Yousaf et al. (2025) A novel Bi-weight Mid Correlation Coefficient Divergence (BMCCD) approach for multi-model ensemble-based drought assessment

Identification

• Journal: Environmental Monitoring and Assessment
• Year: 2025
• Date: 2025-09-11
• Authors: Mahrukh Yousaf, Laiba Shafique, Sadia Qamar, Muhammad Shakeel, Farman Ali, Zulfiqar Ali
• DOI: 10.1007/s10661-025-14578-2

Research Groups

• College of Statistical Sciences, University of the Punjab, Lahore, Pakistan
• Department of Statistics, University of Sargodha, Sargodha, Pakistan
• College of Chemistry and Environmental Engineering, Water Science and Environmental Research Centre, Shenzhen University, Shenzhen, China

Short Summary

This study introduces a novel Bi-weight Mid Correlation Coefficient Divergence (BMCCD) weighting scheme for multi-model ensemble (MME) drought assessment, demonstrating superior performance in correlation and error reduction compared to existing methods, and utilizes it to project future drought characteristics on the Tibetan Plateau.

Objective

• To introduce a novel Bi-weight Mid Correlation Coefficient Divergence (BMCCD) weighting scheme for multi-model ensemble (MME) drought assessment, aiming to improve efficiency and reliability over existing methods.

Study Configuration

• Spatial Scale: 32 distinct locations on the Tibetan Plateau, located between 25° and 40° north latitude and 74° and 104° east longitude, covering approximately 2.5 million square kilometers. Reference data had a spatial resolution of 0.5° × 0.5°.
• Temporal Scale: Historical data from 1961 to 2014, and future projections from 2015 to 2100. Drought characteristics were analyzed across seven time scales (1, 3, 6, 9, 12, 24, and 48 months) under three Shared Socio-economic Pathways (SSP1-2.6, SSP2-4.5, and SSP5-8.5).

Methodology and Data

• Models used: 18 CMIP6 Global Climate Models (GCMs).
• Data sources:
  • Simulated monthly precipitation data from CMIP6 GCMs, obtained from https://cds.climate.copernicus.eu/, initially in kilograms per square meter per second (kg m⁻² s⁻¹).
  • Referenced monthly precipitation data from the CN05.1 model, with a spatial resolution of 0.5° × 0.5°, expressed in millimeters per month (mm/month).
  • Methods: Bi-weight Mid Correlation Coefficient, Point-to-Point Divergence (PPD), Linear Regression, K-Component Gaussian Mixture Model (K-CGMM), Univariate models, and Markov chain for Steady-State Probabilities (SSPs).
  • Software/Tools: R-language (specifically the "WGCNA" library for bicor function), bilinear interpolation for spatial alignment, and unit conversion.

Main Results

• The BMCCD approach achieved the highest average correlation value of 0.749 with the referenced data and the lowest mean absolute error (MAE) value of 1.332, outperforming Simple Model Averaging (SMA) and Weighted Ensemble (WE) methods.
• BMCCD demonstrated better consistency with a minimum correlation of 0.898 and a median correlation of 0.8327.
• The ACCESS-ESM1-5 model was identified as the top-performing GCM in representing observed precipitation data, while CanESM5 showed weaker performance.
• The K-Component Gaussian Mixture Model (K-CGMM) consistently showed lower Bayesian Information Criteria (BIC) values compared to univariate distributions (e.g., triangular, skewed-normal) across all SSP scenarios and time scales, indicating a better fit for standardizing drought indices.
• Long-term projections (2015-2100) for the Tibetan Plateau under all SSP scenarios indicate a predominant "no drought" (ND) condition, with low probabilities for extreme drought (ED) and extreme wet (EW) events.

Contributions

• Introduces a novel Bi-weight Mid Correlation Coefficient Divergence (BMCCD) weighting scheme that robustly combines GCMs by accounting for non-linear relationships and minimizing outlier influence, outperforming traditional SMA and recent WE methods in accuracy and reliability.
• Develops the Standardized Bi-weight Divergence Index (SBDI) based on BMCCD-aggregated data, providing a new and effective tool for drought analysis and projection.
• Integrates K-Component Gaussian Mixture Models (K-CGMM) for drought index standardization, offering a more reliable and efficient approach than unimodal distributions for capturing the complex, multimodal nature of climate extremes.
• Provides long-term, spatially explicit drought projections for the Tibetan Plateau under multiple SSP scenarios, offering actionable insights for policymakers in a vulnerable high-altitude region to mitigate future risks from low-probability, high-impact events.

Funding

Not explicitly mentioned in the provided text.

Citation

@article{Yousaf2025novel,
  author = {Yousaf, Mahrukh and Shafique, Laiba and Qamar, Sadia and Shakeel, Muhammad and Ali, Farman and Ali, Zulfiqar},
  title = {A novel Bi-weight Mid Correlation Coefficient Divergence (BMCCD) approach for multi-model ensemble-based drought assessment},
  journal = {Environmental Monitoring and Assessment},
  year = {2025},
  doi = {10.1007/s10661-025-14578-2},
  url = {https://doi.org/10.1007/s10661-025-14578-2}
}