Conference Agenda

The Soil Moisture and Ocean Salinity satellite launched by the European Space Agency in 2009 represents the first Earth Explorer mission equipped with an L-band microwave radiometer and has operated successfully for fifteen years providing critical data for global water cycle and climate change research. Despite its achievements the mission faces inherent limitations including coarse spatial resolution inconsistency with complementary satellite products and incomplete spatial temporal coverage. This project systematically addresses these challenges through innovative algorithm development and multi-source data fusion achieving substantial progress. Regarding product consistency enhancement the research team developed a monthly sliding window linear regression method to correct systematic brightness temperature biases between SMOS and SMAP satellite products. Following correction the agreement between soil moisture and vegetation optical depth products improved markedly with enhanced correlations against biomass and canopy height measurements. The fusion of these two major L-band satellite missions substantially expanded global spatial coverage beyond what either sensor achieves independently. For spatial temporal continuity reconstruction the project pioneered PhyFill a physics-constrained deep learning approach that incorporates precipitation monotonicity constraints and soil moisture dry-down curve boundary conditions. This method successfully transformed gap-ridden satellite observations into seamless global daily products with reconstruction accuracy comparable to original observations. Complementing this effort the TransCNN architecture was developed by integrating Transformer and convolutional neural network components demonstrating superior long-term dependency modeling and improved correlation performance over conventional methods. Spatial resolution enhancement was pursued through multiple complementary strategies. A multi-frequency passive microwave downscaling framework utilizing higher frequency satellite data enabled the refinement of coarse L-band observations to finer spatial scales. The DespiTe method based on optical trapezoid model theory was introduced as a standalone approach requiring no thermal infrared inputs while achieving competitive accuracy against airborne validation data. Additionally a thermal-inertia-based downscaling algorithm was developed showing favorable performance characteristics and a specialized normalization technique for mountainous terrain areas improved retrieval accuracy through correction of terrain-induced angular effects. Data record extension was accomplished by harmonizing observations from multiple satellite sensors spanning several decades through machine learning techniques. The resulting long-term global daily soil moisture product maintains extensive global land coverage and has been made publicly available through established scientific data repositories providing valuable continuity for climate studies and trend analysis. The added value of these improved products was demonstrated through comprehensive evaluation of their capability to detect hydrological extreme events. Systematic comparison across different satellite bands and retrieval algorithms revealed that L-band sensors consistently outperform higher frequency alternatives in capturing drought and flood conditions with superior sensitivity and spatial accuracy. Real-case validation across diverse extreme weather events confirmed these patterns establishing the preferential suitability of L-band observations for early warning applications. Through these interconnected advances the project has substantially elevated the quality accessibility and practical utility of satellite-based soil moisture monitoring contributing to enhanced capabilities in global water resource assessment.

Glaciers in High Mountain Asia (HMA) are important water sources for millions of people in arid regions. However, their mass balance exhibits strong regional heterogeneity. The reasons are still not well understood mainly due to limited high-resolution temperature and precipitation datasets in high-altitude zones. Here, we propose a glacier ablation and accumulation index based on firn and snow observation from Sentinel-1 synthetic aperture radar data, which represent the seasonal ablation and accumulation patterns across HMA. Evaluations of the index with the summer snow line, precipitation, and temperature records from supraglacial weather stations confirm their reliability. Comparisons to available mass balance data show that glaciers with the lowest ablation and lowest summer accumulation index which are located in the center of the HMA region, have the most balanced state (+0.05 m w.e. a^-1). In contrast, glaciers with a high ablation and median summer accumulation index, show highly negative mass balances (-0.47 m w.e. a^-1). Under similar ablation index, the type of glaciers with higher summer accumulation index is more negative. No matter of the seasonal accumulation type, a year with higher summer accumulation normally leads to less negative or more positive mass balance. This work provides novel methods and knowledge to understand and project the glacier mass balance heterogeneity in HMA.