EVAonline: An open-source web platform for global reference evapotranspiration estimation via multi-source data fusion

This dataset provides the complete validation results and reproducibility package for EVAonline, an open-source system designed for high-accuracy reference evapotranspiration (ETo) estimation using multi-source data fusion. 📦 What's Included 186,287 daily observations spanning 30 years (1991–2020) across 17 Brazilian cities (MATOPIBA region + Piracicaba/SP). Four 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, ~9 km resolution). Fused product: EVAonline Adaptive Kalman Fusion results. 📂 Repository Structure 📁 data/ - Raw inputs, intermediate files, and final figures. 📁 notebooks/ - Demonstration notebooks for each provider (NASA, Open‑Meteo, MET Norway, NWS) and utility examples 📁 scripts/ - API adapters, Kalman fusion core, FAO‑56 Penman‑Monteith engine, and sequential pipeline (scripts 1 to 7) to reproduce all manuscript results. 📄 Root Files - complete_validation_analysis.ipynb (main entry point), tutorial_full_pipeline.ipynb (step‑by‑step guide using Piracicaba/SP), environment.yml and requirements.txt (Python 3.12 environment). 🔬Methodology Notes Input vs. output distinction: open_meteo_raw and nasa_power_raw contain raw meteorological variables (temperature, humidity, wind, radiation) used as inputs to EVAonline’s FAO‑56 PM engine. eto_open_meteo contains ETo values pre‑calculated by the Open‑Meteo service, used for benchmarking. Validation Scope: Focuses on the mathematical core (Kalman filter, weighted fusion, FAO‑56) – not the full web platform (FastAPI, Celery, Redis). Climatological rigor: 30‑year period (1991–2020) follows WMO guidelines for climate normals. 📈Scientific Performance EVAonline significantly outperforms individual global sources: Metric EVAonline Nasa Power Open-Meteo MAE (mm/d) 0.42 0.84 0.86 PBIAS (%) +0.71 +15.78 +13.02 KGE 0.81 0.41 0.43 MAE reduction: 38.7–50.8% Bias reduction: 91.4–95.5% 📄 License and Citation Data license: Creative Commons Attribution 4.0 International (CC BY 4.0). Software license: GNU AGPL‑3.0 (see GitHub repository). Recommended citation: Cunha Soares, Â. S. M., Ribeiro, V., Duarte, S., PERRELLA BALESTIERI, J. A., Padovani, C. R., Bordignon, Á. J. Z., Maciel, C., & Marques, P. (2026). EVAonline: An open-source web platform for global reference evapotranspiration estimation via multi-source data fusion (1.0.1) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.19339254 🔗 Links Live repository: https://github.com/angelacunhasoares/EVAONLINE Associated article: Submitted to Computers and Electronics in Agriculture (2026)