Identification of Traffic Accident Hotspots using Network Kernel Density: Application to Galle District in Sri Lanka

This study aims to identify traffic accident hotspots and establish the correlation between road features and accidents. Network-based Kernel Density Estimators (NKDE) were computed using traffic accident data from 2009 to 2013 for each year for road segments termed lixels. These yearly NKDEs were a...

Full description

Saved in:
Bibliographic Details
Main Authors: T. M. Rengarasu, H. K. I. Arunodi
Format: Article
Language:English
Published: Sri Lanka Society of Logistics & Transport (SLSTL) 2025-03-01
Series:Journal of South Asian Logistics and Transport
Subjects:
Online Access:https://jsalt.sljol.info/articles/92/files/67e63b92654c7.pdf
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850181761116405760
author T. M. Rengarasu
H. K. I. Arunodi
author_facet T. M. Rengarasu
H. K. I. Arunodi
author_sort T. M. Rengarasu
collection DOAJ
description This study aims to identify traffic accident hotspots and establish the correlation between road features and accidents. Network-based Kernel Density Estimators (NKDE) were computed using traffic accident data from 2009 to 2013 for each year for road segments termed lixels. These yearly NKDEs were aggregated to formulate a traffic accident intensity index (AII), serving as the foundation for hotspot identification. A Gamma regression model with a logarithmic link was constructed to relate the AII values of lixels in the A2 highway with macro-level parameters. Population, Junction density (number of junctions in a section), Bendiness, and Urbanisation. The study determined that a cell size of 250 m with a 500 m bandwidth is optimal for NKDE calculations for the road network of the Galle district. Lixels having extreme AII values were defined as the traffic accident hotspots and subdivided into yellow, orange, and red zones. Additionally, the Gamma regression model highlighted significant correlations between road attributes and accident intensity, indicating a 131% increase in AII values in urban areas compared to rural ones. Areas with junctions and bends had a decreasing effect on the AII values. With the increase of a junction, AII decreased significantly by 29 %, and if a bend is present in the lixel, AII decreases significantly by 12%.
format Article
id doaj-art-d5505844a9cd4cd2b8daa2ae992c5bb6
institution OA Journals
issn 2783-8897
2783-8676
language English
publishDate 2025-03-01
publisher Sri Lanka Society of Logistics & Transport (SLSTL)
record_format Article
series Journal of South Asian Logistics and Transport
spelling doaj-art-d5505844a9cd4cd2b8daa2ae992c5bb62025-08-20T02:17:49ZengSri Lanka Society of Logistics & Transport (SLSTL)Journal of South Asian Logistics and Transport2783-88972783-86762025-03-015112310.4038/jsalt.v5i1.92Identification of Traffic Accident Hotspots using Network Kernel Density: Application to Galle District in Sri LankaT. M. Rengarasu0H. K. I. Arunodi1Department of Civil & Environmental Engineering, University of Ruhuna, Sri LankaDepartment of Civil & Environmental Engineering, University of Ruhuna, Sri LankaThis study aims to identify traffic accident hotspots and establish the correlation between road features and accidents. Network-based Kernel Density Estimators (NKDE) were computed using traffic accident data from 2009 to 2013 for each year for road segments termed lixels. These yearly NKDEs were aggregated to formulate a traffic accident intensity index (AII), serving as the foundation for hotspot identification. A Gamma regression model with a logarithmic link was constructed to relate the AII values of lixels in the A2 highway with macro-level parameters. Population, Junction density (number of junctions in a section), Bendiness, and Urbanisation. The study determined that a cell size of 250 m with a 500 m bandwidth is optimal for NKDE calculations for the road network of the Galle district. Lixels having extreme AII values were defined as the traffic accident hotspots and subdivided into yellow, orange, and red zones. Additionally, the Gamma regression model highlighted significant correlations between road attributes and accident intensity, indicating a 131% increase in AII values in urban areas compared to rural ones. Areas with junctions and bends had a decreasing effect on the AII values. With the increase of a junction, AII decreased significantly by 29 %, and if a bend is present in the lixel, AII decreases significantly by 12%.https://jsalt.sljol.info/articles/92/files/67e63b92654c7.pdftraffic accident hotspotsgisnetwork kernel density estimationgamma regression model
spellingShingle T. M. Rengarasu
H. K. I. Arunodi
Identification of Traffic Accident Hotspots using Network Kernel Density: Application to Galle District in Sri Lanka
Journal of South Asian Logistics and Transport
traffic accident hotspots
gis
network kernel density estimation
gamma regression model
title Identification of Traffic Accident Hotspots using Network Kernel Density: Application to Galle District in Sri Lanka
title_full Identification of Traffic Accident Hotspots using Network Kernel Density: Application to Galle District in Sri Lanka
title_fullStr Identification of Traffic Accident Hotspots using Network Kernel Density: Application to Galle District in Sri Lanka
title_full_unstemmed Identification of Traffic Accident Hotspots using Network Kernel Density: Application to Galle District in Sri Lanka
title_short Identification of Traffic Accident Hotspots using Network Kernel Density: Application to Galle District in Sri Lanka
title_sort identification of traffic accident hotspots using network kernel density application to galle district in sri lanka
topic traffic accident hotspots
gis
network kernel density estimation
gamma regression model
url https://jsalt.sljol.info/articles/92/files/67e63b92654c7.pdf
work_keys_str_mv AT tmrengarasu identificationoftrafficaccidenthotspotsusingnetworkkerneldensityapplicationtogalledistrictinsrilanka
AT hkiarunodi identificationoftrafficaccidenthotspotsusingnetworkkerneldensityapplicationtogalledistrictinsrilanka