AUTHOR=Huang Weixiang , Luo Xiaoli , Zhang Wei , Feng Bo , Xia Xiaofei , Yi Chenying , Hu Weijun 

TITLE=Improving prediction of short-duration heavy rainfall in Guangxi, China during the pre-summer rainy season based on Fengyun-4A lightning frequency and a machine learning algorithm

JOURNAL=Frontiers in Earth Science

VOLUME=Volume 13 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/earth-science/articles/10.3389/feart.2025.1669417

DOI=10.3389/feart.2025.1669417

ISSN=2296-6463

ABSTRACT=IntroductionThis study investigated the relationships between short-duration heavy rainfall (SDHR) events and lightning activity over Guangxi, China, during the pre-summer rainy season from 2019 to 2023.MethodsThe analysis was conducted using the satellite-retrieved IMERG precipitation dataset and the Fengyun-4A lightning mapping imager (FY-4A/LMI). We employed a Random Forest machine learning model to assess the value of one-hour antecedent lightning frequency as a predictor for SDHR. The model's interpretability was further examined using SHapley Additive exPlanation (SHAP) value analysis.ResultsThe results revealed distinct spatiotemporal variations. SDHR events first occurred in eastern Guangxi in April, expanded westwards in May, and covered the entire region by June. Lightning activity peaked in April, decreased in May, and increased again in June. Both exhibited a unimodal diurnal cycle, peaking at nocturnal-to-morning hours; however, SDHR intensity consistently reached its maximum at 21:00 UTC, approximately one hour later than the lightning peak. The mean number of lightning flashes per SDHR event decreased from 8.58 in April to 6.14 in May and 6.10 in June. Incorporating lightning frequency into the Random Forest model substantially enhanced SDHR prediction accuracy, reducing the mean absolute error by 4.42% (April), 6.02% (May), and 4.29% (June). The coefficient of determination (R2) for SDHR amount increased from 0.29 to 0.35 in April, 0.38 to 0.45 in May, and 0.22 to 0.29 in June.DiscussionSHAP analysis confirmed the positive contribution of lightning frequency to rainfall intensity prediction throughout the entire study period. This positive contribution exhibited a monotonically increasing trend when lightning frequency was below the threshold of 15, and lightning frequency was found to amplify its influence on the model output through interactions with other predictors. Collectively, these results underscore the value of lightning observations as robust predictors for improving short-term heavy rainfall forecasts.