category: literaturenote citekey: ramaeffectsvariablemessagesigns2000 title: Effects of variable message signs for slippery road conditions on driving speed and headways authors: "Rämä, Pirkko; Kulmala, Risto" year: 2000 date: 2000-00-00 2000 doi: 10.1016/s1369-8478(00)00018-8 publication: "Transportation Research Part F: Traffic Psychology and Behaviour" url: "https://doi.org/10.1016/s1369-8478(00)00018-8" zotero_key: NUXH9XJS zotero_storage: K6GRS8TJ collections: transpordi_planeerimine folder: 001_artiklid firstAuthor: "Rämä, Pirkko"

status: converted

Transportation Research Part F 3 (2000) 85±94

www.elsevier.com/locate/trf

Eects of variable message signs for slippery road conditions on driving speed and headways

Pirkko Rama *, Risto Kulmala

Technical Research Centre of Finland, P.O. Box 1902, FIN-02044 VTT, Finland Received 20 June 1999; received in revised form 29 March 2000; accepted 20 April 2000

Abstract

This ®eld study investigated the eects of two variable message signs (VMS) on driver behaviour. Speci®cally, the signs were a warning sign for slippery road conditions and a minimum headway sign. The study was performed as a before-and-after experiment at three test sites in Finland with an after period covering two winter seasons. The results showed that the slippery road condition sign reduced the mean speed on slippery roads by 1±2 km/h in addition to the decrease caused by the adverse road conditions. The minimum headway sign decreased the proportion of headways shorter than 1.5 s for cars in car-following situations, in addition to a speed reduction of 1 km/h. The eects were somewhat smaller during the second winter than the ®rst. Ó 2000 Elsevier Science Ltd. All rights reserved.

Keywords: Speed; Variable message signs; Headways; Road conditions

1. Introduction

In Finland, accident risk is substantially higher in winter than in summer. This is primarily due to ice and snow on the roads. The risk of injury accidents in winter has been estimated to be over 20 times higher on an icy road surface than on a dry and bare one (Malmivuo & Peltola, 1997).

The usual measure against accidents on icy and snowy roads has been to improve the road surface conditions with the help of winter maintenance operations (i.e. snow removal, de-icing with salt, sanding, etc.). In spite of investments in winter maintenance, such as preventive salting of the roads based on weather forecasts, there is normally a delay between detection of slipperiness and maintenance operations. It has been estimated that accident risk during the hour preceding maintenance actions is 12 times as high as 12 h earlier (Oberg, 1993). However, during

^* Corresponding author. Tel.: +358-9-456-4551; fax: +358-9-464-850. E-mail address: pirkko.rama@vtt.® (P. RaÈmaÈ).

these most risky hours, operators at the Trac Management Centre (TMC) are usually aware of the slippery conditions. Trac safety would probably be improved by providing road users with real-time local information about adverse road conditions, especially during the period from the ®rst indication of slipperiness observed at the TMC to the start of maintenance operations on that road section. This is justi®ed by the fact that problems with slippery road surfaces are often concentrated to certain parts of the road network such as bridges, coastal sections, etc.

The poor ability of drivers to recognise road slipperiness emphasises the need for local warnings. On slippery road surfaces, only 14% of Finnish drivers estimate the road to be slippery, while over half regard the friction as quite normal (Heinijoki, 1994).

On average, drivers adapt their speed behaviour to some extent to road surface conditions. More speci®cally, average speeds on a slippery road surface are roughly 4 km/h lower than in good winter conditions, and the standard deviation of speed is also lower due especially to a drop in the highest speeds (Saastamoinen, 1993). However, this adaptation has been proved insucient to fully compensate for the reduced friction (Malmivuo & Peltola, 1997; Roine, 1993). In general, headways are not substantially aected by winter conditions. However, in south-western Finland where the experiment took place, the percentage of headways shorter than 1.5 s was found to be approximately 25% in slippery conditions compared to 39% in good winter conditions (Saastamoinen, 1993). Overall, these problems call for local warning systems such as those based on variable message signs (VMS).

