Uncertainties in gridded air temperature fields and effects on predictive active layer modeling

Several model-based assessments predict a discernible increase in the depth of seasonal thawing and circumpolar-scale warming of permafrost by the mid-21st century. Quantitative estimates of the environmental and socioeconomic impacts of changing climate in northern regions require robust projection of changes in permafrost, which in turn depend on the availability of appropriate models and forcing data. We examined four high-resolution, hemispheric-scale gridded sets of monthly temperature and precipitation constructed using different interpolation routines and reanalysis of data from a large number of weather stations. At many of 455 Russian weather stations, the four data sets depart from empirical mean annual air temperatures averaged over the 15-year period by 1-2 degrees C and in cumulative daily positive temperature sums (degree days of thawing) by more than 200 degrees C days. A permafrost model, forced with the gridded climatic data sets, was used to calculate the large-scale characteristics of permafrost in northern Eurasia. We analyzed zonal-mean air and ground temperatures, depth of seasonal thawing, and area occupied by near-surface permafrost in Eurasia north of 45 degrees N. The 0.5-1.0 degrees C difference in zonal-mean air temperature between the data sets translates into a range of uncertainty of 10-20% in estimates of near-surface permafrost area, which is comparable to the extent of changes projected for the following several decades. We conclude that more observations and theoretical studies are needed to improve characterization of baseline climatic conditions and to narrow the range of uncertainties in model-based permafrost projections.