Uncertainties in global permafrost area extent estimates from different methods

Abstract

Previous permafrost extent estimates applied one or two methods to calculate the permafrost area, and the uncertainties between the methods were not assessed. Here, we apply seven methods to estimate and project global permafrost area extent and discuss the uncertainties of each approach. These methods are forced with output from CMIP6 and ERA5-Land, and we quantify the seven methods’ differences and uncertainties. During the historical period (1981–2010), the mean global permafrost area from multiple methods is 14.1 ± 4.5 × 10⁶ km², with differences ranging from 2.1% to 31.2%. The variability in future permafrost area extent degradation relative to the historical period based on different methods ranges from 1.8% to 34.7%. Uncertainties in permafrost area extent estimates can reach 35% based on different methods. Under various future emission pathways (e.g., SSP1-2.6, SSP2-4.5, SSP3-7.0, and SSP5-8.5), the worst-case scenario (SSP5-8.5) projects a permafrost extent of only 1.3–8.2 × 10⁶ km² for 2070–2099, corresponding to area decreases of 51.2%–86.9%. Spatially, permafrost near the lower-latitude permafrost boundary may completely disappear by the end of the 21st century, while degradation in the circum-Arctic, Qinghai–Tibet Plateau, and Antarctica will be smaller, but still exceed 50% under the highest emission scenario (SSP5-8.5). Compared to the temperature distribution of existing permafrost maps, the temperature at top of permafrost model and the surface frost index using ground temperature adjusted for snow methods perform best. However, compared to in-situ boreholes, two generalized linear model approaches have the best overall accuracy. These uncertainties using different methods are important to recognize in assessments of the future state of permafrost degradation.