The current literature includes only a few studies that have investigated the eects of variable weather related signs on driver behaviour. Most of the earlier studies concerned variable fog warning signs. Cooper and Sawyer (1993) evaluated the fog warning system on the M25 London orbital motorway in which a text message ``FOG'' was presented when fog was formed. An overall reduction in mean vehicle speeds of about 1.8 miles per hour (3 km/h) was found when the signals were switched on. The authors concluded that the speed reductions indicate that drivers are alerted to the presence of fog ahead, which means that drivers are more likely to respond more quickly to the hazard itself. However, the report does not give any details about how the eect of possibly worsening weather conditions at the measurement site is taken into account.

In the Netherlands, Hogema and Van der Horst (1997) found the eect of the variable fog warning system to be an 8±10 km/h reduction in mean speed. The standard deviation of speed decreased marginally. In extremely dense fog, the system seemed to have the opposite eect on speed, because drivers tended to rely heavily on the speed limits presented. The system included variable speed limits and a warning sign. The warning sign was an exclamation mark with a supplementary text panel ``MIST'' (i.e. fog). The signs showed white symbols on a black background. The speed limits were presented without a red circle and might therefore have been understood as recommendations and not mandatory speed limits as intended (see Rama & Luoma, 1999).

In order to decide which types of pictograms and text messages should be used in VMS for slippery road conditions in Finland, the comprehension and preference of alternative messages was studied (Kosonen & Luoma, 1994). The best-comprehended pictogram was the sign with a car and wavy lines inside a red triangle, with a snow¯ake icon as a supplemental panel (Fig. 1). Among the text message signs, the one giving the recommended following distance was preferred.

Many impacts of the VMS for slippery road conditions on driver behaviour were hypothesised at the start of the present study. Speci®cally, positive impacts while the signs are on might include: improved driver awareness of slipperiness; more alert and attentive drivers; reduced driving speed;

Fig. 1. Diagram showing (a) the slippery road condition sign and (b) the minimum headway sign. The messages displayed depend on the road surface condition classi®cation by the road weather monitoring centre: the slippery road condition sign is on or o and the minimum headway sign is always on, but the recommended minimum headway varies. In addition, the minimum headway depends on vehicle type and speed.

increased headways; and more homogeneous trac ¯ow. However, the sign might have negative impacts as well, such as comprehension problems of the message, excessive anxiety, problems in deciding how to change driving behaviour, decreased attention to other driving tasks, and heterogeneous adaptation of driver behaviour.

In consequence, implementation of a telematic system may end up delegating responsibility to the system rather than the drivers. When the sign is o, the following impacts might result (compared with the situation before implementation of the system): decreased observation of road conditions, decreased anticipation of slipperiness, increased driving speed, and shorter headways between vehicles. In addition, an inappropriately operated system might result in negative impacts: trac safety would probably decrease if the road conditions were poor but the sign was o, and false alarms would probably decrease the credibility of the system. Finally, the system might have various side eects such as behavioural changes on the roads not covered by the system.

Many of the above hypothesised eects could be assumed to be time dependent. There could be some eects due to the novelty of the system, in addition to which any eect is probably highly dependent on driversÕ experiences of the systemÕs reliability. The eects may diminish if the system does not prove to be reliable.

Although the number of potential eects of VMS for slippery road conditions is substantial, this study was limited to investigating how VMS for slippery road conditions aect driver behaviour in terms of speed and headways. A complementary study (Luoma, Rama, Penttinen & Anttila, 2000) aimed to cover other potential eects of these signs, and Rama, Kulmala and Heinonen (1996) have investigated driver acceptance and comprehension of these signs.

2. Method

2.1. Signs

Two VMSs were evaluated: (a) a slippery road condition sign and (b) a sign recommending the minimum headway between vehicles (Fig. 1).

The TMC operators classi®ed road surface conditions in three categories for the purpose of VMS control: (a) good, (b) possibly slippery and (c) veri®ed slippery. The classi®cation was based on data from local automatic weather monitoring stations, a road weather closed circuit television (CCTV) camera and the expertise of ®eld personnel and the TMC operators.

The TMC operators switched on the sign for slippery road conditions when they assumed that the road was becoming slippery because of snow or ice. At the same time, this was reported to the local road authorities, who veri®ed the slipperiness and reported back. If the slipperiness was con®rmed, the TMC operator switched the sign to ¯ashing mode. The sign was switched o when the operator concluded that maintenance action had removed the slipperiness.

