Scholarlink

<p>This research presents an innovative method for producing radar rainfall mosaics in the Chao Phraya River Basin, Thailand, using the integration of data from various C-band weather radars operated by the Thai Meteorological Department (TMD). The approach employs open-source Python tools and integrates a novel quality index (QI) that merges cumulative beam blockage (CBB) and pulse volume to mitigate terrain-induced inaccuracies in radar calculations. A thorough 20-fold cross-validation guaranteed robustness under diverse rainfall situations, evaluating five separate bias correction methods utilizing ground truth data from the Hydro-Informatics Institute (HII) precipitation gauge network. The optimized random forest (RF) regressor exhibited exceptional performance, improving radar-gauge correlation at 111 of 169 HII stations, with a success rate of 65.74% and a median enhancement of 35.34%. Findings indicated a consistent underestimation in highland locations (bias ratios &lt; 0.8) and an overestimation in lowland areas (bias ratios &gt; 1.2). Although the RF model sustains bias ratios between 0.85 and 1.15 across all terrains, there remains a persistent underestimating in northern basins, indicating potential for further enhancement. This study’s open-access source code enables the research community and relevant stakeholders to modify the methodology for monitoring severe atmospheric disasters, storms, and hydrological hazards. This capability offers a substantial opportunity for TMD and other agencies to improve disaster preparedness and water resource management in Thailand.</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

Quality Index-Driven Radar Mosaicking and Machine Learning for Enhanced Rainfall Estimation in the Chao Phraya River Basin and Its Tributaries in Thailand

  • 期刊论文

  • 发表日期：

    2025年2月10日

查看全文
  • Nattapon Mahavik,
  • Fatah Masthawee,
  • Sarawut Arthayakun

摘要

This research presents an innovative method for producing radar rainfall mosaics in the Chao Phraya River Basin, Thailand, using the integration of data from various C-band weather radars operated by the Thai Meteorological Department (TMD). The approach employs open-source Python tools and integrates a novel quality index (QI) that merges cumulative beam blockage (CBB) and pulse volume to mitigate terrain-induced inaccuracies in radar calculations. A thorough 20-fold cross-validation guaranteed robustness under diverse rainfall situations, evaluating five separate bias correction methods utilizing ground truth data from the Hydro-Informatics Institute (HII) precipitation gauge network. The optimized random forest (RF) regressor exhibited exceptional performance, improving radar-gauge correlation at 111 of 169 HII stations, with a success rate of 65.74% and a median enhancement of 35.34%. Findings indicated a consistent underestimation in highland locations (bias ratios < 0.8) and an overestimation in lowland areas (bias ratios > 1.2). Although the RF model sustains bias ratios between 0.85 and 1.15 across all terrains, there remains a persistent underestimating in northern basins, indicating potential for further enhancement. This study’s open-access source code enables the research community and relevant stakeholders to modify the methodology for monitoring severe atmospheric disasters, storms, and hydrological hazards. This capability offers a substantial opportunity for TMD and other agencies to improve disaster preparedness and water resource management in Thailand.