The influence of weather on road safety

Weather conditions affect both crash rate and the exposure to traffic hazards. This influence is strongest for the conditions of precipitation (including snow and hail), fog, low sun, wind, ice forming, and hot temperatures.

Precipitation

Research has shown that motorists adjust their road behaviour during showers. They overtake less, driver slower, and increase their following distance (Hogema, 1996; Agarwal et al., 2005). However, the risk of a crash during rain is still greater than in dry weather. The changes in driving behaviour are, apparently, insufficient to compensate for the greater risk during bad weather (Thoma, 1993).

Road users can have problems with reduced visibility during periods of precipitation. This can be reduced to approximately 50 meters during heavy rain or snow. Splashing water, particularly from lorries, can interfere considerably with the visibility of other motor vehicle drivers (Terpstra, 1995). Clouded windows and windscreens as a result of high humidity during rain can also reduce visibility (Fokkema, 1987). Furthermore, blinding can occur at night because the headlights of oncoming vehicles reflect in the water on the road surface (Ellinghaus, 1983).

The more rain, snow, or hail falls, the less the friction of the road surface. Rain can lead to dynamic aquaplaning. A layer of water on the road surface can cause the vehicle to lose contact with the road surface and to skid. The chance of aquaplaning depends on the skidding resistance of the road, but of course also on the vehicle's speed and tyre tread depths (Ellinghaus, 1983; Terpstra, 1995; Van Ganse, 1981). When it has been dry for a long time, a drizzle can lead to viscous aquaplaning if drops of oil and dust, together with water, produce a thin liquid film on the road surface. When the rain gets heavier, the chance of viscous aquaplaning lessens because the road surface is swept clean (Terpstra, 1995; Eisenberg, 2003).

Fog

In a fog the droplets of water are so small and light that they remain floating in the air. This leads to a reduction in visibility because the light is diffused by the fog droplets. In general, fog occurs when the humidity is 100%. When this happens people generally drive somewhat slower, but simultaneously keep a shorter following distance to the vehicle in front of them. In combination with the decreased field of vision, this increases the risk of crashes (Fokkema, 1987; Oppe, 1988). Fog can also cause viscous aquaplaning when water droplets provide a thin film on the road surface (Terpstra, 1995).

Low sun

Sunrise and sunset can greatly hinder the view that road users have of other traffic. The sun blinds the most when it is low on the horizon. This is the case until about an hour after sunrise and from about an hour before sunset. Motorists can still look through their windscreen, but cannot see clearly anymore. Also indirect sunlight which is reflected by, for example, a glass building, noise barriers, or other cars can be problematic. When lighted by the sun, dirt on the windscreen is more visible, thus hindering driving. Visibility is affected even more when the road surface is wet and reflects the sunlight (Fokkema, 1987).

Wind

Gusts of wind can push relatively high vehicles such as busses, delivery vans, camper vans, caravans, and lorries off course and, under extreme conditions, can even cause them to roll over. This happens mainly on bridges and viaducts. Objects carried by the wind, fallen trees, and broken-off branches can also cause traffic disturbance (Ellinghaus, 1983). Pedestrians and two-wheelers can be troubled by strong gusts of wind and therefore disturb other traffic.

Ice forming

If a road surface has an open structure, such as porous asphalt, wet parts of the road surface will freeze quicker than surfaces with a closed structure. When there is black ice, a thin layer of ice forms so quickly on porous asphalt that the asphalt loses its friction (CROW, 2000). Roads that have just been laid also have a greater risk of slipperiness: the layer of black bitumen has a lower temperature and is thus more sensitive to wet parts freezing. In due time the risk of slipperiness will lessen because of the wear and tear of the upper layer (CROW, 2006).

Temperature

High temperatures especially have a psychological and/or physiological effect on a driver. However, a lot less is found in the literature about the physical effects of weather conditions. According to a German study, emotions rise with the temperature, people are more irritable to others, they get tired, lose their concentration, and their reaction time increases (DVR, 2000). French researchers found an increase in the number of crashes during heat waves. Their explanation was that people possibly drive at other times of the day and that they sleep shorter or less deeply because of the high night-time temperatures. This results in them being more tired when they take to the road (Laaidi & Laaidi, 2002).

