Road Accidents Data
Road accident data is an essential source of information in the battle for road safety. Ideally, the data contain the exact location and relevant data about the conditions during the traffic accident. Also, the moment when the accident took place is essential to create a useful geo-time series.
The road accident dataset typically contains data that is derived from a form that must be filled out by a police officer for each traffic accident with injured or deceased people.
The traffic accident data usually contain information on the different circumstances in which the accidents have occurred: course of the crash (type of collision, type of road users, injury classification e.o), traffic conditions (maximum speed, priority regulation, e.o), environmental conditions (weather, light conditions, time of the accident, e.o), road conditions (road surface, obstacles, e.o), human factors (fatigue, drugs, alcohol, e.o) and geographical conditions (location type, physical characteristics of the road, e.o). Datasets from Belgium, the Czech Republic and the UK learn that the data itself is quite comparable.
The localisation of the accident started before the GPS area. The location was and often is still described in 3 different ways:
- Community, street and house number
- Intersecting streets
- Highway location marker
The modern and accurate method of registration is by providing a coordinate using a coordinate system like WGS.
When the location is provided through an address, intersecting streets or highway location markers, geocoding algorithms are needed. A proper geocoding requires a base map with at least road or building geometry and address information. Mile/kilometre markers can be converted to coordinates by calculating the relevant position along the road geometry based on the markers. Unfortunately, this is harder than geocoding.
To improve the data quality (in Belgium, around 88% of the data can be geocoded), an electronic form with automatic geocoding on the accident spot would improve the data quality substantially.
Accident information is from a human perspective sensitive data. The publication of an accident spot (including exact location and time), including specific data about the injury, drug or alcohol involvement give insight into the circumstances during the accident. Nowadays, it is accepted to make these data publically available.
Accident data has great policy value. By publishing the accident data, community groups have almost the same factual data to discuss with the government about traffic safety in their community and around important public spaces like schools or hospitals. Road accidents are also a subject where policymakers have defined clear long-term goals in reduction of the number of casualties which makes it as an example for other transport and mobility-related policy topics like the modal split or reducing travel times.
When best to use
Main data elements:
Depending on the source, the data elements can differ. However, most datasets provide (based on data from Belgium, the Czech Republic and the UK):
- Location information: address, highway location marker or co-ordinate
- Road information: e.g. classification of the road, nr of lanes
- Road conditions: bad road surface; faulty signals; faulty lighting; road works; queue; downhill; curve; bad visibility
- Weather conditions: normal weather; rain; fog; wind; snow; hail; other weather
- Road conditions: dry road surface; wet road surface; snow on the road surface; clean road surface; dirty road surface
- Light conditions: daylight; twilight; public lighting; night
- Time information: date, time, weekend or weekday
- Types of vehicles involved: car, minibus, light truck
- Direction/movement/dynamics: straight direction, opposite direction, constant speed, brake
- Aggravating circumstances: alcohol, speeding, drugs
- Number of road users
- Number of injured and deceased road users
- Information about road users: sex, age
- Consequences road user: dead road user, seriously injured road user, lightly road user, uninjured road user
- Type of collision: multiple collision, frontal collision