Wang et al. (2025) Gap-Filling for Daily Latent Heat Flux Observations with the Full-factorial method at Global Flux Sites

Identification

• Journal: Scientific Data
• Year: 2025
• Date: 2025-11-28
• Authors: Xiaowei Wang, Fujiao Tang, Yazhen Jiang, Yunsheng Lou
• DOI: 10.1038/s41597-025-06144-9

Research Groups

• State Key Laboratory of Resources and Environment Information System, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences
• School of Ecology and Applied Meteorology, Nanjing University of Information Science & Technology
• School of Transportation Science and Engineering, Harbin Institute of Technology
• University of Chinese Academy of Sciences

Short Summary

This study developed and validated a novel median-adjusted full-factorial and iterative method to accurately fill gaps in daily latent heat flux (LE) observations from 265 global eddy covariance sites, demonstrating superior performance compared to existing LE products and providing a high-quality, continuous dataset.

Objective

• To process site data from multiple global flux networks, gap-fill daily latent heat flux (LE) observations using a median-adjusted full-factorial method, and conduct a comprehensive verification analysis of the filled LE data.

Study Configuration

• Spatial Scale: Global, encompassing 265 flux sites (1509 site-years) with diverse land cover types and climate regimes. Reanalysis data were spatially resampled to a 0.01° grid.
• Temporal Scale: Daily resolution, derived from half-hourly eddy covariance measurements, spanning 1 to 32 years for individual sites.

Methodology and Data

• Models used:
  • Median-adjusted full-factorial method for gap-filling, based on the Penman-Monteith equation and incorporating atmospheric, vegetation, and soil factors.
  • Iterative gap-filling process, where newly filled data are used to enable filling of previously intractable gaps.
• Data sources:
  • Observation: Half-hourly latent heat flux (LE) and meteorological data from global eddy covariance flux networks (AmeriFlux, FLUXNET, EuroFlux, OzFlux, ChinaFlux, National Tibet Plateau Data Center (TPDC), National Cryosphere Desert Data Center (NCDDC)).
  • Reanalysis: ERA5-Land (for air temperature, atmospheric pressure, wind speed, relative humidity, vapor pressure deficit, net radiation), Global Land Data Assimilation System (GLDAS) (for soil heat flux), Modern-Era Retrospective Analysis for Research and Applications Version 2 (MERRA-2) (for roughness length for momentum transfer).
  • Products for intercomparison: Global Land Evaporation Amsterdam Model (GLEAM) v3 and ERA5-Land LE products.

Main Results

• The gap-filling framework achieved high accuracy for isolated gaps, with mean Root Mean Square Error (RMSE) of 16.53 W/m², Mean Absolute Error (MAE) of 10.58 W/m², and R² of 0.81. Shrubland sites showed the highest performance (RMSE: 14.69 W/m², MAE: 8.93 W/m²).
• For long-term continuous gaps (5–50 days), the average RMSE was 24.41 W/m². Accuracy declined with increasing gap duration (e.g., for forests, RMSE increased from 21.69 W/m² for 5-day gaps to 31.23 W/m² for 50-day gaps).
• The gap-filled data significantly outperformed ERA5-Land and GLEAM LE products across all land cover types and gap scenarios. For isolated gaps, the gap-filled data had a mean R² of 0.81 compared to 0.37 for the products.
• Gap-filled LE data effectively captured seasonal trends and long-term variations, showing strong consistency with observed values.
• The energy balance closure for gap-filled LE was satisfactory, with an average Energy Balance Ratio (EBR) of 0.79 across sites and R² values ranging from 0.86 to 0.92 for turbulent fluxes versus available energy.
• Sensitivity analysis revealed that net radiation (Rn) and air temperature (Ta) perturbations had the strongest influence on gap-filling accuracy, with RMSE changes ranging from -2.2 to 8.5 W/m² and MAE changes from -0.4 to 3.6 W/m².

Contributions

• Presents a novel, physically-based median-adjusted full-factorial and iterative framework for robust and accurate gap-filling of daily latent heat flux (LE) observations.
• Generates a comprehensive, high-quality, and continuous daily LE dataset for 265 global flux sites (1509 site-years), addressing a critical need for continuous observational data.
• Demonstrates superior accuracy of the developed gap-filling method compared to widely used reanalysis and model-based LE products (ERA5-Land and GLEAM).
• Provides a valuable, publicly available dataset that can serve as a reference for LE model validation, water demand assessments, and studies on water-energy-carbon cycles.
• Quantifies the sensitivity of gap-filling accuracy to meteorological input uncertainties, highlighting the critical importance of high-quality air temperature and net radiation data.

Funding

• National Natural Science Foundation of China (grant: 42371368)
• Postgraduate Research & Practice Innovation Program of Jiangsu Province (grant: KYCX24_1449)

Citation

@article{Wang2025GapFilling,
  author = {Wang, Xiaowei and Tang, Fujiao and Jiang, Yazhen and Lou, Yunsheng},
  title = {Gap-Filling for Daily Latent Heat Flux Observations with the Full-factorial method at Global Flux Sites},
  journal = {Scientific Data},
  year = {2025},
  doi = {10.1038/s41597-025-06144-9},
  url = {https://doi.org/10.1038/s41597-025-06144-9}
}