his State of the Art report summarizes the work done within the COST Action Nour-Eddin El Faouzi

T R283 TU0702 “Real-time monitoring, surveil- lance and control of networks under Editor adverse weather conditions”. It provides a comprehensive synthesis about the effects of adverse weather conditions on road traf- REAL-TIME MONITORING, fic operations and road safety as well as the best practices which are available in various SURVEILLANCE AND CONTROL countries. Several projects were carried out in the last OF ROAD NETWORKS years on this topic (especially in Austria, UNDER ADVERSE Denmark, Finland, France, Germany, Greece, Poland and USA) and coordina- WEATHER CONDITIONS tion at the European level has only recently emerged, through initiatives such as this COST Action. The next step could logically be the inte- gration of weather forecasting and management capabilities together in an inte- grated framework that captures the effect of weather and weather-related strategies on traffic system performance and achieve weather-sensitive traffic management tools.

Nour-Eddin El Faouzi is a research director at INRETS, head of Transport and traffic engineering laboratory (LICIT), a joint research unit of INRETS - ENTPE and chair of TU0702 COST Action. Photo de couverture : J. Navarro, l'autoroute A64 le 11 février 2009, pour le site www.ladepeche.fr

Effects of weather on traffic and

December 2010 CONDITIONS WEATHER ADVERSE OF ROAD NETWORKS UNDER AND CONTROL SURVEILLANCE REAL-TIME MONITORING, pavement: State of art and best 45  practices

ISSN 0768-9756 ISBN 978-2-85782-688-0 ESF provides the COST Office through an EC contract. COST is © Les collections de l’INRETS supported by the EU RTD Framework programme. Recherches Recherches Conformément à la note du 04/07/2014 de la direction générale de l'Ifsttar précisant la politique de diffusion des ouvrages parus dans les collections éditées par l'Institut, la reproduction de cet ouvrage est autorisée selon les termes de la licence CC BY-NC-ND. Cette licence autorise la redistribution non commerciale de copies identiques à l’original. Dans ce cadre, cet ouvrage peut être copié, distribué et communiqué par tous moyens et sous tous formats.

Attribution — Vous devez créditer l'Oeuvre et intégrer un lien vers la licence. Vous devez indiquer ces informations par tous les moyens possibles mais vous ne pouvez pas suggérer que l'Ifsttar vous soutient ou soutient la façon dont vous avez utilisé son Oeuvre. Pas d’Utilisation Commerciale — Vous n'êtes pas autorisé à faire un usage commercial de cette Oeuvre, tout ou partie du matériel la composant. (CC BY-NC-ND 4.0) Pas de modifications — Dans le cas où vous effectuez une adaptation, que vous transformez, ou créez à partir du matériel composant l'Oeuvre originale (par exemple, une traduction, etc.), vous n'êtes pas autorisé à distribuer ou mettre à disposition l'Oeuvre modifiée.

Le patrimoine scientifique de l'Ifsttar

Le libre accès à l'information scientifique est aujourd'hui devenu essentiel pour favoriser la circulation du savoir et pour contribuer à l'innovation et au développement socio-économique. Pour que les résultats des recherches soient plus largement diffusés, lus et utilisés pour de nouveaux travaux, l’Ifsttar a entrepris la numérisation et la mise en ligne de son fonds documentaire. Ainsi, en complément des ouvrages disponibles à la vente, certaines références des collections de l'INRETS et du LCPC sont dès à présent mises à disposition en téléchargement gratuit selon les termes de la licence Creative Commons CC BY-NC-ND.

Le service Politique éditoriale scientifique et technique de l'Ifsttar diffuse différentes collections qui sont le reflet des recherches menées par l'institut : • Les collections de l'INRETS, Actes • Les collections de l'INRETS, Outils et Méthodes • Les collections de l'INRETS, Recherches • Les collections de l'INRETS, Synthèses • Les collections du LCPC, Actes • Les collections du LCPC, Etudes et recherches des laboratoires des ponts et chaussées • Les collections du LCPC, Rapport de recherche des laboratoires des ponts et chaussées • Les collections du LCPC, Techniques et méthodes des laboratoires des ponts et chaussées, Guide technique • Les collections du LCPC, Techniques et méthodes des laboratoires des ponts et chaussées, Méthode d'essai

www.ifsttar.fr

Institut Français des Sciences et Techniques des Réseaux, de l'Aménagement et des Transports 14-20 Newton, Cité Descartes, Champs sur Marne F-77447 Marne la Vallée Cedex 2 Contact : [email protected]

Nour-Eddin El Faouzi Editor

Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Effects of weather on traffic and pavement: State of the art and best practices

RECHERCHES © Les collections de l’INRETS December 2010 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

The editor Dr. Nour-Eddin El Faouzi LICIT, joint Research Unit of INRETS - ENTPE [email protected]

The laboratory units LICIT: Transport and Traffic Engineering Laboratory, 25 François Mitterrand, case 24, 69675 Bron CEDEX, France INRETS: French National Institute for Transport and Safety Research, 25 avenue François Mitterrand, case 24, 69675 Bron CEDEX, France ENTPE: National School of State Public Works, rue Maurice Audin, 69518 Vaulx en Velin CEDEX, France

The contributors of this report are listed on the next page.