During the last decades, various studies have been made of the road safety effects of weather conditions (Keay & Simmonds, 2005; Stiers, 2005; Bos, 2001; Brijs et al., 2007; Sabir, 2011)). In addition to data from meteorological offices, the police road crash registration is a useful data source because weather conditions is one of its variables. To be able to make a good estimation of weather's influence on road crashes, the average weather should be used as point of departure. People more or less prepare themselves for expected weather, and it is usually extreme weather conditions or sudden changes in the weather that produce fluctuations in the numbers of crashes (Bos, 2001; Eisenberg, 2003).

Crash data

AVV crash data shows that in the 2006-2009 period there was an annual average of 59 fatal crashes when it rained, 4 during snow, and 6 during fog. To find out if these numbers are relatively large, we have to relate them to the numbers of hours that it had been raining, snowing, or was foggy. This was also looked for in the literature study. Bos (2001) concluded that, during the 1997-2000 period, there were more road casualties than usual in a relatively warm and dry summer. A cold and dry winter resulted in fewer casualties, but there were more in a cold and wet winter. No unambiguous conclusions could be made about spring and autumn.

A recent study of the relation between weather and road crashes on Dutch national state roads showed that there was an increase in the number of crashes of between 35% and 182% when it rained. Ice forming on the road surface even led to an increase of between 77% and 245%. However, ice forming is far less frequent than rain, and thus has a smaller impact on the total number of crashes (Stiers, 2005).

A significant relation between rainfall and crashes was also found in a study into the effect of weather conditions on road crashes in three cities in the Netherlands (Brijs et al., 2007). However, the researchers did not find significant relations for air pressure, wind, minimum visibility, and rainfall after a long period of drought.

Recent research into the effect of different types of weather on crashes on crashes on Dutch roads during the period 2000-2009 (Sabir, 2011) indicates that precipitation leads to fewer crashes. Sabir also found that the number of crashes and the injury severity increase as the temperature rises.  According to the author, the most plausible explanation is that more cyclists and pedestrians participate in traffic during nice weather. He finally concludes that the number of crashes also increases for snow, cloudy weather and fog. The injury severity, however, decreases when it snows.

A Scottish study of the effects of weather on road crashes in Glasgow reported a 20% increase in crashes on rainy days. The largest effect of rain was found during the summer and autumn months (Smith, 1982).

An Australian study carried out in Melbourne and its surroundings found a 2.4% increase in the average number of injury crashes during rainfall. At night the number of injury crashes during rainfall even increased with 5.2% (Keay & Simmonds, 2005).

A British study (Edwards, 1998) into the relation between injury severity and the weather found that during rainfall crashes with slight injury were relatively more frequent than crashes with serious injury. According to the author, the relatively large increase in less severe crashes in comparison with the increase in serious crashes could possibly be attributed to the drivers already having a reduced grip on the road at lower speeds and are therefore involved in crashes sooner. Furthermore, the average speeds are in fact lower during rainfall, which will therefore result in the average outcome of the crash being less serious, whereas at the same time the crash frequency increases.

In Denmark, during a period of snow the number of injury crashes decreased by 1.2% per additional snowy day (Fridstrøm et al., 1995). Traffic counts show that there was not less traffic. According to the authors, the reduction in crashes was due to adapted driving behaviour and, in addition, possibly by an improved night-time visibility when there is snow on the road, fewer novice drivers, and, finally, the speed reducing effect of vehicles that had skidded off the road to the roadside.

A study from Saudi Arabia (Al-Ghamdi, 2007) reports the number of crash casualties being higher during fog than under other weather conditions. According to Saudi police reports, more than 95% of these fog crashes are caused by insufficient speed reduction for the visibility. Other studies also report that speeds are insufficiently adjusted to visibility (Hogema & Van der Horst, 1994; Brooks et al., 2011). Moreover, an America study (Abdel-Aty et al., 2011) found that the crash severity was higher during fog.

