Since the end of the "stagnation period" of global warming in 2013, the global near surface temperature has rapidly increased, causing changes in the permafrost state over the years. The lack of temperature data on the permafrost roof before and after the stagnation period for ten years has constrained human understanding of the changes in permafrost status. This study used high-precision soil surface temperature data to drive the TTOP model to produce permafrost roof temperatures for the years 2003-2013 and 2014-2023, and validated the accuracy using station observation data. The results showed that the root mean square errors of permafrost roof temperatures (areas below 0 ℃ are permafrost regions) for the years 2003-2013 and 2014-2023 were 1.52 ℃ and 1.74 ℃, respectively, which were better than the approximate period (2000-2016) permafrost roof temperature data (root mean square error of 1.89 ℃) produced by other parameter data driven TTOP models in existing research. Especially near the permafrost boundary (areas where permafrost roof temperatures are close to 0 ℃), the accuracy was significantly improved. The data can be used to distinguish the range of permafrost in the Arctic region and analyze the trend of temperature changes in the permafrost roof.
The circum-Arctic soil surface temperature data were obtained from Guo et al. (2024), who produced monthly high-precision soil surface temperature data for 2003–2023 using a multi-factor integration and month-by-month modeling approach (https://doi.org/10.1016/j.jag.2024.104114). Land cover type data were derived from the European Space Agency’s GlobCover product (https://due.esrin.esa.int/page_globcover.php). Soil r-factor data were obtained from the lookup table published by Obu et al. (2019) (https://doi.org/10.1016/j.earscirev.2019.04.023).
（1）Based on the soil surface temperature data, the mean surface freezing index and thawing index were calculated for the periods 2003–2013 and 2014–2023, respectively.
（2）Following the lookup table published by Obu et al. (2019), r-factor values were assigned according to different land cover types to generate the circum-Arctic r-factor dataset.
（3）The surface freezing and thawing indices, together with the r-factor data, were used to drive the TTOP model to compute permafrost top temperatures for 2003–2013 and 2014–2023 (regions with temperatures below 0 °C are defined as permafrost areas).
The accuracy of the generated dataset was validated using in-situ observations from test stations, with the root mean square error (RMSE) and coefficient of determination (R²) used as evaluation metrics.
Validation results indicate that the RMSEs of the permafrost top temperature for 2003–2013 and 2014–2023 are 1.52 °C and 1.74 °C, respectively—superior to the RMSE of 1.89 °C reported in existing studies that used other parameter datasets to drive the TTOP model for a comparable period (2000–2016). Notably, accuracy near the permafrost boundary (where permafrost top temperatures approach 0 °C) shows a significant improvement.
| # | number | name | type |
| 1 | 2020YFA0608501 | Research on Arctic Terrestrial Environmental Change and Its Effects | National key R & D plan |
This work is licensed under
CC BY 4.0 (Creative Commons Attribution 4.0 International License).
CSTR:11738.11.NCDC.ARCTIC-CHANGE.DB7142.2026
10.12072/ncdc.arctic-change.db7142.2026
