A High-Resolution Daily Multi-Layer Soil Temperature Dataset for the Tibetan Plateau (2001–2022)

TPSoilTemp v1.0 is a high-resolution daily multi-layer soil temperature dataset for the Tibetan Plateau covering 2001–2022. The dataset provides daily mean, maximum, and minimum soil temperatures at depths of 10, 20, and 40 cm with a spatial resolution of 1/30° (~3 km). The dataset was generated using a hierarchical framework integrating multi-source data evaluation, Triple Collocation uncertainty-weighted fusion, LightGBM spatial downscaling, and residual correction. Eight reanalysis / land surface model products were first evaluated against nearly 300 in situ soil temperature stations from 17 observation networks across the Tibetan Plateau. ERA5-Land, JRA-3Q, and GLDAS-CLSM were selected and fused to produce a 0.25° soil temperature product, which was subsequently downscaled to 1/30° resolution using multiple environmental variables. The final dataset accurately reproduces the spatiotemporal variability of soil temperature across the Tibetan Plateau, with RMSEs of daily maximum soil temperature of 3.15, 2.97, and 3.28 °C at depths of 10, 20, and 40 cm, respectively. The dataset can support studies of soil heatwaves, permafrost degradation, land–atmosphere interactions, and ecosystem responses to climate change in high-altitude cold regions.