3.3. The projection of MAGT and ALT
In view of a strong statistical rule of MAGT and ALT in climatic factors (e.g., TDD and FDD) and topographic factors (e.g., Lon, Lat, and Ele), most studies have begun to use similar statistical methods to investigate the present and future development trends of the periglacial climate realm (Koven et al., 2013; Aalto et al., 2017, 2018; Zhang et al., 2019). In this study, the optimal fitting model for the present state was employed to forecast MAGT and ALT under different future climate scenarios. For ALT, the spatial domain was limited to the area with simulated MAGT ≤ 0°C during each associated period and/or RCP scenario.
Due to climate change, the permafrost temperature exhibits an obvious rising trend on the QTP. We simulated the future change of permafrost on the QTP after half a century. The results revealed that the future changes of MAGT and ALT are predicted to be pronounced, but region-specific (Figure 6). The forecasted average MAGT over the QTP permafrost regions will increase from -1.35°C (present status) to -0.66°C by 2061−2080 (RCP2.6) and to 0.25°C for RCP8.5 (Table 2). The ALT, however, was only predicted to increase from 2.3 m (2000−2015) to 2.7 m (2061−2080) for RCP8.5. The reason for the consistency or small change of the ALT is that, the section of the permafrost with a MAGT near 0°C is forecasted to degrade to seasonally frozen ground, and this part of the permafrost usually corresponds to a thicker active layer. Additionally, the uncertainties related to the forecasts of MAGT and ALT under different RCPs in the future were given. And, the uncertainties are characterized by the range of MAGT value and ALT value. As can be seen in Figure 7, even under the different RCPs scenarios, the fluctuation range of MAGT and ALT is basically the consistent.
Over the next half century, the near-surface permafrost areas are predicted to continue to decrease by 0.13 × 106km2 (12%), 0.42 × 106km2 (40%) and 0.60 × 106km2 (58%) on the QTP by 2070 (2061−2080), under the RCP2.6, RCP4.5 and RCP8.5 scenarios, respectively. The result is basically consistent with the projected change by Chang et al. (2018) (Figure 8). Permafrost is in non-equilibrium under the influence of climate change, and there may be no permafrost that is driven by the current climate. In fact, it may be that permafrost is degrading, so the distribution range of the simulation results may be underestimated (Zhao et al., 2019). The changes in MAGT and ALT are not only related to the changes in temperature and precipitation but also closely related to hydrothermal parameters and surface energy balance (Guo and Wang, 2016; Hu et al., 2019). Based on the existing observation data and improved soil physics, the estimated changes in previous studies are generally larger than that of actual change (Lawrence et al., 2012; Cheng et al., 2019; Wang et al., 2019b).
4. Discussion
In order to project the possible future changes of permafrost, we simulated MAGT and ALT changes under the present state and future scenarios based on statistical and ML methods. The results show that under different RCPs, significant degradation of the QTP permafrost may occur (e.g., MAGT rising and ALT thickening); in particular, under RCP8.5, more than half of the near-surface permafrost will disappear, and regional differences were observed. In this section, to further verify the feasibility of our results, we compared our simulated MAGT and ALT with those of previous studies and then analyzed the vulnerability of permafrost to climate change under the present state. Based on these findings, we proposed urgent action should be taken to adapt climate change. Finally, the model performance and potential sources of the uncertainty in this study were discussed.