AccScience Publishing / AJWEP / Online First / DOI: 10.36922/AJWEP025500379
Submit to AJWEP
Cite this article
11
Download
47
Views
Journal Browser
Volume | Year
Issue
Search
Forthcoming Issue
Current Issue
View All
News and Announcements
View All
ORIGINAL RESEARCH ARTICLE

Light gradient boosting machine-based early warning of dense fog in the Qiantang River Estuary with Shapley additive explanations interpretation

Nan Fang1 ,  Xiaoni Liang2* ,  Fengxue Qiao3,4 ,  Zhaoming Chen3 ,  Chuhan Lu5 ,  Weicai Zheng1
Show Less
1 Early Warning Center in Zhejiang Province, Hangzhou, Zhejiang, China
2 Meteorological Service Center in Zhejiang Province, Hangzhou, Zhejiang, China
3 Department of Geography, School of Geographic Sciences, East China Normal University, Shanghai, China
4 Key Laboratory of Geographic Information Science, Ministry of Education, East China Normal University, Shanghai, China
5 Department of Atmosphere, School of Atmosphere and Remote Sensing, Wuxi University, Wuxi, Jiangsu, China
AJWEP, 025500379 https://doi.org/10.36922/AJWEP025500379
Received: 8 December 2025 | Revised: 7 February 2026 | Accepted: 24 February 2026 | Published online: 12 May 2026
© 2026 by the Author(s). This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution 4.0 International License ( https://creativecommons.org/licenses/by/4.0/ )
Download PDF
XML
Cite
Abstract

Low-visibility events, particularly dense fog, pose significant risks to navigation and transportation safety in coastal estuarine regions, making accurate and timely early warning systems essential. This study develops a lightweight low-visibility warning model based on the light gradient boosting machine (LightGBM) algorithm, using hourly meteorological observation data from the Qiantang River Estuary region for the period 2021–2024. The forecast lead time was set to 3 h, and the model’s performance was evaluated for predicting both low-visibility events (visibility ≤ 2,000 m) and fog events (visibility ≤ 500 m), with interpretability analysis conducted using Shapley additive explanations (SHAP). The results show that: (i) Validation against actual low-visibility events confirms that the model provided effective warnings across the study area, achieving an average accuracy of 98.6% for low-visibility events. (ii) The original LightGBM model requires parameter optimization to handle imbalanced classification, particularly for rare fog events. By adjusting class weights, the false-negative rate for dense fog was effectively reduced, improving recall from 44% to 70%. (iii) Global SHAP analysis revealed that relative humidity is the meteorological factor contributing most to dense fog warnings. Sample characteristics such as low wind speed, high humidity, and a small air–ground temperature difference consistently contribute to the model’s prediction of dense fog events.

Keywords
Light gradient boosting machine
Visibility
Machine learning
Shapley Additive exPlanations analysis
Funding
This research was funded by Zhejiang Provincial Natural Science Foundation of China (Grant No. LZJMY25D050007).
Conflict of interest
The authors declare they have no competing interests.
References
  1. Fu G, Li X, Wei N. Review on the atmospheric visibility research. Period Ocean Univ China. 2009;39(5):855-862. In Chinese.

 

  1. Guijo-Rubio D, Gutiérrez PA, Casanova-Mateo C, Sanz-Justo J, Salcedo-Sanz S, Hervás-Martínez C. Prediction of low-visibility events due to fog using ordinal classification. Atmos Res. 2018;214:64-73. doi: 10.1016/j.atmosres.2018.07.017

 

  1. Qian WH, Leung JCH, Chen YL, Huang SY. Applying anomaly-based weather analysis to the prediction of low visibility associated with the coastal fog at Ningbo-Zhoushan Port in East China. Adv Atmos Sci. 2019;36:1060-1077. doi: 10.1007/s00376-019-8252-5

 

  1. Zhou G, Xu J, Xie Y, et al. Numerical air quality forecasting over eastern China: An operational application of WRF-Chem. Atmos Environ. 2017;153:94-108. doi: 10.1016/j.atmosenv.2017.01.020

 

  1. Xia F, Li CY. Fog weather forecast experiments over Shandong province based on three visibility schemes. J Meteorol Environ. 2018;34(3):48-57. In Chinese. doi: 10.3969/j.issn.1673-503X.2018.03.006

 

  1. Cheng FY, Feng CY, Yang ZM, et al. Evaluation of real-time PM2.5 forecasts with the WRF-CMAQ modeling system and weather-pattern-dependent bias-adjusted PM2.5 forecasts in Taiwan. Atmos Environ. 2021;244:117909. doi: 10.1016/j.atmosenv.2020.117909

 

  1. Kurt A, Oktay AB. Forecasting air pollutant indicator levels with geographic models 3 days in advance using neural networks. Expert Syst Appl. 2010;37(12):7986-7992. doi: 10.1016/j.eswa.2010.05.093

 

  1. Salcedo-Sanz S, Guijo-Rubio D, Pérez-Aracil J, Peláez- Rodríguez C, Gómez-Orellana AM, Gutiérrez-Peña PA. Artificial Intelligence-Based Methods and Algorithms in Fog and Atmospheric Low-Visibility Forecasting. Atmosphere. 2025;16(9):1073. doi: 10.3390/atmos16091073

 

  1. Wang K, Zhao H, Liu AX, Han B, Bai Z. Development and validation of visibility forecast technique based on the risk neural network. China Environ Sci. 2009;29(10):1029-1033. In Chinese. doi: 10.3321/j.issn:1000-6923.2009.10.005

 

  1. Wang Y. Research on Shanghai Visibility Prediction Model Based on Multi-source Data and XGBoost Algorithm. Dissertation. East China Normal University; 2019.