The minimum headway sign was adapted to the road surface condition. In addition, it also took into account vehicle type (based on vehicle length more or less than 8 m) and speed of the vehicle. The minimum headway sign was actually the estimated stopping distance including a 1-s reaction time (2 s for long vehicles taking into account braking delays). During the second winter, in February 1995, the word ``SUOSITUS'' (ÔrecommendationÕ) was added below the headway pictogram to improve comprehension of the VMS.

2.2. Design and sites

The design was a before-and-after study with controls. There were three sites in southwest Finland (Site 1 at Eurajoki, Site 2 at Koski and Site 3 in Salo). The posted speed limit at the sites was 80 km/h. The after period covered two winters to examine possible novelty and adaptation eects. The set-up shown in Fig. 2 is for Site 1 but was the same at Sites 2 and 3, with the exception that there was no minimum headway sign. The study included data from two winter periods at Sites 1 and 2, and from one winter period at Site 3.

Driving speeds and headways were measured automatically with loop detector based trac monitoring stations (TS). Driving behaviour was monitored before the signs 536±1,800 m upstream (TS1), and after the signs 360±1100 m downstream (TS2). In addition, a third measurement point (TS3, not at Site 3) was used 7,670±13,000 m downstream. At Site 1, there was a notable intersection between TS2 and TS3, and drivers at the third point only partly represented the drivers who had encountered the signs. It was considered preferable to install the trac monitoring stations TS1 and TS2 in places where the road curved slightly, as we assumed this to emphasise the eects of road surface conditions, particularly on driving speeds. In general, the installations were realised as planned. However, the curves were somewhat slight at Site 2, and at Site 3 TS1 had to be installed uphill on a section with an overtaking lane.

The experimental design included six experimental sign conditions as follows: (1) steady slippery road condition sign, (2) ¯ashing slippery road condition sign, (3) minimum headway sign- + steady slippery road condition sign, (4) minimum headway sign + ¯ashing slippery road condition sign, (5) minimum headway sign only, and (6) nothing (black sign). The use of a steady or ¯ashing slippery road condition sign de®ned which data were included in the study. The other data were matched by day of week and hour of day. In the matching process, ÔNothing onÕ data and a ÔMinimum headway sign onlyÕ data were selected as comparison data for both the steady and ¯ashing signs. Thus, the ®rst four conditions represented hours when the sign in question had been used (are typically used), and the last two covered wider time periods.

Fig. 2. Experimental set-up (TS Trac monitoring Station).

In the analysis, we primarily compared driver behaviour at TS2 in relation to the behaviour at TS1 both before and after implementation of the signs. The behaviour at TS1 should indicate the ``baseline'' behaviour in the prevailing weather, road surface and trac conditions, and the difference between that and the behaviour at the other two stations, the eects of the system. To control the possible local and before±after dierences in the weather and road surface conditions, we used driver behaviour in the opposite direction as an additional measure. The before period lasted from mid-December 1993 to mid-February 1994 at Sites 1 and 2, and from October 1994 to December 1994 at Site 3. During these periods, the signs were not yet erected but the TMC operators acted as if the system were on and logged their decisions (turning the VMS on, ¯ashing or not) accordingly. The ®rst after period at Sites 1 and 2 lasted from mid-February to mid-April 1994. The second after period at Sites 1 and 2 lasted from mid-October 1994 to mid-April 1995, and the after period at Site 3 from January 1995 to mid-April 1995.

In the main speed analysis, only cars travelling in free trac ¯ow conditions (with a minimum headway of 5 s to the vehicle ahead) were included. The following formula was used to compute the eect of the VMS:

$$\Delta \mu = (\mu{e,TS2} - \mu{e,TS1}) - (\mu{c,TS2} - \mu{c,TS1}),\tag{1}$$

where l is the mean speed, e the experimental direction, c the control direction, TS1 the trac monitoring station 1, and TS2 is the trac monitoring station 2. Dl was estimated for both before and after implementation of the system, and the eect of the VMS was ®nally estimated as Dlafter ÿ Dlbefore. Observations were aggregated hourly and the means were weighted by the number of observations.

In the statistical analyses, signi®cant dierences between mean speeds were determined with t tests. The P-value is given, where statistical testing has been done.

When examining headways, we only included vehicles with a headway of 5 s or less in the analysis. To estimate the eects of various factors we used a log-linear model to describe the proportion of short headways as follows:

$$HW < 1.5{TS2} = HW < 1.5{TS1} + e^{\sum \beta_1 x_1},$$ (2)

where HW < 1:5 is the proportion of headways shorter than 1.5 s out of all headways of 5 s or less, xi's are the explanators in the model, and bⁱ 's are the parameter values in the model. The explanatory variables were the sign condition, the before±after variable and direction (experimental-control). The number of short headways during the 1-h period was assumed to follow the Poisson distribution.

