Qi et al. (2026) Bridging the thermal gap: Generating 10 m, 3-day land surface temperature via Landsat–Sentinel-2 fusion

Identification

• Journal: Remote Sensing of Environment
• Year: 2026
• Date: 2026-01-02
• Authors: Yuan Qi, Bo Huang, Min Zhao, Xiaolu Jiang, Wenfei Mao
• DOI: 10.1016/j.rse.2025.115227

Research Groups

• Department of Geography, Faculty of Social Sciences, The University of Hong Kong, Hong Kong Special Administrative Region
• Urban Systems Institute, The University of Hong Kong, Hong Kong Special Administrative Region
• Department of Urban Planning & Design, Faculty of Architecture, The University of Hong Kong, Hong Kong Special Administrative Region
• Center for Ocean Research in Hong Kong and Macau (CORE), The Hong Kong University of Science and Technology, Hong Kong Special Administrative Region
• School of Geosciences and Info-physics, Central South University, Changsha, Hunan, China

Short Summary

This study proposes a novel fusion framework (fHiSTR-LST) to generate high spatiotemporal resolution (10 m, ~3-day) land surface temperature (LST) by synergizing Landsat-8/9 and Sentinel-2 data, demonstrating robust performance and outperforming state-of-the-art methods.

Objective

• To address the inherent trade-off between spatial and temporal resolution in current satellite LST products by developing a novel fusion framework (fHiSTR-LST) that generates 10 m spatial resolution LST with an effective ~3-day temporal resolution.

Study Configuration

• Spatial Scale: 10 meters
• Temporal Scale: Approximately 3 days

Methodology and Data

• Models used: fHiSTR-LST (fusion framework), which includes deep-learning-based spatiotemporal fusion for reflectance data and spatial-spectral fusion for LST generation.
• Data sources: Landsat-8/9 (L8/9) optical and thermal observations, Sentinel-2 (S2) optical observations.

Main Results

• The fHiSTR-LST framework reliably produces 10 m spatial resolution LST with an effective ~3-day temporal resolution across joint, L8/9-only, and S2-only overpass scenarios.
• Cross-validation against native L8/9 LST products showed a mean R of 0.90 and a Root Mean Square Error (RMSE) of 1.17 K.
• Ground-truth validation confirmed satisfactory accuracy with a mean R of 0.97 and an RMSE of 3.45 K.
• Combined validation indicated that fHiSTR-LST outperforms state-of-the-art methods by 13 % in R and reduces RMSE by 9 %.
• The generated LST data successfully enabled small-area vegetation-phenology tracking and fine-scale urban thermal-pattern delineation.

Contributions

• Proposes a novel fusion framework (fHiSTR-LST) that effectively bridges the thermal gap to generate unprecedented 10 m spatial resolution LST with an effective ~3-day temporal resolution.
• Achieves the first reported ground-truth validation for such high spatiotemporal resolution LST fusion products, confirming its satisfactory accuracy.
• Demonstrates significant performance improvement over existing state-of-the-art methods in both correlation and error metrics.
• Provides a crucial capability for investigating nuanced effects of global warming, particularly for urban thermal environment analyses and vegetation phenology tracking.

Funding

• Not explicitly mentioned in the provided text.

Citation

@article{Qi2026Bridging,
  author = {Qi, Yuan and Huang, Bo and Zhao, Min and Jiang, Xiaolu and Mao, Wenfei},
  title = {Bridging the thermal gap: Generating 10 m, 3-day land surface temperature via Landsat–Sentinel-2 fusion},
  journal = {Remote Sensing of Environment},
  year = {2026},
  doi = {10.1016/j.rse.2025.115227},
  url = {https://doi.org/10.1016/j.rse.2025.115227}
}