Hui Liang’s recent publication on Earth System Science Data

Generation of global 1 km daily land surface-air temperature difference and sensible heat flux products from 2000 to 2020

Congratulations to Hui Liang and colleagues on publishing a research article on Earth System Science Data (Vol. 17, pp. 5571–5600). The title of the article is “Generation of global 1 km daily land surface-air temperature difference and sensible heat flux products from 2000 to 2020” and it is now available at https://doi.org/10.5194/essd-17-5571-2025. The daily mean values for sample days can be downloaded from https://doi.org/10.5281/zenodo.14986255, and the complete product is freely available at https://www.glass.hku.hk/.

Accurate estimation of land surface sensible heat flux (H) is crucial for understanding surface energy transfer and water/carbon cycles. However, existing H products are primarily meteorological reanalysis datasets with coarse spatial resolutions, and the only existing satellite-based product (FLUXCOM) has limited resolution and temporal coverage. To address these gaps, this study generated the first global high-resolution (1 km) daily H product for the period 2000–2020 using long short-term memory (LSTM) deep learning models. The study also introduces the first global, accurate satellite-based land surface-air temperature difference (Ts-a) product, derived using Random Forest models, which serves as a key driver for the H estimation.

Validation against independent in-situ measurements indicates that the new products achieve high accuracy, with an RMSE of 25.54 W/m² for H and 1.46 K for Ts-a. The estimated values outperform current products such as MERRA2, ERA5-Land, ERA5, and FLUXCOM under most conditions, offering more detailed spatial information in diverse landscapes.

Flowchart of the study methodology, illustrating the two-step process for estimating Ts-a and H using machine learning models and GLASS products.

Spatial distributions of daily H values on the 121st day of 2010 from the estimated H product (a1) compared with FLUXCOM (a2), MERRA2 (a3), ERA5-Land (a4), and ERA5 (a5). The zoomed-in views (b1-b5 and c1-c5) highlight the new product's ability to capture intricate details in rugged terrain like the Tibetan Plateau.

Reference: Liang, H., Liang, S., Jiang, B., He, T., Tian, F., Ma, H., Xu, J., Li, W., Ma, Y., Zhang, F., & Fang, H. (2025). Generation of global 1 km daily land surface-air temperature difference and sensible heat flux products from 2000 to 2020. Earth System Science Data, 17, 5571–5600. https://doi.org/10.5194/essd-17-5571-2025

Keywords: Sensible heat flux, Land surface-air temperature difference, Deep learning, LSTM, Global, High resolution