microorganisms and susceptible to decomposition (vulnerable). The vulnerability of this carbon (percentage possibly released into the atmosphere) is believed to be influenced by several factors such as the environmental conditions. This study aimed at developing a methodology for carbon vulnerability assessment based on environmental conditions and spatially assessing this vulnerability across the Arctic permafrost region. The results showed that, given enough carbon emission samples and well-characterised environmental conditions, simple linear regression models could work well in this context. More complex, non-linear or machine-learning regression models would need even more carbon emission samples to be reliable. Different simple linear regression models (Generalised Linear Model and β-regression) converged towards similar results. Vegetation and strong annual variation of environmental conditions were identified as key drivers of the carbon vulnerability. Indirect influences of other environmental conditions (such as temperature, moisture or energy balance) on the carbon vulnerability were also assumed. Central-East Asia, South Asia, South Canada and Alaska were identified as the regions with the highest carbon vulnerability, with 2.5 to 5% of the soil carbon likely to be emitted there. Local vulnerabilities up to 10 to 15% could also be observed. The global spatial patterns were believed reliable while the local quantification was proved uncertain. Limitations in the number and the nature of the carbon emission samples were identified as main contributors to this study’s uncertainty.