Divergent responses of permafrost degradation to precipitation increases at different seasons on the eastern Qinghai–Tibet Plateau based on modeling approach

The Qinghai–Tibet Plateau (QTP) has responded to remarkable climate warming with dramatic permafrost degradation over the past few decades. Previous studies have mostly focused on permafrost responses to rising air temperature, while the effects of accompanying increases in precipitation remain contentious and largely unknown. In this study, a distributed process-based model was applied to quantify the impacts of increased precipitation on permafrost thermal regimes in a warming climate by employing model experiments in the source region of Yellow River (SRYR) on the eastern QTP. The results showed that the active layer thickness (ALT) of permafrost increased by 0.25 m during 2010–2019 compared to 2000 across the SRYR, which was primarily driven by climate warming. In contrast, the increased annual precipitation played a relatively limited role and just slightly mitigated active layer thickening by 0.03 m. Intriguingly, increased precipitation in the cold and warm seasons exerted opposite effects on permafrost across the SRYR. The increased precipitation in the cold season mainly promoted ALT increases, while the increased precipitation in the warm season mitigated ALT increases. In ∼81.0% of the permafrost across the SRYR, the cooling effects of warm season wetting outweighed the warming effects of cold season wetting; while at the transition zone where permafrost was unstable and degrading to seasonally frozen ground, the warming effects of cold season wetting played a relatively larger role which contributed to permafrost degradation. This study explored the physical mechanisms of permafrost thermal responses to climate wetting, thus providing a better understanding of permafrost change in a warmer and wetter climate on the QTP.


Introduction
Permafrost, as one of the most critical components of the terrestrial system, plays an important role in stabilizing the temperature at the interface of the lithosphere, atmosphere and hydrosphere (Dobinski 2011) and provides essential feedbacks to climate change Wu 2007, Koven et al 2011). The vulnerability of permafrost in a warming climate has raised worldwide concern due to its potential impacts on the vegetation dynamics (Jin et al 2021, Wang et al 2022), water resources , carbon cycle (Schuur et al 2015, Wang et al 2020a and even on atmospheric circulation . The active layer is defined as the ground layer subject to annual freeze and thaw in the permafrost region (Dobinski 2020), which regulates the water exchange and energy transfer between the atmosphere and soils (Wu and Zhang 2010, Li et al 2012, Zhang et al 2021b. The increase in active layer thickness (ALT) is one of the most prominent phenomena that indicates the degree of permafrost degradation (Sharkhuu et al 2007, Blok et al 2010, Dobinski 2011, Guo and Wang 2013, Frey et al 2016, Vasiliev et al 2020, Zhao et al 2020. The Qinghai-Tibet Plateau (QTP), known as the 'Third Pole' , is covered by the largest area of elevational permafrost across the world (Yao et al 2017). As one of the most sensitive regions to climate change, the QTP has experienced prominent climate warming in the last five decades, with a warming rate twice as fast as the global average (Chen et al 2015, Yao et al 2019. Due to the drastic climate warming, extensive degradation of permafrost has been observed in the QTP, including the active layer thickening (Luo et al 2016), rising in the lower altitudinal limit and even permafrost disappearance (Cheng and Wu 2007, Kang et al 2010, Wu et al 2013. Meanwhile, previous research suggests that the QTP has become wetter since the 1980s despite large regional differences. For instance, the annual precipitation has increased in most parts of the QTP since the mid-1990s, with the most intensive increase observed in the northern and central QTP (Lei et al 2014. While in some parts of the southeastern and southern QTP, the precipitation has declined since 1998 , Zhang et al 2017, 2021a. The increasing trend of precipitation on the QTP is also expected to persist into the future. According to the projections of Coupled Model Intercomparison Project phase 6 models, the QTP is expected to experience continued warming and increased precipitation in the coming decades of the 21st century (Lalande et al 2021). Precipitation is projected to increase at a faster rate than the global mean but slower than that in the Arctic. Furthermore, larger precipitation increases are expected under higher emission scenarios (Chen et al 2022). Consequently, it raises questions about how increased precipitation could jointly affect permafrost thermal dynamics along with atmospheric warming.
The rising air temperature is thought to be the most important driver of the permafrost degradation (Karjalainen et al 2019, Zheng et al 2020, but the increased precipitation could also modulate the soil thermal regime by affecting the soil moisture and surface energy transfer through competing mechanisms. A wetter soil in summer with an increased thermal conductivity allows heat to penetrate deeper, and the rainfall can intensify the convective heat transfer in the soil layers and promotes thaw (Luo et al 2020, Mekonnen et al 2021, Murton 2021). However, more saturated soil also enhances evaporative cooling and requires more latent heat for phase change, thus protecting permafrost from further thawing (Park et al 2013, Li et al 2019. The snow cover as a buffer layer between atmosphere and soil also have mixed impacts on soil temperature, including insulation effect (warming by insulation from the cold air), and albedo effect (cooling by increased albedo and decreased incoming radiation) , Wang et al 2019, 2020b, Yi et al 2019, Gao et al 2021. It remains unclear which mechanisms play a dominant role. Although some studies have investigated the impacts of increased precipitation on permafrost, conclusions were inconsistent in the magnitude and direction. Some studies suggested both increased precipitation and rising air temperatures are thought to contribute equally to permafrost degradation (Mekonnen et al 2021), while some research indicated the wetting could almost offset the impacts of rising air temperature and prevent permafrost from degradation (Zhang et al 2021a). Since the numerical modeling is the most common method in previous studies, large uncertainty may arise from the differences of model structures when depicting the soil heat and water transfer processes. Besides, the seasonally varied modulating mechanisms could also lead to divergent permafrost response to air temperature and precipitation changes in the warm and cold season , Zheng et al 2020, which is also reported by observational studies (Cai et al 2018). Therefore, the distinct impact of warming and wetting (i.e., precipitation increases) at different seasons on the permafrost changes in a warmer and wetter future QTP needs to be further examined.
From the views above, this study uses a state-ofthe-art process-based model (geomorphology-based eco-hydrological model, GBEHM) to quantify the impacts of precipitation increases on the permafrost active layer at different seasons, based on numerical modeling in the source region of Yellow River (SRYR) on the eastern QTP. The objectives are: (a) to quantify the overall impacts of precipitation increases on the permafrost thermal conditions under a warming climate and (b) to unravel the dominant influencing mechanisms of precipitation changes on the permafrost thermal conditions in the cold and warm seasons, respectively.

