Soares et al. (2026) EVAonline: An open-source web platform for global reference evapotranspiration estimation via multi-source data fusion

Identification

• Journal: Open MIND
• Year: 2026
• Date: 2026-03-30
• Authors: Ângela S M Cunha Soares, Vitor Ribeiro, S.X. Duarte, José Antônio Perrella Balestieri, Carlos Roberto Padovani, Álex Júnior Zanchet Bordignon, Carlos Maciel, P. V. S. Marques
• DOI: 10.5281/zenodo.19339254

Research Groups

• Universidade de São Paulo Escola Superior de Agricultura Luiz de Queiroz
• Universidade de São Paulo
• Universidade Estadual Paulista (Unesp)
• Universidade Estadual Paulista Júlio de Mesquita Filho - Câmpus de Guaratinguetá
• Brazilian Agricultural Research Corporation

Short Summary

EVAonline is an open-source web platform designed for high-accuracy global reference evapotranspiration (ETo) estimation through multi-source data fusion, demonstrating significant performance improvements over individual global data sources.

Objective

• To develop and validate EVAonline, an open-source web platform for high-accuracy global reference evapotranspiration (ETo) estimation using an adaptive Kalman fusion approach with multi-source meteorological data.

Study Configuration

• Spatial Scale: 17 Brazilian cities (MATOPIBA region and Piracicaba/SP). Data sources range in resolution from 0.1° to 0.5° and approximately 9 km.
• Temporal Scale: 30 years (1991–2020) of daily observations.

Methodology and Data

• Models used: FAO-56 Penman-Monteith engine, Adaptive Kalman Filter for data fusion, weighted fusion.
• Data sources:
  • Reference (ground truth): Xavier et al. BR-DWGD (0.1° resolution).
  • Reanalysis A: NASA POWER (MERRA-2, 0.5° resolution).
  • Reanalysis B: Open-Meteo Archive (ERA5-Land, approximately 9 km resolution).
  • Fused product: EVAonline Adaptive Kalman Fusion results.

Main Results

• EVAonline significantly outperforms individual global sources in reference evapotranspiration (ETo) estimation.
• Mean Absolute Error (MAE) for EVAonline is 0.42 mm/d, a reduction of 38.7–50.8% compared to NASA POWER (0.84 mm/d) and Open-Meteo (0.86 mm/d).
• Percentage Bias (PBIAS) for EVAonline is +0.71%, representing a 91.4–95.5% reduction compared to NASA POWER (+15.78%) and Open-Meteo (+13.02%).
• Kling-Gupta Efficiency (KGE) for EVAonline is 0.81, substantially higher than NASA POWER (0.41) and Open-Meteo (0.43).

Contributions

• Development of EVAonline, an open-source web platform for global reference evapotranspiration estimation, integrating multi-source data fusion.
• Implementation and validation of an Adaptive Kalman Filter for combining diverse meteorological datasets to enhance ETo accuracy.
• Provision of a comprehensive validation dataset and a reproducible package for the core mathematical components of the EVAonline system.
• Demonstrated significant improvements in ETo estimation accuracy and bias reduction compared to existing individual global data sources.

Funding

• Funding information is not explicitly provided in the text.

Citation

@article{Soares2026EVAonline,
  author = {Soares, Ângela S M Cunha and Ribeiro, Vitor and Duarte, S.X. and Balestieri, José Antônio Perrella and Padovani, Carlos Roberto and Bordignon, Álex Júnior Zanchet and Maciel, Carlos and Marques, P. V. S.},
  title = {EVAonline: An open-source web platform for global reference evapotranspiration estimation via multi-source data fusion},
  journal = {Open MIND},
  year = {2026},
  doi = {10.5281/zenodo.19339254},
  url = {https://doi.org/10.5281/zenodo.19339254}
}