Scholarlink

<p>General Circulation Models (GCMs) are essential tools for assessing climate change, but it is critical to evaluate their performance to identify suitable models for regional-scale hydrological applications. This study evaluates 34 CMIP6 GCMs from the NEX-GDDP archive to select climate models for climate change impact assessment for the state of Kerala, India, by examining monthly precipitation data across the 86 grids that cover the state. The climate models are analysed for their macro-scale performance using a network analysis approach and for their micro-scale performance using a statistical index-based ranking approach. The macro-scale approach focuses on spatial synchrony of climate variables to look at the performance of the climate models across the spatial domain using network-derived indicators. The micro-scale approach focuses on the temporal performance of the climate models across the different grids. The present study generates a combined ranking based on both the schemes. A pool of 16 GCMs was identified by combining the top 10 models from the macro-scale and micro-scale selection approaches, with ACCESS-ESM1-5, MPI-ESM1-2-LR, CMCC-CM2-SR5 and TaiESM1 consistently ranking top across both methods. To remove interdependent models from the ensemble, interdependency measures like transfer entropy, mutual information and Pearson’s correlation were employed on area-weighted average precipitation data from the climate models. A maximum relevant, minimum redundant ensemble was identified for Kerala, containing the models: MIROC-6, INM-CM4-08, EC-Earth3-Veg-LR and HadGEM3-GC31-MM. Historical simulations of rainfall from the selected models show correlation greater than 0.6 with historical observed rainfall over most of the grids in Kerala, with lower correlations towards the Western Ghats region in Kerala. The statistical properties of the observed rainfall like monsoon mean, coefficient of variation, 95<sup>th</sup> percentile rainfall and highest consecutive dry days are also simulated well in these selected models.</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

Integrating statistical metrics and spatial network analysis for a multi-scale climate model evaluation

  • 期刊论文

  • 发表日期：

    2025年7月18日

查看全文
  • T Anand,
  • K P Indulekha,
  • S Soumya,
  • Jose George

摘要

General Circulation Models (GCMs) are essential tools for assessing climate change, but it is critical to evaluate their performance to identify suitable models for regional-scale hydrological applications. This study evaluates 34 CMIP6 GCMs from the NEX-GDDP archive to select climate models for climate change impact assessment for the state of Kerala, India, by examining monthly precipitation data across the 86 grids that cover the state. The climate models are analysed for their macro-scale performance using a network analysis approach and for their micro-scale performance using a statistical index-based ranking approach. The macro-scale approach focuses on spatial synchrony of climate variables to look at the performance of the climate models across the spatial domain using network-derived indicators. The micro-scale approach focuses on the temporal performance of the climate models across the different grids. The present study generates a combined ranking based on both the schemes. A pool of 16 GCMs was identified by combining the top 10 models from the macro-scale and micro-scale selection approaches, with ACCESS-ESM1-5, MPI-ESM1-2-LR, CMCC-CM2-SR5 and TaiESM1 consistently ranking top across both methods. To remove interdependent models from the ensemble, interdependency measures like transfer entropy, mutual information and Pearson’s correlation were employed on area-weighted average precipitation data from the climate models. A maximum relevant, minimum redundant ensemble was identified for Kerala, containing the models: MIROC-6, INM-CM4-08, EC-Earth3-Veg-LR and HadGEM3-GC31-MM. Historical simulations of rainfall from the selected models show correlation greater than 0.6 with historical observed rainfall over most of the grids in Kerala, with lower correlations towards the Western Ghats region in Kerala. The statistical properties of the observed rainfall like monsoon mean, coefficient of variation, 95th percentile rainfall and highest consecutive dry days are also simulated well in these selected models.