Scholarlink

<p>Reliable regional-scale temperature projections are essential for climate impact assessment and strategizing remedial measures for future. The complex physiography, which is the characteristics of mainland India, introduces large uncertainties in Global Climate Model (GCM) outputs for temperature analysis over the vast expanse of this geographical entity. This study presents a framework for reducing the uncertainties of GCM outputs by applying a multi-model ensemble (MME) technique using CMIP6 Bias Corrected GCM outputs. Monthly maximum, mean, and minimum temperatures for the period 1965–2014 were evaluated against ERA5 reanalysis data. Five Multi-Model Ensemble (MME) techniques—Equal Weight Averaging, Bates–Granger Averaging, Granger–Ramanathan Averaging (GRA), Mallows Model Averaging, and Bayesian Model Averaging—were systematically assessed using k-fold cross-validation and full-sample calibration at multiple spatial resolutions. Results show that GRA consistently outperforms other techniques by minimizing mean absolute error and maximizing the index of agreement across all temperature variables. The proposed framework enhances the robustness of temperature simulations and provides a strong foundation for future climate projections under different emission scenarios.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhancing the robustness of temperature simulations in India through a bias-corrected multi-model ensemble framework

  • 期刊论文

  • 发表日期：

    2026年6月15日

查看全文
  • Avijit Paul,
  • Monomoy Goswami

摘要

Reliable regional-scale temperature projections are essential for climate impact assessment and strategizing remedial measures for future. The complex physiography, which is the characteristics of mainland India, introduces large uncertainties in Global Climate Model (GCM) outputs for temperature analysis over the vast expanse of this geographical entity. This study presents a framework for reducing the uncertainties of GCM outputs by applying a multi-model ensemble (MME) technique using CMIP6 Bias Corrected GCM outputs. Monthly maximum, mean, and minimum temperatures for the period 1965–2014 were evaluated against ERA5 reanalysis data. Five Multi-Model Ensemble (MME) techniques—Equal Weight Averaging, Bates–Granger Averaging, Granger–Ramanathan Averaging (GRA), Mallows Model Averaging, and Bayesian Model Averaging—were systematically assessed using k-fold cross-validation and full-sample calibration at multiple spatial resolutions. Results show that GRA consistently outperforms other techniques by minimizing mean absolute error and maximizing the index of agreement across all temperature variables. The proposed framework enhances the robustness of temperature simulations and provides a strong foundation for future climate projections under different emission scenarios.