Nonlinear Drivers of Extreme Rainfall in Barishal, Bangladesh: Evidence from an Integrated Statistical and Machine Learning Framework

Abstract

In the context of assessing climate risks in vulnerable coastal areas like Barishal, Bangladesh, it is important to understand the determinants of extreme rainfall. Current research mainly focuses on mean rainfall patterns, with limited emphasis on extremes and land–atmosphere interactions. This paper addresses this gap by creating a unified statistical and machine learning framework to identify the key drivers of extreme precipitation. Monthly climatic data for 1984–2025 (N=504) were obtained from the NASA POWER database, including temperature, humidity, wind speed, solar radiation, and soil moisture. Extreme rainfall events were defined using the 90th percentile threshold. Logistic regression and Random Forest models were then employed to identify both linear and nonlinear relationships, while SHAP was used to evaluate the importance of each variable. The results show that temperature (β=3.501, p<0.05) and soil moisture (β=49.788, p<0.001) significantly increase the likelihood of extreme rainfall, whereas solar radiation (β=−1.064, p<0.001) has a negative effect. The Random Forest model outperforms logistic regression (AUC=0.976 vs. 0.958), achieving 85.6% accuracy and a Kappa value of 0.68. SHAP analysis indicates that soil moisture is the most influential predictor, followed by humidity and temperature. Overall, the findings provide a robust and interpretable framework for understanding the occurrence of extreme rainfall, which can support improved climate risk management strategies.

Declarations

Consent to publish: The authors have provided consent to publish.

Competing interests: The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Data availability statement: The data used in this study are available from publicly accessible sources.

Funding: The authors received no financial support for the research, authorship and/or publication of this article.

Clinical Trial Number: Not Applicable.

Declaration of using generative AI: During the preparation of this work the author(s) used ChatGPT in order to correct the grammatical errors. After using this tool/service, the author(s) reviewed and edited the content as needed and take full responsibility for the content of the published article.

Author Contributions: M.O.F.: Conceptualization, methodology, writing—original draft preparation, writing—review and editing. All authors have read and agreed to the published version of the manuscript.