Study area and data
A typical permafrost-affected region, the SRYR (95 • 50 ′ E to 103 • 30 ′ E, 32 • N to 35 • 40 ′ N) on the eastern QTP, was selected for its location at the boundary between permafrost and seasonally frozen ground with a permafrost area proportion of 31.5% during 2010-2019 (2010s) (Yang et al 2023) (figure 1). The SRYR has a catchment area of 1.24 × 10 5 km 2 (figure 1), and snow, lakes, wetlands, and frozen ground characterize the landscape. The SRYR has undergone extensive permafrost degradation due to climate warming over the past several decades (Wang et al 2018a. The mean annual precipitation is 549 mm and is concentrated from June to September when 75%-90% of the precipitation falls during this period due to the influence of monsoon (Zheng et al 2009). From October to April, the air temperature remains below 0 • C, ranging from approximately 2 • C to 8 • C from May to September. The highest temperatures are usually observed in July. According to the monthly air temperature, this study divides a year into the cold season (October to April) and the warm season (May to September).
The precipitation data is from the China Gaugebased Daily Precipitation Analysis (CGDPA), a gridded precipitation dataset with a resolution of 0.25 • × 0.25 • . The CGDPA dataset has been widely used in the QTP and has achieved satisfactory performance (Qin et al 2017, Shi et al 2020. Air temperature is generated by merging TerraClimate (Abatzoglou et al 2018), a monthly dataset of air temperature with a high spatial resolution (1/24 • , ∼4 km), and observations from 22 meteorological stations (figure 1), and the details can be found in Yang et al (2023). Other climate forcing data including wind speed, relative humidity and sunshine duration are interpolated from the meteorological station observations (Yang et al 2004). The key geographic datasets used in the process-based model include the digital elevation data, land use map and soil water parameters. The digital elevation data is from the shuttle radar topography mission (SRTM) (Jarvis et al 2008), while the land use map is from the Land use dataset in China (1980China ( -2015 (Chinese Academy of Sciences and Environmental Science Data 2020), and the soil water parameters are from the Dai et al (2013). This study used a global evapotranspiration product of Global Land Evaporation Amsterdam Model (GLEAM) with a resolution of 0.25 • for comparison. The GLEAM evapotranspiration product is produced by applying the Priestley and Taylor equation to calculate the potential evaporation using a variety of satellite products as input, and it is validated using the station observations at FLUXNET sites (Miralles et al 2011). A surface soil moisture dataset over the Maqu region (101 • 38 ′ E to 102 • 45 ′ E, 33 • 30 ′ N to 34 • 14 ′ N) is a spatially upscaled soil moisture record based on in situ measurements of Maqu monitoring networks (Zhang et al 2021c), which is used for GBEHM soil moisture validation.

