Scholarlink

<p>Thunderstorms provide essential rain for agriculture and other economic activities in the tropics. Therefore, understanding the dynamics of thunderstorms and potential climate change impacts is crucial for tropical countries like Ghana. This study aimed to identify the factors influencing monthly thunderstorm frequency from 1990 to 2013 across 22 synoptic stations in Ghana using a machine learning approach. Additionally, trend analyses were conducted on thunderstorm frequency and its identified drivers to assess consistency. Using climate indices and ERA5 reanalysis data as features, we developed a LightGBM regression model to forecast monthly thunderstorm frequencies. Recursive feature elimination was employed to select the most significant features from over 60 candidates. The model recorded the following metrics: (R²=0.79; MAE = 2.26; model bias = 0.24). SHapley Additive exPlanations (SHAP) plots indicated that convective available potential energy (CAPE) was the main driver of thunderstorm variability, followed by incident solar radiation at the top of the atmosphere (TISR), convective inhibition (CIN), 10&#xa0;m wind speed (SI10), and longitude. The number of thunderstorms in Ghana declined significantly from 1990 to 2013. This trend aligns with a notable decrease in CAPE and an increase in CIN. The declining CAPE and rising CIN indicate increasing stability consistent with fewer thunderstorms. These trends pose serious implications for agriculture and the hydropower sector in Ghana. It is recommended that the machine learning model be used to develop early warning systems for thunderstorms.</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

Thunderstorm drivers and trends in Ghana, West Africa: an interpretable machine learning study

  • 期刊论文

  • 发表日期：

    2026年3月24日

查看全文
  • Robert A. Akum,
  • Gunnar Lischeid,
  • Philip G. Oguntunde,
  • Richard A. Balogun

摘要

Thunderstorms provide essential rain for agriculture and other economic activities in the tropics. Therefore, understanding the dynamics of thunderstorms and potential climate change impacts is crucial for tropical countries like Ghana. This study aimed to identify the factors influencing monthly thunderstorm frequency from 1990 to 2013 across 22 synoptic stations in Ghana using a machine learning approach. Additionally, trend analyses were conducted on thunderstorm frequency and its identified drivers to assess consistency. Using climate indices and ERA5 reanalysis data as features, we developed a LightGBM regression model to forecast monthly thunderstorm frequencies. Recursive feature elimination was employed to select the most significant features from over 60 candidates. The model recorded the following metrics: (R²=0.79; MAE = 2.26; model bias = 0.24). SHapley Additive exPlanations (SHAP) plots indicated that convective available potential energy (CAPE) was the main driver of thunderstorm variability, followed by incident solar radiation at the top of the atmosphere (TISR), convective inhibition (CIN), 10 m wind speed (SI10), and longitude. The number of thunderstorms in Ghana declined significantly from 1990 to 2013. This trend aligns with a notable decrease in CAPE and an increase in CIN. The declining CAPE and rising CIN indicate increasing stability consistent with fewer thunderstorms. These trends pose serious implications for agriculture and the hydropower sector in Ghana. It is recommended that the machine learning model be used to develop early warning systems for thunderstorms.