Hydrology and Earth System Sciences
Hydrology and Earth System Sciences
HESS
Articles | Volume 30, issue 10
Articles | Volume 30, issue 10
https://doi.org/10.5194/hess-30-3283-2026
© Author(s) 2026. This work is distributed under
the Creative Commons Attribution 4.0 License.
https://doi.org/10.5194/hess-30-3283-2026
© Author(s) 2026. This work is distributed under
the Creative Commons Attribution 4.0 License.
Articles | Volume 30, issue 10
Research article
 | Highlight paper
 | 
27 May 2026
Research article | Highlight paper |  | 27 May 2026

A novel classifier-guided ensemble framework for global terrestrial evapotranspiration estimates

Le Ni, Weiguang Wang, Jianyu Fu, and Mingzhu Cao

Data sets

GLDAS Noah Land Surface Model L4 monthly 0.25 x 0.25 degree V2.1 H. Beaudoing et al. https://doi.org/10.5067/SXAVCZFAQLNO

MCD12C1 MODIS/Terra+Aqua Land Cover Type Yearly L3 Global 0.05Deg CMG V006 M. Friedl and D. Sulla-Menashe https://doi.org/10.5067/MODIS/MCD12C1.006

ET Dataset for paper ``A Novel Classifier-Guided Ensemble Framework for Global Terrestrial Evapotranspiration Estimates' L. Ni et al. https://doi.org/10.5281/zenodo.18543737

Download
Editorial statement
Evapotranspiration estimates often contain substantial uncertainties, while direct observations remain sparse. This study offers a practical and innovative framework for selecting the most appropriate model for Global Terrestrial Evapotranspiration Estimates for specific locations. The authors present a Classifier-Guided Ensemble approach that integrates process-based algorithms, machine learning models, and hybrid methods to identify the most suitable model for a given setting and subsequently estimate evapotranspiration (ET).
Read more
Short summary
Existing global evapotranspiration algorithms rely on sparse local measurements and each comes with its own strengths and weaknesses. Here, we proposed an ensemble framework that employed a machine learning system to dynamically select the most appropriate algorithm to be used across spatial and temporal scales, thus fully utilizing the distinct strengths of each method. In multi-scale validations, our framework exhibited enhanced extrapolation performance, stability, and interpretability.
Read more
Share