Zhou et al. (2026) Benchmarking and lightweight correction of daily near-surface air temperature products over the Qinghai-Tibetan Plateau

Identification

• Journal: Atmospheric Research
• Year: 2026
• Date: 2026-01-16
• Authors: Nan Zhou, Yao Xiao, Lin Zhao, Guojie Hu, Cunbao Wang, Li Ren, Guanyue Liu, Defu Zou, Xiangfei Li
• DOI: 10.1016/j.atmosres.2026.108784

Research Groups

• School of Geographical Sciences, Nanjing University of Information Science & Technology, Nanjing 210044, China
• Cryosphere Research Station on the Qinghai-Tibet Plateau, State Key Laboratory of Cryospheric Science and Frozen Soil Engineering, Northwest Institute of Eco- Environment and Resources, Chinese Academy of Sciences, Lanzhou 730030, China

Short Summary

This study benchmarks five daily near-surface air temperature products over the Qinghai-Tibetan Plateau against 20 research-grade stations, revealing pervasive cold biases in reanalysis data and proposing a lightweight two-stage correction method that significantly reduces errors and provides scene-specific usability guidance.

Objective

• To build a daily, lineage-spanning benchmark across 20 research-grade stations to evaluate five representative near-surface air temperature datasets over the Qinghai-Tibetan Plateau under a unified collocation.
• To develop and apply a lightweight two-stage correction method (regime-aware height normalization and quantile mapping) to improve the accuracy of these datasets.
• To provide scene-specific guidance for selecting and minimally correcting near-surface air temperature forcings for the Qinghai-Tibetan Plateau.

Study Configuration

• Spatial Scale: Qinghai-Tibetan Plateau (QTP), stratified by eco-climatic zone (HI/HII), ground-thermal state (permafrost/seasonally frozen), and 500-meter elevation bands. Kilometer-scale products are evaluated.
• Temporal Scale: Daily. Performance stratified by season (winter specifically highlighted).

Methodology and Data

• Models used:
  • Evaluation of five representative datasets: global land reanalysis (e.g., ERA5-Land), observation-conditioned regional fusions (e.g., China Meteorological Forcing Dataset - CMFD), and kilometer-scale observation-based interpolations.
  • Correction methods: Regime-aware height normalization (γ) followed by quantile mapping.
• Data sources:
  • 20 research-grade stations (in-situ observations) for benchmarking.
  • Gridded near-surface air temperature products (e.g., ERA5-Land, CMFD).

Main Results

• Observation-conditioned fields generally outperform purely model-driven reanalysis (e.g., ERA5-Land) at the daily scale.
• ERA5-Land reanalysis exhibits a pervasive cold bias, most pronounced in the HII eco-climatic zone during winter.
• Elevation composites indicate increasing R² and decreasing Mean Absolute Error (MAE) from 3.0 to 4.5 kilometers altitude, with winter performance being the weakest seasonally.
• The lightweight two-stage correction (regime-aware height normalization followed by quantile mapping) reduces MAE by 0.5 to 2.0 °C and effectively removes height-dependent cold bias.
• A unitless usability matrix provides scene-specific guidance (by season, zone, and elevation) for selecting and minimally correcting near-surface air temperature forcings.

Contributions

• Establishes a comprehensive, lineage-spanning benchmark for daily near-surface air temperature products over the Qinghai-Tibetan Plateau using 20 research-grade stations.
• Introduces a novel regime-aware stratification for performance evaluation based on eco-climatic zone, ground-thermal state, and elevation bands.
• Develops and validates a lightweight, two-stage correction method (regime-aware height normalization and quantile mapping) that significantly improves the accuracy of gridded temperature products.
• Creates a unitless usability matrix that translates multi-metric skill into practical, scene-specific guidance for data selection and correction, offering a transferable operational pathway for high-relief regions.

Funding

• Not available in the provided paper text.

Citation

@article{Zhou2026Benchmarking,
  author = {Zhou, Nan and Xiao, Yao and Zhao, Lin and Hu, Guojie and Wang, Lingxiao and Wang, Cunbao and Ren, Li and Liu, Guanyue and Zou, Defu and Li, Xiangfei},
  title = {Benchmarking and lightweight correction of daily near-surface air temperature products over the Qinghai-Tibetan Plateau},
  journal = {Atmospheric Research},
  year = {2026},
  doi = {10.1016/j.atmosres.2026.108784},
  url = {https://doi.org/10.1016/j.atmosres.2026.108784}
}