The seasonal and regional variability of Arctic sea ice area (SIA) and thickness (SIT) were investigated between 1979 and 2020 for the Atlantic sector (AS), Pacific sector (PS) and Barents–Kara Seas (BKSs). We applied the SIA data from remote sensing observations and SIT data from numerical model calculations. We found the large summer variability of SIA and SIT in AS and PS compared with those in winter. The opposite feature was seen in the BKSs. The annual declining rates of SIA and SIT were the largest in PS (−1.73 × 10⁴ km² yr⁻¹) and AS (−3.36 × 10⁻² m yr⁻¹), respectively. The SIA variability was modest for winter PS and the northern Canadian Arctic Archipelago of AS. The annual and winter SIA flux from PS to AS gradually increased in 1979–2020; the summer SIA flux accounted for 11% of the PS summer SIA decline. The annual and seasonal SIA outflow through the Fram Strait during 1979–2020 steadily increased while for annual and winter SIA export, the increase mainly occurred in 1979–2000; the summer SIA outflow was only 1.45% equivalent to the decrease in the entire Arctic summer SIA. We concluded that sea ice export was not a major impact factor on the seasonal and regional decline of SIA and SIT except for the individual years. The near surface air temperature (SAT) and sea surface temperature (SST) were responsible for the retreat and thinning of the sea ice. The dramatic increase in SAT in winter resulted in a strong decrease in winter sea ice in BKS. The outgoing longwave radiation had significant negative correlations with SIA and SIT and positive correlations with SAT and SST. The Atlantic multi-decadal oscillation, related to the North Atlantic Ocean’s SST anomalies, had significant negative correlations with SIA and SIT. The SIT had higher correlations with the atmospheric and oceanic factors compared with SIA, which indicates that SIT is important for predictions of Arctic sea ice and climate change.

Glacier changes in the Karakoram Mountain (KM) region were analyzed using Landsat images from 2000 to 2021. Clean ice boundaries were mapped using supervised classification, while debris-covered ice was identified by integrating surface temperature, temperature vegetation dryness index (TVDI), and slope (ASTER DEM data). Extraction accuracy was 98% for clean ice and 67%–71% for debris-covered ice. In 2021, debris-covered ice represented about 5% of the total glacier area. Glacier area declined from 26.59 × 10³ km² (2000) to 20.66 × 10³ km² (2021), except for Amu Darya Basin (+1%). Tarim Basin and Indus Basin retreated by 14% and 25%, respectively. Rapid glacier loss in the Indus Basin may sharply decrease future river flow, underlining the need for accurate debris-covered glacier mapping for water resource assessment.

Soil moisture (SM) is critical for climate, water, and agriculture, but Soil Moisture Active Passive (SMAP) passive microwave products have coarse resolution, limiting regional applications. This study develops an SM downscaling framework based on Transformer and its variants (PatchTST and iTransformer), integrating multi-source satellite and groundwater data to generate 1 km daily SM products (2015–2022). Compared with Random Forest (RF), Long Short-Term Memory (LSTM), and Convolutional Neural Network–LSTM (CNN-LSTM), Transformer and its variants achieve superior accuracy and generalization. Validated against in situ measurements and SMCI1.0, the Transformer-downscaled SM product achieved the best accuracy with ubRMSE = 0.0372 m³/m³ and RMSE = 0.0591 m³/m³. The downscaled SM dataset not only captured finer spatial details but also preserved the spatial patterns and seasonal dynamics of the original SMAP product and showed good responsiveness to precipitation events. Feature importance analysis revealed that, aside from precipitation, the diurnal land surface temperature difference had a greater impact on SM than individual daytime or nighttime land surface temperature, ranking just below vegetation indices and soil texture factors, while groundwater level showed higher importance than elevation and surface temperature. This study confirms the effectiveness of Transformer-based models for SM spatial downscaling, providing a novel framework integrating remote sensing and deep hydrological information to generate accurate 1 km SM products.