The French National Institute for Transport and Safety Research – INRETS Direction scientifique / politique éditoriale – Aude Lauby 25 avenue François Mitterrand Case 24, 69675 Bron CEDEX, France Tél. : +33 (0)4 72 14 23 20 – Fax : +33 (0)4 72 37 68 37 – www.inrets.fr

© Les collections de l’INRETS – Réf. : R283 ISBN 978-2-85782-688-0 ISSN 0768-9756 En application du code de la propriété intellectuelle, l’INRETS interdit toute reproduction intégrale ou partielle du présent ouvrage par quelque procédé que ce soit, sous réserve des exceptions légales

2 © Les collections de l’INRETS

Contributors

Maurice Aron, Researcher, INRETS-GRETIA, Le Descartes 2, 2 rue de la Butte Verte, 93166 Noisy le Grand Cedex, France, [email protected] Johannes Asamer, Researcher, Austrian Institute of Technology (AIT), Österreichisches Forschungs - und Prüfzentrum Arsenal Ges.m.b.H., Giefinggasse 2, 1210 Vienna, Austria, [email protected] Ashish Bhaskar, Post-Doc Fellow, EPFL LAVOC, station 18, 1015 Lausanne, Switzerland, [email protected] Wouter Van Bijsterveld, GEOCISA, Calle de Los Llanos de Jerez 10-12, 28823 Coslada, Spain, [email protected] Romain Billot, Ph.D Student, INRETS, Transport and Traffic Engineering Laboratory – LICIT, INRETS – ENTPE, 25 avenue François Mitterrand, case 24, 69675 BRON cedex, France, [email protected] Nicolas Bueche, Ph.D. Student, EPFL LAVOC, station 18, 1015 Lausanne, Switzerland, [email protected] Halim Ceylan, Professor, Public, Engineering, Pamukkale University, Kinikli Kapmusu, 20017 Denizli, Turkey, [email protected] Edward Chung, Professor, Queensland University of Technology, School of Urban Development, Faculty of Built Environment and Engineering, Gardens Point Campus, GPO Box 2434 BRISBANE QLD 4001, [email protected] Thorsten Cypra, Ph.D., Boschung Mecatronic AG, Route d’Englisberg 21, 1763 Granges-Paccot, Switzerland, [email protected] Alexandre Dinkel, Technische Universität München, Fakultät für Bauingenieur- und Vermessungswesen Lehrstuhl für Verkehrstechnik, Arcisstr. 21 80333 München, [email protected] Min-Tan Do, Researcher, LCPC, Centre de Nantes, Route de Bouaye, BP 4129, 44341 Bouguenais Cedex, [email protected] Nour-Eddin El Faouzi, Research Director and Head of Transport and Traffic Engineering Laboratory – LICIT, INRETS – ENTPE, 25 avenue François Mitterrand, case 24, 69675 BRON Cedex, France. [email protected] Bernhard Heilmann, Researcher, Austrian Institute of Technology (AIT), Österreichisches Forschungs - und Prüfzentrum Arsenal Ges.m.b.H., Giefinggasse 2, 1210 Vienna, Austria, [email protected]