3. Results

3.1. Speed

We assumed the eect of the VMS on mean speed to be dependent on the speed level before the VMS. In good road conditions, the mean speeds at TS1 were 87 km/h (Site 1), 91 km/h (Site 2) and 84 km/h (Site 3). In adverse road conditions, the mean speeds were 5 km/h lower on average.

The main results showed that the slippery road condition sign (at Site 1 in combination with the minimum headway sign) decreased the mean speed of cars travelling in free ¯ow trac at Sites 1 and 2 on average by 1.2 km/h (P < :05) with the steady display. When the sign was ¯ashing (Table 1)

Table 1 Eects of variable road condition signs on mean speed (km/h) for cars travelling in free ¯ow trac, and in parenthesis for total trac ¯ow (n varied between 603 and 32,361)

| Site | VMS | Speed e€ect (km/h) (TS2±TS1) | | | | | |------|--------------------------------------------------------------------|------------------------------|--------|----------|--------|--| | | | Winter 1 | | Winter 2 | | | | 1 | Slippery road condition sign (steady) and
minimum headway sign | )0.9 | ()0.5) | )0.8 | ()0.1) | | | | Slippery road condition sign (¯ashing) and
minimum headway sign | )2.2 | ()2.2) | )1.9 | ()1.8) | | | | Minimum headway sign | )1.1 | ()1.2) | )1.0 | ()1.1) | | | 2 | Slippery road condition sign (steady) | )1.8 | ()1.5) | ) | ()0.6) | | | | Slippery road condition sign (¯ashing) | )2.3 | ()1.7) | ) | ()1.3) | | | | No message | +0.1 | ()0.3) | ) | ()0.2) | | | 3 | Slippery road condition sign (steady) | ) | | +1.4 | (+1.2) | | | | Slippery road condition sign (¯ashing) | ) | | +0.8 | (+0.5) | | | | No message | ) | | +0.1 | (+0.1) | |

^* P < :05:

the decrease in speed at Sites 1 and 2 was on average 2.1 km/h (P < :05). At Site 3 the system had the opposite eect, the average speed increasing by 1.1 km/h (P < :05, Table 1).

The eects were on an average 0.3 km/h smaller during the second winter than during the ®rst. Also, the ®gures for all vehicles (results for the second winter season for Site 2 were only available for all vehicles) indicated a smaller eect (on average 0.3 km/h) than the ®gures for cars travelling in free ¯ow trac (Table 1). When no message was shown, the signs had no signi®cant eect on mean speed.

Three analyses were performed in addition to the main speed analysis. First, at the monitoring station further downstream (TS3) the ¯ashing slippery road condition sign decreased the mean speed on average by 2 km/h (P < :05). The steady display at TS3 had no signi®cant eects at Site 2 and signi®cant increasing eects (on average 0.9 km/h, P < :05) at Site 1.

Second, the mean speed eect was checked for light conditions (i.e. day versus night). The results showed that the eect tended to be more substantial at night.

Third, the eects on the speed distribution were examined. The 85 percentile speeds changed in similar fashion to the mean speed. At Site 3, the standard deviation of speed seemed to be slightly decreased by the VMS. However, the signs did not have any substantial eect on the standard deviation of speed.

3.2. Headways

Investigation of the headway distributions showed that at Site 1, with the minimum headway sign, there were signi®cant changes (P < :05) in the percentage of short headways during the after period (Table 2). In the before period, the percentage of drivers with a headway of less than 1.5 s was 25±35% in good road surface conditions and 16±28% in slippery road conditions. In the after period, the percentage of short headways was 16±18% in good road surface conditions and 13% in slippery conditions. The relative eects were large for all surface conditions, including good road conditions. The minimum headway sign seemed to aect headways in the desired manner, and even seemed to shift the whole headway distribution. At Site 1, during adverse weather and road conditions the mode of the distribution was between 2 and 3 s during the after studies with the VMS on, and between 1.5 and 2 s during the before studies.

