Chinese Researchers Launch Global High-Precision Soil Freeze-Thaw Dataset (2002-2023)

Understanding Soil Freeze-Thaw Cycles: A Key Indicator of Global Climate Dynamics

Soil freeze-thaw (F/T) cycles refer to the repeated freezing and thawing of the near-surface soil layer, typically the top 0-5 cm, driven by temperature fluctuations. These cycles are fundamental to numerous Earth system processes. In hydrology, frozen soil acts as an impermeable barrier, reducing infiltration and increasing surface runoff, which can lead to flooding during thaws. Ecologically, F/T timing synchronizes with vegetation phenology, influencing spring green-up and autumn dormancy. In permafrost regions, thawing releases ancient organic carbon, amplifying greenhouse gas emissions and accelerating climate warming—a positive feedback loop.7376 With global warming intensifying, accurate, long-term monitoring of F/T states is essential for modeling these interactions and predicting environmental changes.

Historically, datasets have been limited by coarse spatial resolution (often 25 km or more) from passive microwave sensors like AMSR-E/AMSR2, missing fine-scale heterogeneity due to topography, vegetation, and soil properties. Optical data offers higher resolution but suffers from cloud cover gaps. The need for seamless, high-resolution global records has long been recognized by the Global Climate Observing System (GCOS), designating F/T as an Essential Climate Variable (ECV).

Breakthrough Release: The FT-HiDFA Dataset by Chinese Researchers

A team from the Aerospace Information Research Institute (AIR) of the Chinese Academy of Sciences (CAS), in collaboration with the University of Chinese Academy of Sciences (UCAS), has unveiled the Global Near-Surface Soil Freeze-Thaw Dataset (FT-HiDFA). Spanning 2002 to 2023, this daily product achieves an unprecedented 0.05° spatial resolution (approximately 5 km at the equator), covering all global land surfaces without gaps.60

Published in Earth System Science Data in November 2025, the dataset addresses previous limitations through innovative data fusion, providing researchers with a 'physical examination' of global soil health over two decades.60 Lead author Defeng Feng and colleagues highlight its role in 'accurately monitoring FT states and providing detailed information for a refined understanding of hydrological and ecological effects globally.'

Innovative Fusion of Microwave and Optical Data

The FT-HiDFA employs the Discriminant Function Algorithm (DFA) on AMSR-E/AMSR2 brightness temperatures at 36.5 GHz vertical polarization to initially classify F/T states at 0.25° resolution. Quasi-emissivity (Qe) corrects for vegetation and atmospheric effects. To downscale to 0.05°, the team integrates MODIS Land Surface Temperature (LST) and GLASS albedo-derived Apparent Thermal Inertia (ATI). A bivariate linear regression model (FTI = a · LST + b · ATI + c) is fitted annually, resampling coarse F/T to fine grids while filling gaps.60

• Microwave Input: Daily descending/ascending orbits, robust to clouds.
• Optical Enhancement: High-res (1 km) LST/ATI for spatial detail.
• Downscaling: Pixel-by-pixel regression ensures continuity.

This hybrid approach yields seamless daily maps, validated against 1027 in situ stations from 44 networks worldwide, achieving 83.78% overall accuracy—comparable to coarse products but with vastly improved detail.

Rigorous Validation and Superior Performance

Validation compared FT-HiDFA against ground soil temperatures at 0-5 cm depth. Accuracies exceed 90% in over 60% of networks, with RMSE for frost days under 30 days. The product excels in heterogeneous terrains like the Tibetan Plateau, where coarse data often fails. Sen's slope and Mann-Kendall tests confirm reliability for trend analysis.60

Users can access the full dataset via the National Tibetan Plateau and Third Pole Environment Data Center at doi:10.11888/Cryos.tpdc.301551.

Revealing Global Trends in Frost Days and Freeze Onset

Analysis shows northern latitudes (>45°N) average 187.8 frost days annually. Decreasing trends dominate (14.35% of land, 2.67% significant), signaling warming, while 11.17% show increases. Freeze onset averages Julian day 240.3, with earlier onset in 9.10% areas.60 These patterns underscore regional climate variability, e.g., reduced persistence in Eurasia vs. delayed freezing in North America.

Tibetan Plateau: A Permafrost Hotspot Under Scrutiny

The Qinghai-Tibet Plateau, 'Third Pole,' holds 1.6% of global permafrost. FT-HiDFA reveals detailed frost day distributions aligning with permafrost zones, aiding thaw impact studies. Over 20 years, thawing accelerates carbon release, risking 0.1-0.2 GtC/year emissions.71 Chinese researchers' high-res data fills critical gaps for regional models.

Hydrological Impacts: Flooding, Drought, and Water Cycles

F/T controls soil permeability; frozen states boost runoff, thaws enhance infiltration. Dataset trends predict altered river flows, e.g., earlier snowmelt floods. In boreal forests, frequent cycles erode soil, worsening droughts.75

Ecological Consequences: Vegetation, Biodiversity, and Phenology

Thawing desynchronizes plant cycles, reducing productivity. Dataset enables phenology modeling, revealing F/T as a spring green-up trigger or inhibitor.72 Biodiversity suffers from habitat shifts in permafrost edges.

Carbon Cycle Feedbacks and Permafrost Thaw Risks

Permafrost stores 1300-1600 GtC; thaw mobilizes it via microbial activity, potentially adding 100 GtC by 2100. FT-HiDFA quantifies cycle frequency, linking to CH4/CO2 pulses.80

Applications in Climate Modeling and Policy

Integrate into CMIP7 models for better ECV representation. Supports IPCC assessments, disaster risk (erosion), agriculture (crop damage). Freely available, it democratizes access for global scientists.

China's Leadership in Remote Sensing Earth Observation

AIR CAS exemplifies China's remote sensing prowess, with UCAS training next-gen researchers. This advances 'Double First-Class' initiatives, positioning Chinese academia at forefront of cryosphere science.

Future Outlook: Enhancing Resolution and Extending Records

Ongoing work fuses Sentinel-1 SAR for all-weather coverage. Dataset paves way for real-time monitoring, AI-driven predictions, and interdisciplinary studies. Researchers worldwide can now dissect F/T-climate feedbacks with precision.

Photo by Conor Murphy on Unsplash