Chen et al. (2025) Bias correction of subseasonal to seasonal precipitation forecasts over the Tibetan Plateau based on CMA climate prediction models

Identification

• Journal: Atmospheric Research
• Year: 2025
• Date: 2025-11-29
• Authors: Xinyu Chen, Minhong Song, Ziqiang Zhou, Yaqi Wang, Tongwen Wu, Zhiqiang Lin
• DOI: 10.1016/j.atmosres.2025.108667

Research Groups

• Climate Change and Resource Utilization in Complex Terrain Regions Key Laboratory of Sichuan Province, School of Atmospheric Sciences, Chengdu University of Information Technology, Chengdu, China
• Earth System Numerical Prediction Center, China Meteorological Administration, Beijing, China

Short Summary

This study evaluates the effectiveness of CDF and KEM bias correction methods on subseasonal to seasonal precipitation forecasts over the Tibetan Plateau using CMA hindcast data, finding that CDF excels in systematic bias reduction while KEM improves spatial correlation and anomaly trends, with a combined approach offering synergistic benefits.

Objective

• To systematically evaluate and compare the effectiveness of non-parametric percentile mapping (CDF) and Kalman filter-type adaptive (KEM) bias correction methods in reducing precipitation prediction biases from the China Meteorological Administration (CMA) operational climate prediction model over the Tibetan Plateau during summer, particularly in July.

Study Configuration

• Spatial Scale: Tibetan Plateau
• Temporal Scale: Subseasonal to seasonal (summer, particularly July), hindcast period from 2006 to 2020

Methodology and Data

• Models used:
  • Operational climate prediction model of the China Meteorological Administration (CMA)
  • Non-parametric percentile mapping (CDF) bias correction method
  • Kalman filter-type adaptive (KEM) bias correction method
• Data sources:
  • Hindcast data from the operational climate prediction model of the China Meteorological Administration (CMA)

Main Results

• The CDF method is more effective in correcting systematic biases, while the KEM method significantly improves spatial correlation and anomalous trends.
• On a seasonal scale, the CDF method reduced overall systematic precipitation bias in the plateau, decreasing bias by 80 % in the south, with spatial correlations between 0.72 and 0.79.
• The KEM method primarily reduced precipitation bias in the central plateau by 60 % to 80 %, achieving spatial correlations above 0.8 in six years.
• On a subseasonal scale, both methods showed similar effects on bias and spatial correlation as on the seasonal scale, with a modest improvement in precipitation event discrimination (0.01 increase in AROC).
• The KEM method enhanced prediction capability for precipitation anomaly trends, increasing the overall PS score by 6.34 to 84.59.
• For varying precipitation thresholds in July, the KEM method performed optimally for moderate to heavy precipitation, aligning more closely with observations.
• CDF and KEM are complementary: CDF removes pointwise systematic bias and standardizes amplitude distributions, while KEM applied subsequently to CDF-corrected fields restores spatial-phase coherence, refines anomaly trend spatial structure, and suppresses low-end outliers.
• A "CDF first, then KEM" sequence, integrated via performance-driven weighted fusion, leverages their strengths for targeted improvement of precipitation anomaly-trend correction.

Contributions

• Provides a systematic evaluation and comparison of two distinct bias correction methods (CDF and KEM) for subseasonal to seasonal precipitation forecasts over the complex terrain of the Tibetan Plateau.
• Quantifies the specific strengths of each method: CDF for systematic bias reduction and amplitude standardization, and KEM for improving spatial correlation and anomaly trend prediction.
• Proposes a novel, functionally complementary combined approach ("CDF first, then KEM" with weighted fusion) that leverages the strengths of both methods without requiring additional observations, offering a more targeted improvement for precipitation anomaly-trend correction across sub-seasonal to seasonal scales.

Funding

No funding information was provided in the text.

Citation

@article{Chen2025Bias,
  author = {Chen, Xinyu and Song, Minhong and Zhou, Ziqiang and Wang, Yaqi and Wu, Tongwen and Lin, Zhiqiang},
  title = {Bias correction of subseasonal to seasonal precipitation forecasts over the Tibetan Plateau based on CMA climate prediction models},
  journal = {Atmospheric Research},
  year = {2025},
  doi = {10.1016/j.atmosres.2025.108667},
  url = {https://doi.org/10.1016/j.atmosres.2025.108667}
}