The results at Site 2 showed that the slippery road condition sign alone did not in¯uence headways in a substantial manner. The percentage of drivers with a headway of less than 1.5 s

Table 2
	Changes in the percentual shares of headways shorter than 1.5 s out of all headways of 5 s or lessa
Site/road condition	Change in percentage of short headways

| Site/road condition | Change in percentage of short headways | | | |--------------------------|----------------------------------------|--------------|--| | | Winter 1 (%) | Winter 2 (%) | | | Site 1/good condition | )38 | )28 | | | Site 1/possibly slippery | )47 | )37 | | | Site 1/veri®ed slippery | )37 | )31 | | | Site 3/possibly slippery | ) | )12 | |

^a Only statistically signi®cant (P < :05) eects are shown.

stayed the same as at Site 1 before installation of the VMS. However, at Site 3 there was a slight signi®cant decrease in the proportion of headways shorter than 1.5 s (P < :05).

During the second winter, the eects were approximately one third smaller. Additional analysis showed that at Site 1 the mean decrease in the proportion of headways shorter than 1.5 s was 41% during the ®rst winter, 35% during the second winter before the word ÔrecommendationÕ was added to the sign in early February, and 24% after the word was added.

4. Discussion

The purpose of the study was to investigate how a VMS warning of slippery road conditions and a minimum headway sign in¯uenced driver behaviour in terms of speed and headways. The slippery road condition sign decreased driving speeds by 1±2 km/h at a distance of 500±1100 m after the signs. The eects were more substantial at Site 1 than at Site 2, where the road environment was more monotonous (i.e. less hilly and less curvy). Overall, these eects are meaningful because it has been shown that even small decreases in average speed are related to reduced accident risks (e.g. Ranta & Kallberg, 1996). At distances of 3±14 km after the signs, the speed reductions were smaller and in part not statistically signi®cant, even the opposite. Admittedly the measures for longer distances were not optimal because of local dierences both in weather conditions and in the driver sample. The eects seemed to be more substantial at night, implying that these VMS are more eective when they are more conspicuous.

At Site 3, the slippery road condition sign increased the mean speed slightly. The reason for this negative eect might be the unrepresentativeness of road conditions in the before data (because of the short data collection period). There were much heavier snowfalls during the after period. Because there was a moderate hill with an overtaking lane at the measurement point before the sign, snow on the road might lower the mean speed more at that measurement point than at the measurement point after the sign. The experimental design of the present study might be sensitive to this type of error source. In addition, the slippery road condition sign might not be eective in all road environments.

The minimum headway sign in¯uenced headways in the desired manner. The proportion of short headways was decreased and the VMS seemed to shift the whole headway distribution so that headways became longer. The percentage of short headways before installation of the sign corresponded well to those observed in an earlier study (Saastamoinen, 1993). When interpreting the changes it should remembered that the percentage of short headways was considerably higher in good road surface conditions than in slippery conditions. Hence the high relative changes in good conditions in comparison to those in slippery conditions are not surprising. The slippery road condition sign alone did not seem to aect headways substantially.

The ¯ashing mode aected behaviour the most. The eect was more substantial and lasted longer, which was expected because ¯ashing is frequently used to signal danger. On the other hand, the results of Rama et al. (1996) showed that drivers had problems understanding the meaning of the ¯ashing mode. Consequently, a ¯ashing VMS cannot be recommended based on this study. In general, use of ¯ashing lights is a principal and system level issue, and could not be based on an individual study concerning one type of sign (Rama & Luoma, 1999).

In our experiment, comprehension of the minimum headway sign was ensured by adding the text ``recommendation''. This change, however, weakened the eect of the sign on driver behaviour. There are several speculative explanations for this. One is that by adding the word ÔrecommendationÕ the value of the message as real-time information or on behavioural feedback was diminished, and the message was understood as a more general type of information or education. Also, driversÕ ability to estimate the distance to the vehicle ahead is not very well known. There is a clear need for further development and research of this type of sign. According to the results of Rama et al. (1996), the majority of drivers regarded both signs useful: 65% approved of the slippery road condition sign and 72% the minimum headway sign. Seventy percent of drivers in the ®rst winter and 80% in the second winter interpreted the meaning of the slippery road condition sign correctly.

The signs had a very speci®c eect. The slippery road condition sign aected driving speeds but did not in¯uence headways (except for the slight decrease at one site). The minimum headway sign aected headways substantially, and also had an eect on speeds (a decrease of approximately 1 km/h).

