Brown et al. (2026) Evaluation of high-resolution meteorological data products using flux tower observations across Brazil

Identification

• Journal: Hydrology and earth system sciences
• Year: 2026
• Date: 2026-01-13
• Authors: Jamie R. C. Brown, Ross Woods, Débora Regina Roberti, Rafael Rosolem
• DOI: 10.5194/hess-30-141-2026

Research Groups

• School of Civil Aerospace and Design Engineering, University of Bristol, Bristol, UK
• Instituto de Astronomia, Geofísica, e Ciências Atmosféricas, Universidade de São Paulo, São Paulo, Brazil
• Departamento de Física, Universidade Federal de Santa Maria, Santa Maria, Brazil
• Cabot Institute for the Environment, University of Bristol, Bristol, UK

Short Summary

This study evaluates the quality of five high-resolution global and regional meteorological datasets against flux tower observations at 11 sites across Brazil. It finds that ERA5-Land generally performs best for most variables, while MSWEPv2.2 excels in daily precipitation and the Brazilian National Meteorological Database (BNMD) in monthly precipitation, with no single product consistently superior across all variables and temporal scales.

Objective

• To assess the quality of four global reanalysis products (ERA5-Land, GLDAS2.0, GLDAS2.1, MSWEPv2.2) and one regional gridded dataset (BNMD) using local in-situ measurements across Brazil.
• To identify which high-resolution gridded product is most accurate overall and for each specific meteorological variable.
• To determine the dominant types of error (correlation, bias, variation) associated with each product and how these errors vary spatially and seasonally.

Study Configuration

• Spatial Scale: Continental scale, covering 11 flux tower sites across Brazil's main land cover types (tropical rainforest, woodland savanna, various croplands, tropical dry forests). Gridded products have resolutions ranging from 0.1° to 0.25°.
• Temporal Scale: Flux tower observations span periods between 1999 and 2014. Gridded products cover longer periods (e.g., 1948-present). Analysis was performed at daily and monthly temporal resolutions.

Methodology and Data

• Models used:
  • ERA5-Land: Generated using the tiled ECMWF Scheme for Surface Exchanges over Land incorporating land surface hydrology (H-TESSEL).
  • GLDAS2.0/2.1: Products of the NOAH-3.6 Land Surface Model (LSM) forced with meteorological datasets (Princeton meteorological forcing dataset for GLDAS2.0; combined forcing including Global Precipitation Climatology Project for GLDAS2.1).
  • MSWEPv2.2: Combines satellite remote sensing data with multiple sources of reanalysis products, bias-corrects, and weights between multiple nearby observation gauges.
  • BNMD: Developed from local interpolation of meteorological variables across Brazil.
• Data sources:
  • In-situ observations: Meteorological data from 11 FLUXNET flux tower sites across Brazil, providing independent measurements of precipitation, air temperature, wind speed, atmospheric pressure, downward shortwave and longwave radiation, and specific humidity.
  • Gridded products:
    • Brazilian National Meteorological Gridded Database (BNMD)
    • Global Land Data Assimilation System (GLDAS) 2.0
    • Global Land Data Assimilation System (GLDAS) 2.1
    • ECMWF Reanalysis 5-Land (ERA5-Land)
    • Multi-Source Weighted-Ensemble Precipitation v2.2 (MSWEPv2.2)
  • Methodology: Data quality control, temporal averaging to daily and monthly scales, vertical interpolation for wind speed, estimation of atmospheric pressure (for BNMD) and specific humidity (for ERA5-Land and BNMD). Mean Squared Error (MSE) decomposition into correlation, variation, and bias contributions, and a ranking system based on MSE for performance evaluation.

Main Results

• ERA5-Land consistently achieved the best ranking (lowest errors) for air temperature, atmospheric pressure, downward longwave radiation, and specific humidity at both daily and monthly scales, and for wind speed and downward shortwave radiation at the daily scale.
• For precipitation, MSWEPv2.2 outperformed other datasets at daily scales, while BNMD performed best at monthly scales.
• GLDAS2.0 generally performed least well at both temporal scales for most variables, though GLDAS2.1 showed improvements over its predecessor.
• BNMD wind speed and GLDAS2.0 shortwave radiation outperformed other datasets at a monthly scale.
• The largest contribution to the MSE at the daily scale for most variables was the correlation contribution, indicating issues with capturing temporal patterns.
• At the monthly scale, the bias contribution became the dominant source of error for most variables, suggesting systematic over- or under-predictions.
• Errors exhibited seasonality, with precipitation errors generally lower in dry seasons and increasing in wetter seasons, and air temperature biases varying with warmer/cooler months.

Contributions

• Provides a comprehensive and independent validation of five high-resolution global and regional meteorological data products against flux tower observations in Brazil, a region with limited observational coverage.
• Utilizes Mean Squared Error (MSE) decomposition to offer detailed insights into the specific sources of error (correlation, bias, variation) for each variable and how these contributions shift with temporal resolution.
• Offers practical recommendations for selecting the most suitable gridded meteorological product for different variables and temporal scales, which is crucial for hydrological and land surface modeling applications in Brazil.
• Highlights the critical importance of regional validation for global data products, as no single product consistently outperforms others across all variables or temporal scales.

Funding

• The Engineering and Physical Sciences Research Council (EPSRC) Water Informatics: Science and Engineering Centre for Doctoral Training (WISE-CDT; grant no. EP/L016214/1)
• Brazilian Experimental datasets for MUlti-Scale interactions in the critical zone under Extreme Drought (BEMUSED; grant no. NE/R004897/1)
• COSMIC-SWAMP project (Natural Environment Research Council (NERC) grant NE/W004364/1 and São Paulo Research Foundation (FAPESP) grant 2021/03032-7)
• FAPESP grant 2021/11762-5
• CTG-IAG P&D/Aneel – Sistema inteligente para benefício da qualidade das informações climáticas e ampliação da rede de estações meteorológicas no setor de energia hidro-eólica-solar do Brasil

Citation

@article{Brown2026Evaluation,
  author = {Brown, Jamie R. C. and Woods, Ross and Rocha, Humberto Ribeiro da and Roberti, Débora Regina and Rosolem, Rafael},
  title = {Evaluation of high-resolution meteorological data products using flux tower observations across Brazil},
  journal = {Hydrology and earth system sciences},
  year = {2026},
  doi = {10.5194/hess-30-141-2026},
  url = {https://doi.org/10.5194/hess-30-141-2026}
}