Spatial database creation and spatial analysis of road traffic crashes in the Ashanti region, Ghana

Abstract

This study addresses the limitations of Ghana's road traffic crash database, which lacks geographical coordinates, making it difficult to conduct spatial analyses for effective road safety interventions. The research focuses on creating a spatial database and performing spatiotemporal and statistical analyses of road traffic crashes along the N6 and N10 highways in the Ashanti Region from 2018 to 2020. A combination of primary and secondary data sources was used, with crash location data geocoded using postmile linear referencing. Validation of geocoded data was conducted using ground truth points, and positional accuracy was assessed through Root Mean Square Error (RMSE) calculations. Various spatial statistical methods, including Kernel Density Estimation and Hotspot Analysis using Getis-Ord Gi*, were applied to identify crash clusters and high-risk locations. The results indicate that geocoded crash points were accurate, with RMSE values of ±15 m for northings and ± 14 m for eastings, aligning with the recommended 100m buffer for crash location accuracy. The findings highlight critical crash hotspots, particularly at major intersections, emphasizing the need for targeted road safety interventions. The study demonstrates the potential of integrating geospatial data with traditional crash databases to improve road safety planning in Ghana.