Zhang et al. (2025) Global optimization of a water-constrained two-leaf light use efficiency model through multi-biome FLUXNET observations

Identification

• Journal: Agricultural and Forest Meteorology
• Year: 2025
• Date: 2025-09-22
• Authors: Sha Zhang, Wenchao Wang, Jinguo Yuan, Yun Bai
• DOI: 10.1016/j.agrformet.2025.110845

Research Groups

• School of Geographic Sciences, Hebei Normal University, Shijiazhuang, China
• Hebei Technology Innovation Center for Remote Sensing Identification of Environmental Change, Shijiazhuang, China
• Hebei Key Laboratory of Environmental Change and Ecological Construction, Shijiazhuang, China
• Hebei Key Research Institute of Humanities and Social Sciences at Universities “GeoComputation and Planning Center of Hebei Normal University”, Shijiazhuang, China

Short Summary

This study developed and globally optimized a water-constrained two-leaf light use efficiency (WTL-LUE) model using multi-biome FLUXNET observations to improve terrestrial gross primary productivity (GPP) estimation, particularly under water stress. The optimized WTL-LUE model significantly enhanced GPP simulation accuracy and stability across various ecosystems compared to existing LUE and machine learning models.

Objective

• To develop and globally optimize a water-constrained two-leaf light use efficiency (WTL-LUE) model by refining its parameterizations for temperature, vapor pressure deficit, and light response functions using multi-biome FLUXNET observations, thereby improving the accuracy and stability of terrestrial gross primary productivity (GPP) estimation.

Study Configuration

• Spatial Scale: Global, utilizing observations from 201 FLUXNET sites covering ten distinct ecosystems.
• Temporal Scale: Model performance evaluated at daily and 8-day scales.

Methodology and Data

• Models used:
  • Developed: WTL-LUE (Water-constrained Two-Leaf Light Use Efficiency model)
  • Basis: RTL-LUE (Revised Two-Leaf Light Use Efficiency model)
  • Benchmarked against: MOD17, VPM (Vegetation Photosynthesis Model), TL-LUE (Two-Leaf Light Use Efficiency model), RTL-LUE, and XGBoost (machine learning approach).
• Data sources: FLUXNET2015 dataset (eddy covariance observations of GPP and meteorological variables).

Main Results

• The optimized WTL-LUE model achieved an R² of 0.71 (RMSE = 2.23 gC m⁻² d⁻¹) for daily GPP estimation and an R² of 0.74 (RMSE = 2.03 gC m⁻² d⁻¹) for 8-day GPP estimation.
• WTL-LUE demonstrated superior performance compared to existing LUE models (MOD17, VPM, TL-LUE, RTL-LUE), especially in dryland ecosystems (savannas, shrublands) and specific vegetation types (croplands, deciduous broadleaf forests, wetlands).
• The model also exhibited relative stability advantages over the benchmarked XGBoost machine learning approach across environmental gradients.

Contributions

• Developed an improved water-constrained two-leaf light use efficiency model (WTL-LUE) with refined parameterizations for meteorological constraints and a nonlinear photosynthesis response to light.
• Performed a global optimization of the WTL-LUE model using extensive multi-biome FLUXNET observations, leading to significantly enhanced GPP estimation accuracy.
• Demonstrated the critical importance of integrating meteorological data with remote sensing for accurate water stress representation in GPP models.
• Provided a robust and stable tool for analyzing global ecosystem dynamics and their responses to climate change, outperforming both traditional LUE models and a machine learning approach.

Funding

• Not specified in the provided text.

Citation

@article{Zhang2025Global,
  author = {Zhang, Sha and Wang, Wenchao and Yuan, Jinguo and Bai, Yun},
  title = {Global optimization of a water-constrained two-leaf light use efficiency model through multi-biome FLUXNET observations},
  journal = {Agricultural and Forest Meteorology},
  year = {2025},
  doi = {10.1016/j.agrformet.2025.110845},
  url = {https://doi.org/10.1016/j.agrformet.2025.110845}
}