The eects seem partly to be ``novelty eects'' which gradually diminish as drivers get used to the signs. On the other hand, as drivers get used to the system the eects may also become more substantial, especially if the system proves to be reliable in its operation. In the present study, the eects were measured for one short period during the ®rst winter and for a second winter season. It is dicult to decide what would be the correct follow-up time period. It is probable that the eects are dependent both on driver experiences of the current system, but also on more general information and knowledge about real time information systems.

The weather controlled variable warning signs are essentially meant for spots of exceptional weather and road surface conditions, and to be controlled with the help of a neighbouring road weather station or CCTV. For the purpose of informing and warning drivers about area-wide problems of slippery road surfaces, other types of services should be used.

In conclusion, the VMS increased trac safety by decreasing mean speeds and increasing the following distances. These positive eects were not found at all test sites. Careful use of the VMS at speci®c sites only is recommended. In addition, the variable message signs have other eects besides those that can be measured in terms of speed and headway. Some of those eects have been investigated in a complementary study by Luoma et al. (in press).

Acknowledgement

Appreciation is extended to the Finnish National Road Administration (Finnra) for support of this research. The authors wish to thank the following persons for their assistance and support with this research: Pekka Kulmala, Erkki Ritari and Arja Wuolijoki of the Technical Research Centre of Finland (VTT) for helping with the data collection and initial data analyses, Jorma Helin, Kari Karessuo and Eini Hirvenoja of the Finnish National Road Administration (Finnra) for their valuable consultation and advice, and Juha Luoma (VTT) for his helpful suggestions on earlier drafts of this paper. However, the authors are solely responsible for the contents of this paper.

References

Cooper, B. R., & Sawyer, H. (1993). Assessment of M25 automatic fog-warning system (Final report, Project Report 16). Crowthorne: Transport Research Laboratory.
Heinijoki, H. (1994). Kelin kokemisen, rengaskunnon ja rengastyypin vaikutus nopeuskayttaytymiseen In¯uence of the type and conditions of tyres and drivers perceptions of road conditions on driving speed. Helsinki: Finnish National Road Administration.
Hogema, J. H., & Van der Horst, R. (1997). Evaluation of A16 motorway fog-signalling system with respect to driving behaviour. Transportation Research Record, 1573, 63±67.
Kosonen, E., & Luoma, J. (1994). Comprehension of and preference among variable message signs for poor road conditions. In Proceedings of the 12th Triennal Congress of the International Ergonomics Association (Vol. 4, pp. 343± 345). Toronto: International Ergonomics Association.
Luoma, J., Rama, P., Penttinen, M., & Anttila, V. (2000). Eects of variable message signs for slippery road conditions on reported driver behaviour. Transportation Research, Part F, 3, 75±84.
Malmivuo, M., & Peltola, H. (1997). Talviajan liikenneturvallisuus- tilastollinen tarkastelu Trac safety at wintertime ± a statistical investigation. Helsinki: Finnish National Road Administration.
Oberg, G. (1993). Trac on ice and snow. Nordic Road and Transport Research, 3, 32±33.
Rama, P., Kulmala, R., & Heinonen, M. (1996). Muuttuvien kelivaroitusmerkkien vaikutus ajonopeuksiin, aikavaleihin ja kuljettajien kasityksiin The eect of variable road condition warning signs. Helsinki: Finnish National Road Administration.
Rama, P., & Luoma, J (1999). Comprehension of VMS in Europe. In Proceedings of Sixth World Congress on Intelligent Transport Systems (ITS 99). Washington, DC: ITS America and ITS Canada.
Ranta, S., & Kallberg, V.-P. (1996). Ajonopeuden turvallisuusvaikutuksia koskevien tilastollisten tutkimusten analyysi Analysis of statistical studies of the eects of speed on safety. Helsinki: Finnish National Road Administration.
Roine, M. (1993). Kuljettajakayttaytyminen kaarre-ja jonoajossa Driver behaviour in sharp curves and queues on main roads. Helsinki: Finnish National Road Administration.
Saastamoinen, K. (1993). Kelin vaikutus ajokayttaytymiseen ja liikennevirran ominaisuuksiin Eect of road conditions on driving behaviour and properties of the trac ¯ow. Helsinki: Finnish National Road Administration.

ramaeffectsvariablemessagesigns2000.md 31 KB Előzmények Nyers