Crash rate

Based on literature, we can assume that the crash rate approximately doubles during rain. The size of the crash rate depends on, among others, the speed limit, the day of the week, and the time of day. Less research has been done on crash rates during other weather conditions. Snow seems to lower the crash rate because it makes people drive more carefully and there probably are fewer vulnerable road users on the road (Fridstrøm et al., 1995). Some examples:

• Dutch research into the effect of weather conditions on road safety (Bijleveld & Churchill, 2009) found that the number of road fatalities doubled during precipitation, with the effects being greater in autumn and in winter than in spring and summer.
• Canadian research in the 1979-1983 period found a 75% higher crash rate during rainfall. This increase was particularly sharp during the first hours of rainfall (Andrey & Yagar, 1991).
• Swiss research of driving speed and crash rate during rainfall, after correction for exposure, found that during rainy daylight hours, in spite of slower driving, the crash rate was 2.5 times larger on non-motorways and 5 times larger on motorways (Thoma, 1993). At night-time these crash rates again more than doubled, to 6 times and 11 times larger respectively.
• German research during the 1970s also showed that the wet, night-time crash rate was nearly twice as high as that during wet, daylight hours (Brühning et al., 1978).

Exposure (mobility)

Weather conditions have an indirect influence on the number of casualties by changes in mobility. However, the influence of weather on the number of vehicle kilometres is generally limited and restricted to rural, recreational traffic. Commuter traffic hardly changes (Hogema, 1996; Bijleveld & Churchill, 2009; Kilpeläinen & Summala, 2007).

Weather conditions also influence the choice of transport mode (Bos, 2001). For example, in bad weather there are fewer cyclists. This can also be seen in the crash statistics: the relative proportion of car-bicycle crashes is a lot lower when it rains, snows, or is foggy than when it is dry (Ellinghaus, 1983). These findings are supported by recent SWOV research on the influence of weather conditions on road safety (Bijleveld & Churchill, 2009). During periods of precipitation, fewer journeys are made with transport modes which are sensitive to weather conditions, such as moped and bicycle.

Obviously, the weather itself cannot be influenced. However, its negative road safety effects can be reduced by measures aimed at man, vehicle, and road.

Behavioural measures

In the driver training, the Netherlands already pays attention to the negative effects of bad weather conditions on driving conditions. However, there have been few campaigns to warn road users about certain weather conditions. The Dutch Directorate-General for Public Works and Water Management can give a weather and traffic alarm when the weather conditions are very bad. Then the media warns road users not to travel unless it is absolutely necessary.

Technological developments during the last few years have resulted in more ways to draw road users' attention to bad weather. Fog detectors such as the Fog Prediction System are installed at a number of locations where poor visibility leads to problems. They warn road users with signs above the road if they approach a road section with poor visibility. These signs could possibly be used to give extra information to drivers about the recommended headway distances.

A fog warning system was found to have a positive effect on the average speed on the A16 motorway in the Netherlands (Hogema & Van der Horst, 1997). Compared with other road sections a decrease in speed of 8 to 10 km/h was measured in the test sections.

A recent evaluation study in Saudi Arabia found that during thick fog (visibility of less than 50 m) a significantly lower speed (‑6,5 km/h) was driven on road sections with a fog warning system than on comparable road sections without such a system. Effects of the warning system were no longer found for fog with a visibility of more than 50 m (Al-Ghamdi, 2007). For none of the fog densities that were investigated the fog warning system was found to have an effect on the variation in speed.

Besides fog warning systems, there are also systems that warn for slipperiness. These systems continuously measure the road surface temperature, the relative humidity, the precipitation, and the quantity of anti-icing chemicals on the road. This data is used to determine whether or not there should be preventative scattering to prevent skidding as much as possible (CROW, 2000).

In France, the speed limit on motorways during rainfall is 20 km/h lower than during dry weather. Such an adaptation also acts as a signal to road users that it is riskier to travel during rain.

Vehicle measures

For cars that were sold after 1^st January 1998 it is mandatory to have a third, rear fog light. The light is located at one precisely defined place on the vehicle so as to avoid mistaking it for a brake light. Since 1^st January 2006, a rear fog light is also compulsory for practically all trailers. Furthermore it is recommended that motor vehicles have an attention-increasing signal installed which is activated with sudden braking. Rain and winter tyres with better road grip during bad weather conditions are also for sale. Technological developments such as ABS and ESP ensure that cars skid less quickly. Lorries can also fitted with adaptations that lessen the amount of splashing water. Finally, testing tyres and windscreen wipers, as is done during vehicle testing, can also prevent crashes during bad weather conditions.