 

  1. Zhang XT, Liu H, Liang M, Ju F, Gao XX. Prediction and analysis of PM2.5 mass concentration in Fenwei Plain based on different algorithms of machine learning. J Shaanxi Meteorol. 2023;(3):8-16. In Chinese. doi: 10.3969/j.issn.1006-4354.2023.03.002

 

  1. Fang N, Jiang SJ, Yan XM, Ruan SJ, Ma XY. Research on Ultra Short-Term Fast Rolling Prediction Technology of Wind Speed Based on LSTM Neural Network. Meteorol Sci Technol. 2022;50(6):842-850. In Chinese. doi: 10.19517/j.1671-6345.20220064

 

  1. Fang N, Xie GQ, Ruan XJ, Ren CP, Jiang SJ, Zhang WW. Application of long short-term memory neural network (LSTM) model in low visibility forecast. J Meteorol Environ. 2022;38(5):34-41. In Chinese. doi: 10.3969/j.issn.1673-503X.2022.05.004

 

  1. Castillo-Botón C, Casillas-Pérez D, Casanova-Mateo C, et al. Machine learning regression and classification methods for fog events prediction. Atmos Res. 2022;272:106157. doi: 10.1016/j.atmosres.2022.106157

 

  1. Ding YX, Li Z, Zhang CD, Ma J, Cheng JCP, Wan Z. Prediction of ambient PM2.5 concentrations using a correlation filtered spatial-temporal long short-term memory model. Appl Sci. 2019;10(1):14. doi: 10.3390/app10010014

 

  1. Ke G, Meng Q, Finley T, et al. LightGBM: A highly efficient gradient boosting decision tree. In: Proceedings of the 31st International Conference on Neural Information Processing Systems (NIPS 2017); December 4-9, 2017; Long Beach, CA, USA. Neural Information Processing Systems Foundation; 2017:3146-3154. Available from: https://proceedings. neurips.cc/paper_files/paper/2017/file/6449f44a102fde8486 69bdd9eb6b76fa-Paper.pdf [Last accessed on December 8, 2025].

 

  1. Zhang TY, Su H, Yang X, Yan XH. Remote sensing prediction of global subsurface thermohaline and the impact of longitude and latitude based on LightGBM. Nat Remote Sens Bull. 2020;24(10):1255-1269. doi: 10.11834/jrs.20200007

 

  1. Wang LL, Tan JH. Objective forecast research of dense fog in southern Henan based on LightGBM. Arid Meteorol. 2024;42(5):702-709. In Chinese. doi: 10.11755/j.issn.1006-7639-2024-05-0702

 

  1. Wang ZY. The Correction of Atmospheric Visibility Prediction in Shanghai Based on LightGBM Framework. Dissertation. East China Normal University; 2019.

 

  1. Lundberg SM, Lee SI. A unified approach to interpreting model predictions. In: Proceedings of the 31st International Conference on Neural Information Processing Systems (NIPS 2017); December 4-9, 2017; Long Beach, CA, USA. Neural Information Processing Systems Foundation; 2017:4768-4777. Available from: https://proceedings.neurips.cc/paper_files/paper/2017/ file/8a20a8621978632d76c43dfd28b67767-Paper.pdf [Last accessed on December 8, 2025].

 

  1. Hou L, Dai Q, Song C, et al. Revealing drivers of haze pollution by explainable machine learning. Environ Sci Technol Lett. 2022;9(2):112-119. doi: 10.1021/acs.estlett.1c00865

 

  1. Madhushani C, Dananjaya K, Ekanayake IU, Meddage DPP, Kantamaneni K, Rathnayake U. Modeling streamflow in non-gauged watersheds with sparse data considering physiographic, dynamic climate, and anthropogenic factors using explainable soft computing techniques. J Hydrol. 2024;631:130846. doi: 10.1016/j.jhydrol.2024.130846

 

  1. Guo ZZ, He J, Huang D, Zhou YQ, Zhu YH. Fast assessment model for rainfall-induced shallow landslide hazard and application. Chin J Rock Mech Eng. 2023;42(5):1188-1201. In Chinese. doi: 10.13722/j.cnki.jrme.2022.0605

 

  1. Covert I, Lundberg S, Lee SI. Explaining by removing: A unified framework for model explanation. J Mach Learn Res. 2021;22(33):1-90. Available from: https://www.jmlr.org/ papers/volume22/20-1316/20-1316.pdf [Last accessed on].

 

  1. Xiahou J, Xiao A. Importance analysis on warm season rainstorm forecast factors in Jiangxi province based on machine learning model of Shapely values. Meteorol Disaster Reduct Res. 2024;47(1):12-23. In Chinese. doi: 10.12013/qxyjzyj2024-002

 

  1. Dong JQ. Prediction of atmospheric pollutant concentrations and driver factor mining based on interpretable machine learning. Dissertation. North China Electric Power University; 2023.

 

  1. Rathnayake N, Jayasinghe J, Semasinghe R, Rathnayake U. Predicting short-term wind power generation at Musalpetti Wind Farm: Model development and analysis. Comput Model Eng Sci. 2025;143(2):2287-2305. doi: 10.32604/cmes.2025.064464
Share
Back to top
Asian Journal of Water, Environment and Pollution, Electronic ISSN: 1875-8568 Print ISSN: 0972-9860, Published by AccScience Publishing
We use cookies on our website to ensure you get the best experience.
Read more about our cookies here
Accept