RESUMEN
As a major component of the cryosphere, permafrost plays important roles in the climate system and land surface processes of the Earth. Owing to the rapidly warming climate, permafrost over the globe has degraded in recent decades. However, quantifying the distribution and temporal changes in permafrost is challenging. In this study, we modified the widely used surface frost number model by considering the spatial distribution of soil hydrothermal properties and then revisited the spatiotemporal patterns of permafrost distribution and its changes during the past decades (1961-2017) in China. We found that the modified surface frost number model performs well in simulating permafrost extent in China, with the overall accuracy and kappa coefficients being 0.92 and 0.78 in the calibration (1980s) and 0.94 and 0.77 in the validation period (2000s). Based on the modified model, we also found that permafrost extent in China demonstrated a significant decreasing trend over the past decades, especially on the Qinghai-Tibet Plateau (QTP), with a trend of -1.15 × 104 km2/yr (P < 0.01). Moreover, there is a significant relationship between ground surface temperature and permafrost distribution area, with the R2 being 0.41, 0.42, and 0.77 in NE and NW China and on the QTP. The sensitivity of permafrost extent to ground surface temperature in NE China, NW China, and the QTP, respectively, was -8.56 × 104, -1.97 × 104, and -34.60 × 104 km2/°C, respectively. Permafrost degradation has accelerated since the late 1980s, possibly due to increased climate warming. This study is of great significance for improving permafrost distribution simulation at large spatial scales (trans-regional) and for offering vital information for adapting to climate change in cold regions.