Infrastructural measures

During the past ten years it has become more or less standard to use porous asphalt on road surfaces in the Netherlands. Porous asphalt allows rainwater to drain away quicker than normal asphalt. This reduces the risk of aquaplaning and the road markings remain better visible during rain. In contrast to normal asphalt, porous asphalt also has hardly any road rutting. Furthermore, when it rains, the effect of splashing water is considerably less than on other road surface types. However, its positive effects can be undone by the average road user's risk compensation; during  rain and on a wet road surface road users are thought to drive faster on porous asphalt than they would do on normal asphalt (Tromp, 1993).

With a Slipperiness Warning System, a road authority may receive timely warning to scatter salt if slipperiness is predicted. These systems have been installed on state, provincial, and many municipal roads by placing sensors in the road surface that keep track of various weather parameters. They can give an estimate of the risk of skidding at the location at which they have been placed.

Finally, quality control and maintenance of the road surface material contribute to keeping the risk of aquaplaning small.

In the past, the study of weather conditions has particularly focussed on the influence of rain. The risk of a road crash during rainfall is about twice as high as when it is dry. Although the risk during fog, snow, and fierce gusts of wind is probably even larger, there are four times more crashes during rainfall in the Netherlands, simply because it rains far more often.

Besides influencing the crash rate, weather also influences exposure. However, this influence is limited to the mobility of weekend recreation traffic and to the choice of transport mode. In bad weather, the car traffic volume does not change much, but there are considerably fewer cyclists.

During the last few years, roads and vehicles have been improved to such an extent that one would expect that the crash rate during bad weather had become smaller. However, risk compensating behaviour by motorists on wet porous asphalt can undo vehicle and infrastructural improvements by insufficient speed reduction for the bad weather conditions.

Current literature on weather effects on road safety does not make it possible to determine the effectiveness of various measures. Many international studies were used in the literature study and their results cannot straightforwardly be applied  to the Netherlands.

Abdel-Aty, M., Ekram, A.-A, Huang, H., & Choi, K. (2011). A study on crashes related to visibility obstruction due to fog and smoke. In: Accident Analysis and Prevention, vol. 43, nr. 5, p. 1730-1737.

Agarwal, M., Maze, T. & Souleyrette, R. (2005). Impact of weather on urban freeway traffic flow characteristics and facility capacity. Iowa State University, Center for Transportation Research and Education. Ames, IA.

Al-Ghamdi, A.S. (2007). Experimental evaluation of fog warning system. In: Accident Analysis and Prevention. vol. 39, nr. 6,  p. 1065-1072.

Andrey J. & Yagar, S. (1993). A temporal analysis of rain-related crash risk. In: Accident Analysis and Prevention, vol. 25, nr. 4, p. 465-472.

Bijleveld, F. & Churchill, T. (2009). The influence of weather conditions on road safety. R-2009-9. SWOV Institute for Road Safety Research, Leidschendam.

Bos, J.M.J. (2001). Door weer en wind; Gevolgen van perioden met extreem weer voor de verkeersveiligheid. R-2001-23. Stichting Wetenschappelijk Onderzoek Verkeersveiligheid SWOV, Leidschendam.

Brijs, T., Karlis, D. & Wets, G. (2007). Studying the effect on weather conditions on daily crash counts. In: Proceedings of the 14th International Conference on Road Safety on Four Continents, 14-16 November 2007, Bangkok, Thailand.

Brooks, J., Crisler, M., Klein, N., Goodenough, R., et al. (2011). Speed choice and driving performance in simulated foggy conditions. In: Accident Analysis and Prevention, vol. 43, nr. 3, p 698-705.

Brühning, E., Ammong, D. von, Hippchen, L., Lierz, W., et al. (1978). Forschungsorientierter Zugriff zum Datenbestand der amtlichen Strassenverkehrsunfallstatistik. Bundesanstalt für Straßenwesen BASt, Köln.

CROW (2000). Gladheid: preventie en bestrijding. Publicatie 152. CROW kenniscentrum voor verkeer, vervoer en infrastructuur, Ede.

CROW (2006). Gladheidsbestrijding op maat; Ervaringen met wintergladheidsbestrijding op tweelaags zoab. Infoblad Openbare Ruimte 963. CROW kenniscentrum voor verkeer, vervoer en infrastructuur, Ede.

