Salvi et al. (2026) Discrepancy in the sign of temperature trends in reanalysis datasets

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

Identification

• Journal: Environmental Research Letters
• Year: 2026
• Date: 2026-03-25
• Authors: Kaustubh Salvi, Pushpendra Raghav, Mukesh Kumar, Nicholas R. Magliocca, Gautam Bisht
• DOI: 10.1088/1748-9326/ae5726

Research Groups

[Information not available in the provided abstract.]

Short Summary

This study evaluates the alignment of annual mean daily maximum and minimum temperature trends from three reanalysis datasets (ERA5, MERRA-2, NLDAS-2) against observed trends from 7,059 stations across the continental United States, revealing substantial trend misalignment (21-31%) that persists across various regions and record lengths.

Objective

• To evaluate the sign of annual mean daily maximum (TMAX) and minimum (TMIN) temperature trends from ERA5, MERRA-2, and NLDAS-2 reanalysis datasets against long-term Global Historical Climatology Network daily (GHCN) observations across the continental United States.

Study Configuration

• Spatial Scale: Continental United States (CONUS), covering 7,059 (TMAX) and 6,983 (TMIN) stations.
• Temporal Scale: Long-term annual mean daily temperature trends, including records of 10 to 30 years or more.

Methodology and Data

• Models used: ERA5, MERRA-2, NLDAS-2 (reanalysis datasets).
• Data sources: Global Historical Climatology Network daily (GHCN) observations.

Main Results

• Substantial trend misalignment was observed between GHCN observations and reanalysis data, with approximately 27–31% of TMAX locations and 21–30% of TMIN locations exhibiting discrepancies.
• The misalignment is primarily driven by false positive trends in the reanalysis datasets.
• Misalignment persists even for longer records (≥10-30 years) and is concentrated at stations with strong negative observed trends.
• Trend misalignment prevails irrespective of aggregation approach, including across snow and rain-affected regions, western and eastern snow-dominated areas, urban and non-urban locations, and various elevation classes.
• The NLDAS-2 dataset demonstrated distinct behavior, showing a greater percentage of stations with trend misalignment in snow-affected regions compared to rain-affected areas.
• Within snow-affected domains, NLDAS-2 displayed elevated misalignment across the western United States, contrasting with other datasets.

Contributions

• This study provides a continental-scale evaluation of the directional alignment of temperature trends between widely used reanalysis datasets and ground observations.
• It quantifies significant discrepancies in temperature trend signs, highlighting the need for caution when using reanalysis datasets for regional temperature trend analyses and derived applications.

Funding

[Information not available in the provided abstract.]

Citation

@article{Salvi2026Discrepancy,
  author = {Salvi, Kaustubh and Raghav, Pushpendra and Kumar, Mukesh and Magliocca, Nicholas R. and Bisht, Gautam},
  title = {Discrepancy in the sign of temperature trends in reanalysis datasets},
  journal = {Environmental Research Letters},
  year = {2026},
  doi = {10.1088/1748-9326/ae5726},
  url = {https://doi.org/10.1088/1748-9326/ae5726}
}