© Les collections de l’INRETS 3 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Christian Holldorb, Professor, Biberach University of Applied Sciences, Karlstrasse11, 88400 Biberach, Germany, [email protected] Nicolai Jonasson, The Icelandic Road Administration, Reykjavik, Iceland [email protected] Mila Mihaylova, Reader, School of Computing and Communications, InfoLab21, South Drive, Lancaster University, Lancaster LA1 4WA, United Kingdom, [email protected] Pertti Nurmi, Finnish Meteorological Institute, Erik Palmenin aukio 1, PO Box 503, 00101 Helsinki, Finland, [email protected] Bent Juhl Pedersen, Danish Road Institute, Denmark Road Directorate Jyllandsvej 100 DK-5500 Middelfart Denmark, [email protected] Michal Karkowski, Road and Bridge Research Institute – IBDIM, 1, Instytutowa Str. 03-302 Warsaw, Poland. [email protected] Matthew Karlaftis, Professor, Department of Transportation Planning and Engineering, School of Civil Engineering, National Technical University of Athens, 5 Iroon Polytechniou Str., Zografou Campus, Athens. Greece. [email protected] Karol J. Kowalski, Department of Building Materials Engineering, Faculty of Civil Engineering, Warsaw University of Technology, [email protected] Finn Krog Kristensen, Road Directorate, Danish Road Institute, Gulgalderen 12, 2640 Hedehusene, Denmark, [email protected] Einar Palsson, The Icelandic Road Administration, Reykjavik, Iceland. [email protected] Luis Picado Santos, Professor, DECivil, Instituto Superior Tecnico, Technical University of Lisbon, Av. Rovisco Pais, 1049-001 Lisbon, Portugal, [email protected] Yrjo Pilli-Sihvola, Centre for Economic Development, Transport and the Environment, Kauppamiehenkatu 4, FI 45100 Kouvola, Finland, yrjo.pilli- [email protected] Patrick Rychen, Ph.D. Student, EPFL LAVOC, station 18, 1015 Lausanne, Switzerland, [email protected] Suzanne Schulz, Karlsruhe Institute of Technology, Institute of and Railroad Engineering, Otto-Ammann-Platz 1, 76131 Karlsruhe, Germany, [email protected] Serdal Terzi, Associate Professor, Suleyman Demirel University Technical Education Faculty, Construction Education Department, 32260 Isparta, Turkey, [email protected] Eleni I. Vlahogianni, Lecturer, Department of Transportation Planning and Engineering, School of Civil Engineering, National Technical University of Athens, 5 Iroon Polytechniou Str., Zografou Campus, Athens. Greece. [email protected]

4 © Les collections de l’INRETS

Contents

Acknowledgments ...... 7 Introduction ...... 9 1. Definition and terminology associated with the adverse weather ...... 13 2. Effects of weather on traffic and safety ...... 21 2.1. Impact of weather on traffic operations and Drivers’ behaviours ..... 21 Introduction ...... 21 2.1.1. Impact on macroscopic traffic characteristics ...... 22 2.1.2. Impacts on microscopic traffic characteristics ...... 31 2.2. Impact of weather on road safety ...... 36 Synthesis ...... 47 3. Effects of weather on pavement ...... 49 3.1. Effects of wetness on pavement skid resistance ...... 49 3.1.1. Frictions ...... 49 3.1.2. Skid resistance ...... 49 3.1.3. Tyre/road contact area ...... 49 3.1.4. mechanisms ...... 50 3.1.5. Effect of water film thickness on tyre/road friction ...... 56 3.1.6. Skid resistance on dry versus wet pavements ...... 56 3.2. Crash risks related to wet ...... 60 3.3. Assessment of road surface condition ...... 62 3.4. Effect of snow and ice on pavement skid resistance ...... 70 3.4.1. Tyre and wintry road friction ...... 70 3.4.2. Tyre and road contact area ...... 70 3.4.3. Effect of snow/ice on tyre/road friction ...... 71 3.4.4. Accident risks related to wintry road surface ...... 72 3.5. Assessment and prediction of road surface condition ...... 76 3.5.1. Measurements of the pavement condition ...... 76 3.5.2. Prediction of pavement condition ...... 79 4. Operational state of practice and best practices ...... 83 4.1. Weather sensing and RWIS ...... 83 Introduction ...... 83 4.1.1. RWIS Infrastructure ...... 85 4.1.2. Weather data granularity and applications ...... 85 4.1.3. Quality Needs Assessment ...... 86

© Les collections de l’INRETS 5 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

4.1.4. Data quality ...... 88 4.2. National projects and initiatives ...... 88 4.2.1. US Department of Transport ...... 88 4.2.2. Road Pilot - Austria ...... 91 4.2.3. Danish road directorate weather information system ...... 93 4.2.4. Finish Road Weather Information System ...... 98 4.2.5. French Road Weather Information System ...... 104 4.2.6. Greek Road Weather Information System ...... 109 4.2.7. Polish general directorate weather information system and motorways ...... 111 4.2.8. Ongoing projects ...... 115

Conclusion ...... 127 Lists of figures and tables ...... 129 References ...... 133 Publication data form ...... 145 Fiche bibliographique ...... 146