Brief introduction of GBEHM
The GBEHM is used to simulate the soil water and thermal dynamics. GBEHM originated from the geomorphology-based model (GBHM) (Yang et al 2002(Yang et al , 2015, and was subsequently improved to include ecohydrological processes (Yang et al 2015) and cryospheric processes (Qin et al 2017, Gao et al 2018. The study basin was discretized into 240 subcatchments for this study using a 3 km grid system, which is the same as the model settings in Yang et al (2023). At each grid, a soil column configuration of 31 layers with a total depth of 50 m below the ground surface was adopted (figure S1). The thickness of the soil layer increased with depth, ranging from 0.05 m for the top soil layers to >10 m for the bottom soil layers (table S1). The soil freezing and thawing processes are simulated by solving the coupled heat and water transfer equation. For each soil layer, the energy balance is solved using the equation from Flerchinger and Saxton (1989): Note: The subscript 'detrend' or 'origin' denotes whether the precipitation is processed by the detrending method or not processed.
where C S is the volumetric heat capacity of the soil (J m −3 K −1 ); T is the soil temperature (K); t is the time step of calculation (s); ρ i and ρ l are the densities of ice and liquid water (kg m −3 ); L f is the latent heat of fusion (J kg −1 ); θ i is the soil ice content (m 3 m −3 ); z is the soil layer depth (m); k s is the soil thermal conductivity (W m −1 K −1 ) calculated using the scheme from Farouki (1981); c l is the specific heat capacity of water (J kg −1 K −1 ); and q l is the liquid water flux between soil layers, which is calculated by the Richards equation. The upper boundary condition for heat transfer is determined by surface energy balance, using the same methods of the Simple Biosphere Model 2 (Sellers et al 1996). Specifically, the snow cover is parameterized by a multilayer snow cover model provided by Jordan (1991). As for the lower boundary for heat transfer, a zeroheat flux is assumed at the depth of 50 m. By incorporating different processes, GBEHM considers both conductive and the non-conductive heat transfer processes, including conductive heat transfer, latent heat exchange and convective heat transfer from snowmelt water and rainwater, and more details can be found in Yang et al (2015), Qin et al (2017) and Gao et al (2018). The permafrost distribution is determined by the soil temperature simulated by GBEHM, and permafrost is considered to exist at a 3 km grid if there is a soil layer where the simulated temperature keeps at or lower than 0 • C throughout the study period (2000-2019) (figure 1). The GBEHM has been successfully applied in simulations of frozen ground dynamics, hydrological processes and ecosystem changes in many regions across the QTP (Qin et al 2017, Gao et al 2018, Shi et al 2020. In the SRYR, the GBEHM has been comprehensively validated in the hydrological and frozen soil modeling using multi-source data, and has obtained satisfactory results . In this study, the simulated snow depth was compared with the observations from seven meteorological stations in the SRYR. Figure S2 shows that the simulated monthly snow depth can effectively reflect the monthly variations of the observations (R 2 = 0.79, Mean bias = 0.07 cm). Regarding frozen soil modeling, GBEHM demonstrated good performance in reproducing the depth of frozen ground at 11 observational sites, and the type of frozen ground at 53 boreholes ( (b)). This study is a follow-up study of Yang et al (2023) to further explore the distinct impacts of warming and wetting on permafrost in this region.

