Permafrost variability over the Northern Hemisphere based on the MERRA-2 reanalysis

• Published in: The Cryosphere
• Main Authors: Tao, Jing, Koster, Randal D., Reichle, Rolf H., Forman, Barton A., Xue, Yuan, Chen, Richard H., Moghaddam, Mahta
• Format: Article in Journal/Newspaper
• Language: English
• Published: Copernicus Publications 2019
• Subjects: article; Verlagsveröffentlichung; Merra; Active layer thickness; permafrost; The Cryosphere; Alaska; Siberia
• Online Access: https://doi.org/10.5194/tc-13-2087-2019; https://noa.gwlb.de/receive/cop_mods_00000619; https://noa.gwlb.de/servlets/MCRFileNodeServlet/cop_derivate_00000586/tc-13-2087-2019.pdf; https://tc.copernicus.org/articles/13/2087/2019/tc-13-2087-2019.pdf

This study introduces and evaluates a comprehensive, model-generated dataset of Northern Hemisphere permafrost conditions at 81 km2 resolution. Surface meteorological forcing fields from the Modern-Era Retrospective Analysis for Research and Applications 2 (MERRA-2) reanalysis were used to drive an improved version of the land component of MERRA-2 in middle-to-high northern latitudes from 1980 to 2017. The resulting simulated permafrost distribution across the Northern Hemisphere mostly captures the observed extent of continuous and discontinuous permafrost but misses the ecosystem-protected permafrost zones in western Siberia. Noticeable discrepancies also appear along the southern edge of the permafrost regions where sporadic and isolated permafrost types dominate. The evaluation of the simulated active layer thickness (ALT) against remote sensing retrievals and in situ measurements demonstrates reasonable skill except in Mongolia. The RMSE (bias) of climatological ALT is 1.22 m (−0.48 m) across all sites and 0.33 m (−0.04 m) without the Mongolia sites. In northern Alaska, both ALT retrievals from airborne remote sensing for 2015 and the corresponding simulated ALT exhibit limited skill versus in situ measurements at the model scale. In addition, the simulated ALT has larger spatial variability than the remotely sensed ALT, although it agrees well with the retrievals when considering measurement uncertainty. Controls on the spatial variability of ALT are examined with idealized numerical experiments focusing on northern Alaska; meteorological forcing and soil types are found to have dominant impacts on the spatial variability of ALT, with vegetation also playing a role through its modulation of snow accumulation. A correlation analysis further reveals that accumulated above-freezing air temperature and maximum snow water equivalent explain most of the year-to-year variability of ALT nearly everywhere over the model-simulated permafrost regions.