Debus, G., Heller, D., Wille, M., Dütschke, E., et al. (2005). Risikoanalyse von Massenunfällen bei Nebel. Berichte der Bundesanstalt für Straßenwesen: Mensch und Sicherheit, Heft M 169. Bundesanstalt für Straßenwesen BASt, Bergisch-Gladbach.

DVR (2000). Wind und Wetter. Wie das Wetter auf den Menschen wirkt. DVR-Report 2000, nr. 4, p. 8‑11. Deutscher Verkehrssicherheitsrat e.V., Bonn.

Edwards, J.B. (1998). The relationship between road accident severity and recorded weather. In: Journal of Safety Research, vol. 29, nr. 4, p. 249-262.

Eisenberg, D. (2004). The mixed effects of precipitation on traffic crashes. In: Accident Analysis and Prevention, vol. 36, nr. 4, p. 637-647.

Ellinghaus, D. (1983). Wetter und Autofahren; Eine Untersuchung über den Einfluss des Wetters auf das unfallgeschehen und die Verkehrssicherheit. Uniroyal Verkehrsuntersuchung, Heft 10. Gesellschaft für angewandte Sozialforschung und Planung, GmbH Köln.

Fokkema, H.J. (1987). Weersgesteldheid en verkeersveiligheid. Traffic Test in opdracht van Directie Verkeersveiligheid van het Ministerie van Verkeer en Waterstaat, Veenendaal.

Fridstrøm, L., Ifver, J., Ingebritsen, S., Kulmala, R, et al. (1995). Measuring the contribution of randomness, exposure, weather and daylight to the variation in road accident counts. In: Accidents Analysis and Prevention, vol. 27, nr. 1, p. 1-20.

Hogema, J.H. (1996). Effects of rain on daily traffic volume and on driving behaviour. A study as part of the Project Road and Weather Conditions. Rapport TNO-TM 1996-B019. TNO Human Factors Research Institute TM, Soesterberg.

Hogema, J.H., & Horst, A.R.A. van der (1994). Driving behaviour in fog: analysis of inductive loop data. Report TNO-TM 1994 C-6. TNO Human Factors Research Institute TM, Soesterberg.

Hogema, J.H. & Horst, R. van der (1997). Evaluation of A16 motorway fog-signalling system with respect to driving behaviour. In: Transportation Research Record, vol. 1573, p. 63-67.Laaidi, K. & Laaidi, M. (2002). Météorologie et sécurité routière. In: Via Secura, vol. 56, p. 22-24.

Keay, K. & Simmonds, I. (2005). The association of rainfall and other weather variables with road traffic volume in Melbourne. In: Accidents Analysis and Prevention, vol. 37, nr. 1, p. 109-124.

Kilpeläinen, M. & Summala, H. (2007). Effects of weather and weather forecasts on driver behaviour. In: Transportation Research Part F, vol. 10, nr. 4, p 288-299.

Oppe, S. (1988). Verkeersonveiligheid bij mist. R-88-49. Stichting Wetenschappelijk Onderzoek Verkeersveiligheid SWOV, Leidschendam.

Sabir, M. (2011). Weather and travel behaviour. Vrije Universiteit Amsterdam, Amsterdam.

Smith, K. (1982). How seasonal and weather conditions influence road accidents in Glasgow. In: Scottish Geographical Magazine, vol. 98, p. 103-114.Terpstra, J.M. (1995). Over slecht zicht, bewolking, windstoten en gladheid. Koninklijk Nederlands Meteorologisch Instituut, De Bilt.

Stiers, T. (2005). Studie van de impact van weercondities op de verkeersveiligheid op uurniveau. Limburgs Universitair Centrum, Faculteit Toegepaste Economische Wetenschappen, Leuven.

Thoma, J. (1993). Geschwindigkeitsverhalten und Risiken bei verschiedenen Strassenzuständen, Wochentagen und Tageszeiten. Schweizerische Beratungsstelle für Unfallverhütung BfU, Bern.

Tromp, J.P.M. (1993). Verkeersveiligheid en drainerend asfaltbeton (ZOAB). R-93-35. Stichting Wetenschappelijk Onderzoek Verkeersveiligheid SWOV, Leidschendam.