Pedestrian Injury Severity Modelling with Additional Secondary Data Extracted using Satellite Images

Pedestrian-vehicle crashes are a leading cause of road traffic crash death all over the world. Since the factors affecting the crashes depends on many factors such as road infrastructure and environmental settings, determining area-specific characteristics which influence the crashes is important. This study focuses on pedestrian crash severity modelling in Chennai, India. We develop an ordered probit model to identify the factors affecting different levels of injury severities for the years 2017- 2019. The severity levels were divided into three categories, simple injury, grievous injury, and fatality. The primary data was taken from Road Accident Database Management System and additional data about location-specific information such as intersection presence, type of intersection and bus stop presence were extracted manually using Google Earth Pro. The results showed that environmental settings (urban or rural) significantly impact pedestrian-vehicle crashes with higher injury severities in rural areas. Time of the day, presence of bus stops within 250ft of crashes and intersection-related crashes also have significant impacts on the severity. Gender related factors were explored. Further interpretation of results was done by computing the marginal effects. Site specific treatments and need for advance pedestrian safety measures are highlighted after analysing the modelling results.