Imaging permafrost active layer thickness under forest for climate model improvement

Abstract

Permafrost is an important but poorly known carbon reservoir which is vulnerable to the high latitude accelerated warming. The projected thickening of its superficial seasonally thawed active layer and its induced spatial reorganization will hasten Carbon release in the atmosphere while impacting hydrology, geochemical transfers, vegetation repartition and ground stability. Active layer thickness (ALT) is only assessed by northern stations, therefore, its spatial distribution remains unknown and lacks for model evaluation, especially under the boreal forest. The all-weather spaceborne InSAR technique has shown only partial sensitivity to ALT through ground movements and remained restricted to non-forested areas. To overcome these limitations, we generalized the ground movement estimation under the omnipresent forest by exploiting the SAR polarimetric information, on the one side, and we isolated the thermodynamical component from the hydrological one during freezeback using a land surface model, on the other side, to extract ALT. Based on a one year TerraSAR-X time serie acquired over the region of Yakustk, we obtained a first high resolution ALT image which reveals unexpected short scale spatial heterogeneity, arranged along anisotrotopic patterns. Its poor comparison with the ALT simulated by the ISBA land surface model, currently used in climate modeling, highlights that climate models, and thus their simulations of greenhouse gas emissions, remain very uncertain over northern high latitudes in absence of regionalized ALT information under the boreal forest. This novel approach, operable using current and future sensors over wide areas, offers a new way forward to improve modeling as well as to optimally monitor global warming from the high latitudes.