Weligamage et al. (2026) Assessing deficiencies in remotely sensed actual evapotranspiration (AET): introducing AET signatures

Identification

• Journal: Hydrology and earth system sciences
• Year: 2026
• Date: 2026-06-30
• Authors: Hansini Gardiya Weligamage, Keirnan Fowler, Margarita Saft, Tim J. Peterson, Dongryeol Ryu, Murray Peel
• DOI: 10.5194/hess-30-4057-2026

Research Groups

• Department of Infrastructure Engineering, The University of Melbourne, Parkville, Victoria, Australia
• Institute of Applied Geosciences, Technische Universität Berlin, Berlin, Germany
• Department of Civil Engineering, Monash University, Clayton, Victoria, Australia

Short Summary

This study introduces eight novel actual evapotranspiration (AET) signatures across daily to annual temporal scales and applies them to systematically evaluate two remotely sensed AET products against flux tower observations at 17 sites in Australia, revealing significant deficiencies in the remotely sensed products' ability to capture AET variability and seasonality.

Objective

• To systematically define and test a set of eight actual evapotranspiration (AET) signatures across various temporal scales (daily, monthly, annual).
• To apply these AET signatures to assess the quality and diagnose deficiencies in two remotely sensed AET products (MODIS and CMRSET) against flux tower observations in Australia.

Study Configuration

• Spatial Scale: 17 OzFlux eddy covariance flux tower sites across diverse climate and ecosystem regions in Australia. Remotely sensed products have pixel resolutions of 500 m (MODIS) and 30 m (CMRSET).
• Temporal Scale: Data periods typically range from 7 to 20 years for flux tower observations. AET signatures are defined and assessed at daily, monthly, and annual scales.

Methodology and Data

• Models used:
  • Moderate Resolution Imaging Spectroradiometer (MODIS, 16A2GFv06.1) AET product.
  • CSIRO MODIS Reflectance-based Scaling Evapotranspiration (CMRSET) product.
• Data sources:
  • Actual evapotranspiration (AET) data from 17 OzFlux eddy covariance flux towers (TERN data portal).
  • Remotely sensed AET (AETRS) products: MODIS (8-day composites, 500 m resolution) and CMRSET (monthly composites, 30 m resolution).
  • Potential evapotranspiration (PET) data: monthly Morton’s wet environment PET from the SILO database.
  • Rainfall data (for event-scale AET responsiveness signature).

Main Results

• Eight AET signatures were proposed: median of annual AET, coefficient of variation of annual AET, periodicity (lag-12 autocorrelation of monthly AET), timing of seasonal peak, coefficient of variation of monthly AET, water stress, AET asynchronicity to PET, and AET responsiveness to a rainfall event.
• Annually, median AETRS generally matched flux tower AET, except for a strong overestimation at three tropical northern Queensland sites.
• AETRS products largely underestimated the interannual and intra-annual (monthly) variability of flux tower AET.
• Monthly indices like water stress and AET asynchronicity to PET were moderately well predicted by AETRS, though MODIS showed higher asynchronicity and CMRSET lower asynchronicity compared to flux tower data for sites with higher than average asynchronicity.
• Seasonal signatures revealed significant discrepancies: MODIS AET exhibited a phase shift and stronger periodicity than observed, while CMRSET showed better timing of seasonal peaks but with larger scatter.
• The regionally-developed CMRSET generally outperformed the globally-developed MOD16A2GFv061, particularly in capturing the timing of seasonal peaks.
• Event-scale AET responsiveness to rainfall (for events greater than 5 mm d⁻¹) showed no discernible correlation at most flux tower sites.

Contributions

• This study is the first to systematically define and test a comprehensive set of eight actual evapotranspiration (AET) signatures across multiple temporal scales (daily to annual).
• It provides a novel framework for assessing various aspects of AET dynamics, offering more nuanced insights into AET behavior than traditional aggregate performance metrics (e.g., Nash-Sutcliffe Efficiency, Kling-Gupta Efficiency).
• The research highlights specific deficiencies in widely used remotely sensed AET products (MODIS and CMRSET) in diverse Australian environments, particularly concerning their representation of AET variability and seasonality.
• The proposed AET signatures offer a valuable tool for selecting appropriate remotely sensed AET products and for diagnosing deficiencies in hydrological models, thereby improving the realism of AET process representation.

Funding

• Australian Research Council (ARC) Linkage Project (LP180100796)
• Victorian Department of Environment, Land, Water and Planning (DELWP)
• Melbourne Water

Citation

@article{Weligamage2026Assessing,
  author = {Weligamage, Hansini Gardiya and Fowler, Keirnan and Saft, Margarita and Peterson, Tim J. and Ryu, Dongryeol and Peel, Murray},
  title = {Assessing deficiencies in remotely sensed actual evapotranspiration (AET): introducing AET signatures},
  journal = {Hydrology and earth system sciences},
  year = {2026},
  doi = {10.5194/hess-30-4057-2026},
  url = {https://doi.org/10.5194/hess-30-4057-2026}
}