Ranjbar et al. (2026) Near real-time estimation of daytime and nighttime evapotranspiration using GOES-R observations and machine learning models

Identification

• Journal: Hydrology and earth system sciences
• Year: 2026
• Date: 2026-08-05
• Authors: Sadegh Ranjbar, Danielle Losos, Sophie Hoffman, Yafang Zhong, Jason A. Otkin, Ankur Rashmikant Desai, Martha C. Anderson, Christopher Hain, Paul C. Stoy
• DOI: 10.5194/hess-30-4927-2026

Research Groups

• Department of Biological Systems Engineering, University of Wisconsin – Madison
• Department of Earth and Planetary Sciences, Yale University
• Department of Geography, University of Colorado Boulder
• Department of Atmospheric and Oceanic Sciences, University of Wisconsin – Madison
• Space Sciences and Engineering Center, University of Wisconsin – Madison
• USDA-ARS Hydrology and Remote Sensing Laboratory
• NASA Marshall Space Flight Center, Earth Science Branch

Short Summary

This study introduces a near real-time estimation framework for daytime and nighttime evapotranspiration using geostationary satellite observations from the GOES-R Advanced Baseline Imager (ABI) and machine learning models. The framework, called ALIVE ET, leverages high-frequency data to estimate sub-daily ET dynamics under both clear and cloudy conditions.

Objective

• To develop a near real-time estimation framework for evapotranspiration using geostationary satellite observations and machine learning models.
• To evaluate the performance of the framework in estimating daytime and nighttime ET under various sky conditions.

Study Configuration

• Spatial Scale: Continental scale, covering the Contiguous United States (CONUS).
• Temporal Scale: 5-minute interval, with a focus on sub-daily ET dynamics.

Methodology and Data

• Models used:
  • Gradient Boosting Regression (GBR)
  • Long Short-Term Memory (LSTM) networks
• Data sources:
  • GOES-R Advanced Baseline Imager (ABI) observations
  • Eddy covariance tower data from AmeriFlux and NEON sites

Main Results

• The ALIVE ET framework demonstrates strong performance in estimating daytime and nighttime ET under various sky conditions.
• GBR outperforms LSTM for daytime ET estimation, achieving an R2 of 0.74 and a normalized RMSE of 0.91.
• LSTM performs better than GBR for nighttime ET estimation, achieving an R2 of 0.24 and a normalized RMSE of 1.29.

Contributions

• This study demonstrates the potential of integrating machine learning with geostationary remote sensing to advance high-temporal-resolution ET estimation.
• The ALIVE ET framework provides near real-time estimates of sub-daily ET dynamics under both clear and cloudy conditions, enabling improved water cycle monitoring and management.

Funding

• National Science Foundation (NSF)
• NASA
• USDA ARS

Citation

@article{Ranjbar2026Near,
  author = {Ranjbar, Sadegh and Losos, Danielle and Hoffman, Sophie and Zhong, Yafang and Otkin, Jason A. and Desai, Ankur Rashmikant and Anderson, Martha C. and Hain, Christopher and Stoy, Paul C.},
  title = {Near real-time estimation of daytime and nighttime evapotranspiration using GOES-R observations and machine learning models},
  journal = {Hydrology and earth system sciences},
  year = {2026},
  doi = {10.5194/hess-30-4927-2026},
  url = {https://doi.org/10.5194/hess-30-4927-2026}
}