Scholarlink

<p>This study employs the Micro-Genetic Algorithm (<i>μ</i>-GA) to identify the optimal physical parameterization configuration for RegCM5 with the new MOLOCH dynamical core over Vietnam. Three experimental setups at 10-km resolution, driven by ERA5 reanalysis data, are designed. EXP-1 utilizes <i>μ</i>-GA to explore various combinations of convective (CP), microphysics (MP), and planetary boundary layer (PBL) schemes for summer 2000. After 50 generations, <i>μ</i>-GA converges at the 9th generation, identifying an optimal configuration (GA-best): Tiedtke for land CP, Emanuel for ocean CP, Holtslag for PBL, and SUBEX for MP. EXP-2 extends simulations to 1995-2004, comparing GA-best with two widely used configurations, lToT (Tiedtke for both land and ocean) and lEoE (Emanuel for both land and ocean). GA-best outperforms both, though its results are relatively similar to lToT. Given Tiedtke's strong performance over land, EXP-3 applies <i>μ</i>-GA for further parameter tuning within this scheme, again for summer 2000. The algorithm converges after 33 generations, avoiding the need for exhaustive testing of millions of cases, which would exceed most computing capabilities. These results highlight <i>μ</i>-GA's efficiency in optimizing model physical configurations, representing a crucial first step toward producing more reliable climate change projections for Vietnam.</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

Optimizing Configurations for the Regional Climate Model (RegCM5) Using a Micro-GA Approach: A Case Study over Vietnam

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年1月1日

查看全文
  • Thanh Nguyen-Xuan,
  • Zixuan Zhou,
  • Tung Nguyen-Duy,
  • Dzung Nguyen-Le,
  • Thanh Ngo-Duc

摘要

This study employs the Micro-Genetic Algorithm (μ-GA) to identify the optimal physical parameterization configuration for RegCM5 with the new MOLOCH dynamical core over Vietnam. Three experimental setups at 10-km resolution, driven by ERA5 reanalysis data, are designed. EXP-1 utilizes μ-GA to explore various combinations of convective (CP), microphysics (MP), and planetary boundary layer (PBL) schemes for summer 2000. After 50 generations, μ-GA converges at the 9th generation, identifying an optimal configuration (GA-best): Tiedtke for land CP, Emanuel for ocean CP, Holtslag for PBL, and SUBEX for MP. EXP-2 extends simulations to 1995-2004, comparing GA-best with two widely used configurations, lToT (Tiedtke for both land and ocean) and lEoE (Emanuel for both land and ocean). GA-best outperforms both, though its results are relatively similar to lToT. Given Tiedtke's strong performance over land, EXP-3 applies μ-GA for further parameter tuning within this scheme, again for summer 2000. The algorithm converges after 33 generations, avoiding the need for exhaustive testing of millions of cases, which would exceed most computing capabilities. These results highlight μ-GA's efficiency in optimizing model physical configurations, representing a crucial first step toward producing more reliable climate change projections for Vietnam.