Experimental design
The annual precipitation in SRYR experienced a dramatic decrease in the 1990s and then rebounded in the 2000s and 2010s (Meng et al 2016, Qin et al 2017, Wang et al 2018b. The air temperature rose at a rate of 0.04 • C/a (p < 0.05) and the annual precipitation increased by 7.46 mm a −1 (p < 0.05) during 2000-2019. Thus, we selected 2000-2019 as the study period since both significant climate warming and wetting happened in the past 20 years. Four simulation experiments (table 1), consisting of one historical scenario (Hist) and three hypothetical scenarios (Exp1-3), were conducted to separate the effects of increased precipitation during different seasons on permafrost thermal state. In Exp3, precipitation in all months was detrended. Precipitation was detrended in warm season (from May to September) in Exp1 and was detrended in cold season (from October to April) in Exp2. Therefore, Exp3 serves as the benchmark assuming that precipitation showed no trend all year around. The differences between Exp3 and other scenarios denote the effects of wetting on ALT at different seasons, including the effects of coldseason wetting (Exp1-Exp3), warm-season wetting (Exp2-Exp3) and year-round wetting (Hist-Exp3) (table 1). The linear detrending method was applied for monthly precipitation data by MATLAB R2019a (Wu and Chen 2019). The original precipitation and its detrended series at different seasons are shown in figure 2. During 2000-2019, the precipitation mainly fell in the warm season with an average value of 466 mm, account for 82.6% of the annual precipitation. The cold season precipitation just accounted for 17.4% of the annual precipitation but increased fast at a rate of 2.38 mm a −1 , contributing about 31.8% of the annual precipitation increase rate (7.47 mm a −1 ). Figure 3(a) depicts the simulated changes in ALT in the permafrost region under four distinct scenarios. During the 2010s, the permafrost ALT has increased by 0.25 m compared to the value in 2000 (Hist), indicating a dramatic permafrost degradation in the past 20 years over SRYR. The Exp3 removed the wetting trend but conserved the warming trend, and the ALT has increased by 0.28 m in this scenario, indicating the dominant role of rising air temperature in permafrost degradation. Figure 3(a) shows notable variations in the permafrost responses to different seasonal wetting conditions, particularly during the 2010s. It is observed that during the 2010s, Exp1 exhibits the highest ALT, and the average ALT difference between Exp1 and Exp3 is 0.05 m. This implies that a rise in precipitation in the cold season has intensified permafrost degradation. In the 2010s, the ALT in the Exp2 is the smallest among all scenarios, and the average ALT difference between Exp2 and Exp3 is −0.10 m ( figure 3(a)), indicating that permafrost degradation is less severe when increased precipitation only occurs during the warm season. Influenced by both coldseason wetting and warm-season wetting, the ALT in the Hist scenario varies in a range between the Exp1 and Exp2 (figure 3(b)). Overall, the ALT in Hist shows negative differences relative to Exp3, and the mean ALT difference between Hist and Exp3 is −0.03 m during the 2010s. The above results suggest the time when wetting happens (i.e., in the warm or cold seasons) could determine whether the wetting protects permafrost or promotes permafrost thawing. Figure 4 illustrates the spatial distribution of the ALT difference between Exp3 and three wetting scenarios during the 2010s. In the Exp1 scenario, 58.8% of the permafrost area shows a positive difference in ALT relative to the Exp3, while 41.2% exhibits a smaller ALT. However, in the Exp2 scenario, ALT shows negative differences across almost the entire permafrost region. And in the Hist scenario, negative differences in ALT are observed in the majority of areas (81.0%), but positive differences are found at the permafrost region margins (19.0%). Figures 3 and 4 reveal distinct ALT changes under three wetting scenarios. In order to investigate the mechanisms underlying the influence of rising precipitation on permafrost, we present simulated results for snow depth, evapotranspiration and soil temperature under four scenarios and explore the mechanisms by seasons as follows.

Insulation effect of snow cover in the cold season
Precipitation in the cold season is mainly in the form of snowfall. Figure S6 displays the basin-averaged annual snow depth in the Exp1 and Exp3 over the SRYR. The data indicates that the snow depth in the Exp1 is significantly larger than that in the Exp3, particularly during the 2010s, when precipitation increased markedly. Figures 5(a) and (b) show the spatial distributions of the snow depth during the 2010s in the Exp3 and Exp1 over the permafrost region. The simulated soil temperatures of the surface soil at the depth of 0-0.05 m are also shown in figures 5(d) and (e). Figures 5(c) and (f) reveal a spatial distribution of soil temperature difference similar to that of snow depth difference, which implies that the overlying snow cover is likely to influence soil temperature during the cold season. Differences in the surface soil temperature (∆Tsoil) are positively correlated with differences in the snow depth (∆Snow depth, figure 6). The positive correlation between ∆Tsoil and ∆Snow depth is more significant in the whole SRYR ( figure 6(a)), where the  correlation coefficient (r) is 0.73 (p < 0.05), than in the permafrost region (r = 0.47, p < 0.05). The spatial pattern is also similar for the difference of soil temperature at 1 m depth, indicating the transmission of the insulation effect to deeper layers ( figure  S7). Despite the overall warming effects of cold season wetting, cooling effects also exist in some regions with little snow cover (figure 5), where the increased cold season precipitation might enhance evapotranspiration cooling from snowmelt that outweighs the relatively weak snow insulation effects.

