Ojo et al. (2026) Machine Learning‐Based Correction of Reanalysis Surface Radiation Fluxes Using Ground‐Measured Data in the Southern Brazilian Pampa Biome

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: International Journal of Climatology
• Year: 2026
• Date: 2026-08-05
• Authors: Olusola Samuel Ojo, Michel Stefanello, Alecsander Mergen, Maria Eduarda Pinheiro, Vanessa de Arruda Souza, Débora Regina Roberti
• DOI: 10.1002/joc.70528

Research Groups

• University of São Paulo, Institute of Astronomy, Geophysics and Atmospheric Sciences (IAG-USP)
• European Centre for Medium-Range Weather Forecasts (ECMWF)

Short Summary

This study evaluates the performance of ERA-5 and MERRA-2 surface radiation products against hourly observations in Southern Brazilian Pampa, demonstrating that ERA-5 outperforms MERRA-2 for all radiation components. A regression model was developed for gap-filling using reanalysis predictors and refined with machine learning models.

Objective

• Evaluate the accuracy of ERA-5 and MERRA-2 surface radiation products against in-situ observations in subtropical grassland environments.

Study Configuration

• Spatial Scale: Southern Brazilian Pampa, three stations (Santa Maria, Aceguá, and Pedras Altas)
• Temporal Scale: Hourly data from 2014 to 2023

Methodology and Data

• Models used: ERA-5, MERRA-2, Random Forest (RF), K-Nearest Neighbours (KNN), Multilayer Perceptron (MLP)
• Data sources: In-situ observations, European Centre for Medium-Range Weather Forecasts Reanalysis 5th Generation (ERA-5) and Modern-Era Retrospective Analysis for Research and Applications, Version 2 (MERRA-2)

Main Results

• ERA-5 outperformed MERRA-2 for all radiation components, with minimum S↓ RMSE of 5.94 W m−2 (3.1%) and bias of 4.05 W m−2 (2.1%) at Santa Maria.
• A bivariate regression model achieved high predictive skill (R² = 0.986 for S↓ and 0.983 for Q) while RF provided the best performance (R² = 0.9995 for S↓ and 0.9928 for Q at Santa Maria).

Contributions

• This study provides a comprehensive evaluation of ERA-5 and MERRA-2 surface radiation products in subtropical grassland environments, highlighting the potential of ERA-5 for accurate radiation flux reconstruction.
• The developed regression model demonstrates strong predictive skill for gap-filling using reanalysis predictors and machine learning models.

Funding

• This research was funded by the São Paulo Research Foundation (FAPESP) under grant number 2019/07751-2.

Citation

@article{Ojo2026Machine,
  author = {Ojo, Olusola Samuel and Stefanello, Michel and Mergen, Alecsander and Pinheiro, Maria Eduarda and Souza, Vanessa de Arruda and Roberti, Débora Regina},
  title = {Machine Learning‐Based Correction of Reanalysis Surface Radiation Fluxes Using Ground‐Measured Data in the Southern Brazilian Pampa Biome},
  journal = {International Journal of Climatology},
  year = {2026},
  doi = {10.1002/joc.70528},
  url = {https://doi.org/10.1002/joc.70528}
}