The hydrothermal relationship between vegetation and seasonal frozen soil is one of the key research contents in the fields of permafrost ecological environment, hydrology and climate change in alpine mountainous areas. Based on the monitoring data of air temperature, precipitation and soil hydrothermal conditions at the depth of 0–5 m from site TS-04 (with high vegetation coverage) and site TS-05 (with low vegetation coverage) in the alpine grassland of the Tianshan Mountains, this study compared and analyzed the characteristics of freezing-thawing process, temperature and moisture changes of seasonal frozen soil with different vegetation coverage. The results show that the maximum seasonal freezing depth of the two sites is almost comparable, but site TS-04 has a smaller freezing and thawing rate, and a shorter duration of freeze-thaw at all depths. TS-04 also has a smaller annual range of surface temperature and ground-air temperature difference. The analysis indicates that vegetation acts as a thermal buffer and has a good thermal insulation effect on the ground surface. Site TS-04 had high unfrozen water content in the unfrozen period and the water content increased with depth, while the unfrozen water content was low in site TS-05. In addition, the thresholds of soil water content response to rainfall events at 5 cm depth of site TS-04 and site TS-05 were 5 and 11 mm precipitation respectively, which indicated that the high vegetation coverage is conducive to rainwater infiltration, and the underlying soil of the site has a faster response to rainfall events.

The land surface temperature obtained from remote sensing was widely used in the simulation of permafrost mapping instead of air temperature with the rapid development of remote sensing technology. The land surface freezing and thawing index (LFI and LTI), which is commonly regarded as the ground surface freezing and thawing index (GFI and GTI), can produce certain errors in the simulation of permafrost distribution on the Qinghai–Tibet Plateau. This paper improved the accuracy of the thermal condition of the surface soil in the Qinghai–Tibet Engineering Corridor (QTEC) by calculating the LFI (or LTI) and N-factors. The environmental factors affecting the spatial distribution of the GFI and GTI were detected by the GeoDetector model. Finally, the multiple linear relationships between the GFI (or GTI) and the environmental factors were established. The results from 25 monitoring sites in the QTEC show that the N_f (ratio of GFI to LFI) is 1.088, and the N_t (ratio of GTI to LTI) is 0.554. The explanatory power of the interaction between elevation and latitude for the GFI and GTI is 79.3% and 85.6%, respectively. The multiple linear regression model with six explanatory variables established by GFI (or GTI) has good accuracy. This study can provide relatively accurate upper boundary conditions for the simulation of permafrost distribution in the QTEC region.

Aboveground biomass (AGB) serves as a crucial measure of ecosystem productivity and carbon storage in alpine grasslands, playing a pivotal role in understanding the dynamics of the carbon cycle and the impacts of climate change on the Qinghai–Xizang Plateau. This study utilized Google Earth Engine to amalgamate Landsat 8 and Sentinel-2 satellite imagery and applied the Random Forest algorithm to estimate the spatial distribution of AGB in the alpine grasslands of the Beiliu River Basin in the Qinghai–Xizang Plateau permafrost zone during the 2022 growing season. Additionally, the geodetector technique was employed to identify the primary drivers of AGB distribution. The results indicated that the random forest model, which incorporated the normalized vegetation index (NDVI), the enhanced vegetation index (EVI), the soil-adjusted vegetation index (SAVI), and the normalized burn ratio index (NBR2), demonstrated robust performance in regards to AGB estimation, achieving an average coefficient of determination (R²) of 0.76 and a root mean square error (RMSE) of 70 g/m². The average AGB for alpine meadows was determined to be 285 g/m², while for alpine steppes, it was 204 g/m², both surpassing the regional averages in the Qinghai–Xizang Plateau. The spatial pattern of AGB was primarily driven by grassland type and soil moisture, with q-values of 0.63 and 0.52, and the active layer thickness (ALT) also played a important role in AGB change, with a q-value of 0.38, demonstrating that the influences of ALT should not be neglected in regards to grassland change.

Accurate mapping of thermokarst lakes in the Qinghai–Tibet Engineering Corridor is important for understanding permafrost degradation and assessing environmental risks to major infrastructure. However, thermokarst lake extraction from medium-resolution satellite imagery remains challenging because these lakes are often small, fragmented, seasonally variable, and spectrally confused with wetlands, shadows, and other surface water bodies. In this study, Sentinel-2 imagery from the 2020 thaw season was used to map thermokarst lakes in the Qinghai–Tibet Engineering Corridor. A 16-feature dataset was constructed by integrating spectral bands, water indices, texture features, and topographic variables, and a convolutional neural network (CNN) was compared with five conventional machine learning classifiers. In the pixel-based validation, the CNN slightly outperformed the other evaluated models, achieving the overall accuracy of 98.04% and an F1-score of 97.18%. Independent polygon-based validation using the Jilin-1 visual interpretation reference showed that the final CNN-derived inventory achieved an IoU of 0.79, with omission and commission ratios of 0.15 and 0.09, respectively. The CNN more effectively suppressed salt-and-pepper noise, reduced fragmented and serrated lake boundaries, and improved the spatial continuity of mapped water bodies compared with traditional machine learning classifiers. SHAP-based attribution suggested that CNN predictions were more strongly associated with water indices and texture features, whereas Random Forest predictions were mainly associated with near-infrared and shortwave-infrared bands. Thermokarst lakes showed higher lake area proportions and densities in areas with relatively high ground ice content, gentle slopes, thicker active layers, warmer permafrost, and unstable permafrost conditions. These results demonstrate the potential of Sentinel-2 imagery and convolutional models for regional thermokarst lake mapping and permafrost degradation monitoring.