Satellite-based modeling of permafrost temperatures in a tundra lowland landscape

- Thermal state of permafrost & thaw depth can be modeled using satellite data.
- The accuracy of MODIS LST & GlobSnow SWE is adequate for modeling permafrost.
- Unknown snow properties and soil composition induce large uncertainties.

Remote sensing offers great potential for detecting changes of the thermal state of permafrost and active layer dynamics in the context of Arctic warming. This study presents a comprehensive feasibility analysis of satellite-based permafrost modeling for a typical lowland tundra landscape in the Lena River Delta, Siberia. We assessed the performance of a transient permafrost model which is forced by time series of land surface temperatures (LSTs) and snow water equivalents (SWEs) obtained from MODIS and GlobSnow products. Both the satellite products and the model output were evaluated on the basis of long-term field measurements from the Samoylov permafrost observatory. The model was found to successfully reproduce the evolution of the permafrost temperature and freeze-thaw dynamics when calibrated with ground measurements. Monte-Carlo simulations were performed in order to evaluate the impact of inaccuracies in the model forcing and uncertainties in the parameterization. The sensitivity analysis showed that a correct SWE forcing and parameterization of the snow's thermal properties are essential for reliable permafrost modeling. In the worst case, the bias in the modeled permafrost temperatures can amount to 5 °C. For the thaw depth, a maximum uncertainty of about ± 15 cm is found due to possible uncertainties in the soil composition.