Evaporative cooling in the warm season
In the warm season, it is difficult to form longterm snow cover due to the relatively high air temperature, and little differences are shown in the snow depth between the Exp2 and Exp3 scenarios. Figures 7(a) and (b) shows the mean evapotranspiration in the Exp2 and Exp3 scenarios in the warm season during the 2010s, and figure 7(c) shows the evapotranspiration difference (∆ET). It can be seen that the evapotranspiration is higher in the Exp2 than in the Exp3. Figures 7(d) and (e) displays the surface soil (0-0.05 m) temperature in the two scenarios. Figure 7(F) is the difference in surface soil temperature (∆Tsoil) by subtracting the Exp3 results from the Exp2 results. The ∆Tsoil is generally negative (figure 7(f)) with a similar spatial distribution to that of ∆ET (figure 7(c)). Figure 8 presents the correlation between the ∆Tsoil and ∆ET. From figure 8, the ∆Tsoil has a significant negative correlation with ∆ET. The permafrost region exhibits a highly negative correlation between ∆Tsoil and ∆ET, as evidenced by a strong correlation coefficient (r) of −0.85 (p < 0.05). In the whole SRYR, the negative correlation is relatively weaker (r = −0.55, p < 0.05). The spatial distribution of soil temperature difference at 1 m depth is similar to that at 0-0.05 m depth ( figure  S8), but the magnitudes are smaller, which indicates that the evaporative cooling effect is stronger at  the near surface soil layers. The results above indicate that the increased precipitation during the warm season would lower soil temperature and protect permafrost.

Permafrost response to year-round wetting and the dominant mechanism
From 3.2.1 and 3.2.2, wetting in cold season causes soil warming through snow insulation, while wetting in warm season results in soil cooling due to stronger evaporation. In the Hist scenario, both of these mechanisms exist and counteract each other. As shown in figure 3, the ALT in the Hist is smaller than in Exp3, representing an overall cooling effect with increased annual precipitation. The cooling effect due to evaporation on soil during the warm season is more pronounced than the warming effect caused by snow insulation. Based on the results presented in figure 4(c) and table 2, it can be observed that 81.0% (29 961 km 2 ) of the permafrost area displays a negative ALT difference in the Hist scenario compared to Exp3, primarily driven by the cooling effect. On the other hand, 19.0% (7038 km 2 ) of the permafrost area exhibits a positive ALT difference in the Hist compared to Exp3, mainly caused by the warming effect. Overall, these findings suggest that the increase in precipitation all year round plays a protective role for permafrost over the study area. The regions where ALT exhibits positive differences are located in the transition zone where permafrost is unstable and degrading to seasonally frozen ground with an elevation ranging from 4300 to 4800 m. In the

