Land surface temperature (LST) is a critical state variable of land surface energy equilibrium and a key indicator of environmental change such as climate change, urban heat island, and freezing-thawing hazard. The high spatial and temporal resolution datasets are urgently needed for a variety of environmental change studies, especially in remote areas with few LST observation stations. MODIS and Landsat satellites have complementary characteristics in terms of spatial and temporal resolution for LST retrieval. To make full use of their respective advantages, this paper developed a pixel-based multi-spatial resolution adaptive fusion modeling framework (called pMSRAFM). As an instance of this framework, the data fusion model for joint retrieval of LST from Landsat-8 and MODIS data was implemented to generate the synthetic LST with Landsat-like spatial resolution and MODIS temporal information. The performance of pMSRAFM was tested and validated in the Heihe River Basin located in China. The results of six experiments showed that the fused LST was high similarity to the direct Landsat-derived LST with structural similarity index (SSIM) of 0.83 and the index of agreement (d) of 0.84. The range of SSIM was 0.65–0.88, the root mean square error (RMSE) yielded a range of 1.6–3.4 °C, and the averaged bias was 0.6 °C. Furthermore, the temporal information of MODIS LST was retained and optimized in the synthetic LST. The RMSE ranged from 0.7 °C to 1.5 °C with an average value of 1.1 °C. When compared with in situ LST observations, the mean absolute error and bias were reduced after fusion with the mean absolute bias of 1.3 °C. The validation results that fused LST possesses the spatial pattern of Landsat-derived LSTs and inherits most of the temporal properties of MODIS LSTs at the same time, so it can provide more accurate and credible information. Consequently, pMSRAFM can be served as a promising and practical fusion framework to prepare a high-quality LST spatiotemporal dataset for various applications in environment studies.

Soil salinization threatens agricultural sustainability in arid zones, yet quantitative attribution of its spatiotemporal dynamics to multi-source drivers remains scarce at regional scales. To address this, we developed an explainable framework merging Sentinel-1/2, ERA5-Land, and topographic-hydrological indices with XGBoost, trained under weak supervision with proxy labels and independently validated using field-measured ECe. A 7-group, 44-feature ensemble with spatial block 5-fold cross-validation ensured robust assessment. SHapley Additive exPlanations (SHAP) quantified driver contributions and enabled a novel dominant driver zoning (DDZ) framework. Monitoring the Hexi Corridor and Tarim Basin (2017–2024) revealed contrasting trajectories: Hexi’s dynamics were primarily climate-driven (Aridity Index), whereas 19.2% of Tarim showed significant salinization along oasis–desert margins co-dominated by elevation, soil indices, and temperature. The model achieved spatial cross-validation R² values around 0.65. DDZ mapping showed climate dominance in 98.2% of Hexi compared to 76.5% in Tarim, where terrain and optical factors were more influential. The weak supervision strategy overcomes scarce in-situ measurements, while the DDZ maps identified that Land-use-dominated zones recorded the highest salinity, offering clear directives for targeted salinity control in arid basins.

In summary, near-surface icing risk in complex alpine terrain is jointly controlled by freezing conditions, moisture supply, freeze–thaw transitions, and topographic energy processes. Traditional approaches relying on sparse station data or single temperature thresholds fail to capture spatial heterogeneity, and frequent cloud cover together with topographic errors severely limit the application of thermal infrared remote sensing. Taking the area along the Duku Highway in the Tianshan Mountains as the study region, a daily icing risk assessment framework at 250 m resolution was constructed using multi-source remote sensing, ERA5-Land reanalysis data, topographic correction, and an energy–moisture dual-constrained model. A diurnal temperature cycle model, the CAP index, and physics-constrained machine learning were integrated to reconstruct the daily minimum land surface temperature ($T_{s,\mathrm{m}\mathrm{i}\mathrm{n}}$) at 250 m resolution under all weather conditions. A probabilistic two-tier risk assessment model was then established by incorporating moisture, topography, and freeze–thaw transitions. The results show that high-risk zones occur primarily in valleys and topographically constrained corridors rather than the coldest elevations. Validation against Landsat LST (r = 0.886) and the Bayanbulak station (bias −0.76 °C, RMSE 5.62 °C, r = 0.91) confirms spatial and seasonal accuracy. Sensitivity and Monte Carlo analyses indicate the RiskScore is mainly controlled by the low-temperature weight, while upstream parameters are less influential. The framework is best applied as a screening and early-warning product to identify sub-kilometer potential icing corridors, complementing point measurements and short-range forecasts.