Lightning-induced vulnerability assessment in Bangladesh using machine learning and GIS-based approach
This study investigates the lightning-induced vulnerability in Bangladesh using Geographic Information Systems (GIS) and Machine Learning (ML) techniques, addressing the limited research in this area. Lightning, especially prevalent during April to June, is a significant threat in Bangladesh, causin...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-01-01
|
| Series: | Progress in Disaster Science |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2590061725000031 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1823856793731727360 |
|---|---|
| author | Tanmoy Mazumder Md. Mustafa Saroar |
| author_facet | Tanmoy Mazumder Md. Mustafa Saroar |
| author_sort | Tanmoy Mazumder |
| collection | DOAJ |
| description | This study investigates the lightning-induced vulnerability in Bangladesh using Geographic Information Systems (GIS) and Machine Learning (ML) techniques, addressing the limited research in this area. Lightning, especially prevalent during April to June, is a significant threat in Bangladesh, causing fatalities, injuries, and economic losses. By analyzing spatiotemporal patterns of lightning and casualties, and incorporating meteorological, geographical, and socio-economic factors into ML models (Random Forest, Multinomial Logistic Regression, Support Vector Machine, and Artificial Neural Networks), this research provides a nuanced understanding of lightning impacts. Findings indicate a downward trend in lightning strikes but not necessarily in fatalities, revealing the complexity of contributing factors. Northern Bangladesh experiences more lightning strikes, whereas the northeast has higher casualty rates. Correlation analysis indicates that lightning fatalities are influenced by multiple factors, with high correlations to cropland area (0.69), agricultural population (0.61), and lightning flashes (0.45). The Random Forest model has appeared to be the best model to predict [with high accuracy] the influence of lightning vulnerability factors. The most significant predictors of lightning vulnerability are cropland area (32 %) followed by literacy rate (19 %), rural population (18 %), lightning flashes (16 %), and water area (15 %) in Bangladesh. The extensive presence of croplands and rural populations increases exposure to lightning during peak farming seasons, while low literacy rates exacerbate risks by limiting awareness of safety measures. Additionally, large water bodies influence local microclimates and pose risks to those working or travelling in and around these wetlands such as agriculture laborers and fishermen. Changes in lightning flash frequencies due to climate variability, combined with socio-economic disparities and infrastructure deficits, further amplify vulnerabilities. A district-level vulnerability map developed in this study provides actionable insights for geographically/area-based targeted policy interventions to address these interlinked factors driving vulnerability. This comprehensive, data-driven approach marks a significant advancement in our understanding of lightening vulnerability and offrrs valuable insight for strategy developed to combat the fatalities of lightning in Bangladesh. |
| format | Article |
| id | doaj-art-fda441bb746d474087c5781f02ea7c0a |
| institution | Kabale University |
| issn | 2590-0617 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Progress in Disaster Science |
| spelling | doaj-art-fda441bb746d474087c5781f02ea7c0a2025-02-12T05:31:43ZengElsevierProgress in Disaster Science2590-06172025-01-0125100406Lightning-induced vulnerability assessment in Bangladesh using machine learning and GIS-based approachTanmoy Mazumder0Md. Mustafa Saroar1Corresponding author at: Department of Urban and Regional Planning, Khulna University of Engineering & Technology (KUET), Room 404, URP Building, Khulna 9203, Bangladesh.; Department of Urban and Regional Planning, Khulna University of Engineering & Technology (KUET), Khulna 9203, BangladeshDepartment of Urban and Regional Planning, Khulna University of Engineering & Technology (KUET), Khulna 9203, BangladeshThis study investigates the lightning-induced vulnerability in Bangladesh using Geographic Information Systems (GIS) and Machine Learning (ML) techniques, addressing the limited research in this area. Lightning, especially prevalent during April to June, is a significant threat in Bangladesh, causing fatalities, injuries, and economic losses. By analyzing spatiotemporal patterns of lightning and casualties, and incorporating meteorological, geographical, and socio-economic factors into ML models (Random Forest, Multinomial Logistic Regression, Support Vector Machine, and Artificial Neural Networks), this research provides a nuanced understanding of lightning impacts. Findings indicate a downward trend in lightning strikes but not necessarily in fatalities, revealing the complexity of contributing factors. Northern Bangladesh experiences more lightning strikes, whereas the northeast has higher casualty rates. Correlation analysis indicates that lightning fatalities are influenced by multiple factors, with high correlations to cropland area (0.69), agricultural population (0.61), and lightning flashes (0.45). The Random Forest model has appeared to be the best model to predict [with high accuracy] the influence of lightning vulnerability factors. The most significant predictors of lightning vulnerability are cropland area (32 %) followed by literacy rate (19 %), rural population (18 %), lightning flashes (16 %), and water area (15 %) in Bangladesh. The extensive presence of croplands and rural populations increases exposure to lightning during peak farming seasons, while low literacy rates exacerbate risks by limiting awareness of safety measures. Additionally, large water bodies influence local microclimates and pose risks to those working or travelling in and around these wetlands such as agriculture laborers and fishermen. Changes in lightning flash frequencies due to climate variability, combined with socio-economic disparities and infrastructure deficits, further amplify vulnerabilities. A district-level vulnerability map developed in this study provides actionable insights for geographically/area-based targeted policy interventions to address these interlinked factors driving vulnerability. This comprehensive, data-driven approach marks a significant advancement in our understanding of lightening vulnerability and offrrs valuable insight for strategy developed to combat the fatalities of lightning in Bangladesh.http://www.sciencedirect.com/science/article/pii/S2590061725000031Data-driven decision makingLightning vulnerabilityRandom Forest modelThunderstormBangladesh |
| spellingShingle | Tanmoy Mazumder Md. Mustafa Saroar Lightning-induced vulnerability assessment in Bangladesh using machine learning and GIS-based approach Progress in Disaster Science Data-driven decision making Lightning vulnerability Random Forest model Thunderstorm Bangladesh |
| title | Lightning-induced vulnerability assessment in Bangladesh using machine learning and GIS-based approach |
| title_full | Lightning-induced vulnerability assessment in Bangladesh using machine learning and GIS-based approach |
| title_fullStr | Lightning-induced vulnerability assessment in Bangladesh using machine learning and GIS-based approach |
| title_full_unstemmed | Lightning-induced vulnerability assessment in Bangladesh using machine learning and GIS-based approach |
| title_short | Lightning-induced vulnerability assessment in Bangladesh using machine learning and GIS-based approach |
| title_sort | lightning induced vulnerability assessment in bangladesh using machine learning and gis based approach |
| topic | Data-driven decision making Lightning vulnerability Random Forest model Thunderstorm Bangladesh |
| url | http://www.sciencedirect.com/science/article/pii/S2590061725000031 |
| work_keys_str_mv | AT tanmoymazumder lightninginducedvulnerabilityassessmentinbangladeshusingmachinelearningandgisbasedapproach AT mdmustafasaroar lightninginducedvulnerabilityassessmentinbangladeshusingmachinelearningandgisbasedapproach |