Discussion
Some prior work has investigated the effects of precipitation increases on permafrost, but the conclusions were largely inconsistent in the direction and magnitude. Some studies reported that climate wetting can dramatically reduce permafrost thermal response to atmospheric warming (Zhang et al 2021a(Zhang et al , 2021b, while other studies concluded that increased precipitation in a warmer climate would promote permafrost degradation (Douglas et al 2020, Mekonnen et al 2021, Magnusson et al 2022. The findings of this study indicate that while precipitation changes can indeed influence the thermal state of permafrost, the magnitude of impact is relatively minor when compared with the impact of atmospheric warming (figure 3). On the one hand, the discrepancies in previous research could be due to the lack of consideration of the differential impact of precipitation at different seasons. If the effects of increased precipitation are only considered in either the cold or warm season, a considerable effect could arise. However, since the effects of increasing precipitation in different seasons offset each other, the overall impact of the precipitation increase is not substantial. On the other hand, some studies argued that the rainfall in the warm summer penetrating into soil could accelerate active layer deepening through enhanced convective heat transfer (Mekonnen et al 2021, Magnusson et al 2022, Zhao et al 2022. In this study, the increased convective heat transfer due to rainwater penetration is limited compared with the decreased conductive heat and increased latent heat, which is also consistent with the findings of Zhang et al (2021b). This study instead finds a cooling effect of summertime rainfall events on soil, which can be illustrated by half-hourly observations from a grassland site (GS) (97 • 33 ′ 16 ′′ , 34 • 54 ′ 51 ′′ ) in the SRYR (figure 1) . Figure S9 shows the observed rainfall events and the responses of soil water content, soil temperature and ground heat flux at the GS site from 23 June to 23 July, 2016. After heavy rainfall events and consecutive light rainfall events, there are rises in the soil water content, declines in the soil temperatures and decreases in ground heat flux ( figure S9(b)). Consistent change directions are shown between the soil temperatures and surface heat fluxes (figures S9(b) and (c)). The summer rainfall events could lower soil temperature because of decreased conductive heat with lower air temperature on rainy days (figure S9(b)) and increased latent heat flux due to enhanced evapotranspiration with higher soil water content (figure S9(a)), leading to decreased ground heat flux (figure S9(c)) and mitigated permafrost thaw. The different responses to increased summertime precipitation in the QTP and the Arctic region could be attributed to the different climate conditions. On the one hand, the QTP has relatively low snowpack, which leads to stronger coupling between air temperature and soil temperature thus causing more rapid soil cooling when air temperature drops in rainy days ( figure S9). On the other hand, the evaporation processes could respond more quickly to the increased rainfall due to drier soils on the QTP, which could in turn result in more intense evaporative cooling effect. By applying a physically-based model, this study identified the physical mechanisms causing opposing effects at different seasons. When the cold season precipitation increases, due to the stronger insulation effect of the increased snow cover thickness, less energy can be transferred from ground to air in winter. By comparing the results of Exp3 and Exp1 scenarios, the ground heat flux difference (∆G) has a significant positive correlation with the snow depth difference (∆Snow depth) (r = 0.74, p < 0.05) (figure S10). The increased snow cover thickness is thought to be a major reason for permafrost degradation in the circum-Arctic region (Park et al 2014(Park et al , 2015. The overall impact of snow cover on permafrost thermal regime depends on the thickness, density, timing and duration (Zhang 2005, Wang et al 2019. In contrast, the warm season wetting could increase the soil moisture content. A significant positive correlation between the evapotranspiration difference (∆ET) and the liquid soil moisture (∆soiwliq) (r = 0.56, p < 0.05) (figure S11) indicates strong water limitation in the SRYR (Liu et al 2017). Other studies have also reported the evaporative cooling effect of summer rainfall (Hinkel et al 2001, Lawrence andSwenson 2011). Overall, the cooling effects of warm season wetting outweighed the warming effects of cold season wetting in the SRYR (figure 3). However, in the transition zone where permafrost is unstable and degrading to seasonally frozen ground, the response of permafrost is primarily driven by the warming effect induced by cold season wetting (figure 4(c)). This is because the evaporative cooling effect is strong near the ground surface but becomes limited in the deeper thermal field (Hinkel et al 2001), and in these regions the ALT has already become quite deep (table 2), thus the cooling effects can hardly affect the deep soil layers near the ALT. It should be noted that although the physically-based model adopted in this study has already considered many interactions between cryospheric and eco-hydrological processes, some limitations still exist. For example, climate warming, wetting and concomitant permafrost degradation could affect evapotranspiration and surface energy balance by altering vegetation dynamics and nutrient availability, which is not discussed in this study and can be further explored by future studies.

Conclusion
This study investigated the distinct permafrost thermal responses to increased precipitation at different seasons in the SRYR on the eastern QTP through numerical experiments based on a state-of-the-art physically-based permafrost-hydrological model. In summary, the major conclusions of this study are: (1) The SRYR has experienced significant climate warming and wetting in the past 20 years, accompanied by dramatic permafrost degradation with an increase of 0.25 m in ALT during the 2010s compared with 2000. The ALT increase is primarily driven by atmospheric warming, while wetting just reduced the ALT by 0.03 m, highlighting the dominant role of warming in permafrost degradation.
(2) The simulation results revealed the effects of increased precipitation on permafrost differ in the warm and cold seasons. In the 2010s, the cold-season wetting leads to an increase of 0.05 m in ALT, which promotes permafrost thawing. However, the warm-season wetting reduced the ALT by 0.10 m in the 2010s and protected the permafrost from degradation. (3) The increased precipitation in the cold season exerted a warming effect on soil mainly through the insulation effect of snow cover, while increased precipitation in the warm season had a cooling effect on soil primarily driven by the evaporative cooling effect. These mechanisms counteract each other, leading to an overall milder permafrost response to climate wetting. (4) During 2000-2019, the increases in annual precipitation exerted an overall cooling effect on the SRYR. 81.0% of the permafrost ALT change is dominated by the cooling effect of warm-season wetting, while the remaining area is primarily driven by the warming effect of cold-season wetting.

Data availability statements
The data cannot be made publicly available upon publication because they are not available in a format that is sufficiently accessible or reusable by other researchers. The data that support the findings of this study are available upon reasonable request from the authors.