6 © Les collections de l’INRETS

Acknowledgments

We would like to thank COST for supporting the COST TU0702 network and this publication. COST – the European cooperation in the field of scientific and technical research – is the oldest and widest European intergovernmental network for cooperation in research. Established by the ministerial conference in November 1971, COST is presently used by the scientific communities of 35 European countries to cooperate in common research projects supported by national funds1. The funds provided by COST – less than 1 % of the total value of the projects – support the cooperation networks (COST actions) through which, with 30 million € per year, more than 30 000 European scientists are involved in research having a total value which exceeds 2 billion € per year. This is the financial worth of the European added value, which COST achieves. The main characteristics of COST are a “bottom up approach” (the initiative of launching a COST Action comes from the European scientists themselves), à la carte participation” (only countries interested in the Action participate), “equality of access” (participation is also open to the scientific communities of countries not belonging to the European union) and flexible structure” (easy implementation and light management of the research initiatives). As precursor of advanced multidisciplinary research COST has a very important role for the realisation of the European research area (ERA) anticipating and complementing the activities of the Framework programmes, constituting a “bridge” towards the scientific communities of emerging countries, increasing the mobility of researchers across Europe and fostering the establishment of “Networks of Excellence” in many key scientific domains such as: biomedicine and molecular biosciences ; food and agriculture ; forests, their products and services ; materials, physical and nanosciences ; chemistry and molecular sciences and technologies ; earth system science and environmental management ; information and communication technologies ; transport and urban development ; individuals, societies, cultures and health. It covers basic and more applied research and also addresses issues of pre-normative nature or of societal importance. This state of the art report summarizes the work done within the COST Action TU0702 “Real-time monitoring, surveillance and control of road networks under adverse weather conditions”. It gives an overview about the effects of inclement weather conditions on road traffic operations, pavement and road safety as well as the best practices which are available in various countries. Several projects carried out in the last years on this topic (especially in Austria, Denmark, Finland, France, Germany, Greece, Poland and USA) are described.

1 See www.cost.esf.org

© Les collections de l’INRETS 7 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

We hope that this state of the art report along with the results of recent researches in the field will be the starting point of a more in-depth understanding of the effects of adverse weather conditions on mobility of goods and peoples and pave the way for more comprehensive approaches accounting for those effects into traffic management to achieve weather-sensitive traffic management strategies. The editor thanks all authors and co-authors of the report, and also those members of the COST Action TU0702, who did not take part in the writing, but otherwise provided valuable work, ideas, and inspiration.

8 © Les collections de l’INRETS

Introduction

Adverse weather conditions can significantly influence traffic operations, traffic, Level of Service (LOS), and safety. Advanced technologies for collecting and archiving weather data are valuable adjuncts in the development of intelligent weather-based traffic management strategies and monitoring and control systems. In view of the paramount importance of having weather-based tools available, this project focuses on the development of techniques to improve road management and safety in adverse weather. More specifically, the action TU0702 will develop tools for achieving weather-sensitive traffic management and control. The project will also address issues related to road surface state. The planned models and estimators will support advanced control strategies that incorporate fusion of multiple sensor data and information. The action brings together researchers actively engaged in weather and road network management. It concentrates on mutually complementary methods for modeling, estimation and control that improve the safety and efficiency of traffic flow networks. Below is a heavily simplified and formal listing of some major weather parameters causing potentially adverse effects on road traffic and safety: - Precipitation, mainly rainfall: - Rainfall intensity (causing flooding), - Rainfall type (rain, snow, freezing rain, sleet), - Rainfall accumulation (causing flooding), - Precipitation start/end time; - Poor visibility: - Fog, - Dense/heavy snowfall, - Drifting snow (wind-blown snow), - Dust and smoke, - Air quality; - Air/road temperature: - Low temperature, - High temperature, - Drew point; - High wind speed and direction; - Humidity (formation of hoar frost); - Water level data;

© Les collections de l’INRETS 9 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

- Stream, river, lake, water levels near the road; - Tide levels for near the sea cost. One of the major activities of COST Action TU0702 is to consider “adverse weather conditions” in a global setting. The multi-disciplinary approach of the project prevents from ignoring any major aspects of adverse weather. For example, in many countries, very high temperatures can have a significant impact on road surface conditions and, consequently, on traffic safety and operations. Hence, even such a feature can be considered as adverse weather in the scope of the Action. So, the goal of Action TU0702 is to better understand the impacts of weather on freeways/motorways and urban highway network operations, and to develop, promote and implement strategies and tools to mitigate those impacts. The Cost Action TU0702 is divided in three working groups (key areas), addressing collaborative work at different levels: - WG 1. Modelling Weather Impact on Traffic: - State of the art, traffic and weather data needs and integration, - Traffic modelling, estimation and control under different weather conditions, - Decision Support System (DSS) for traffic monitoring and users’ information; - WG 2. Modelling Weather Impact on Road Surface and Pavement: - Tools for road surface monitoring, - Analysis of the weather impact on different types of pavement, - Tools to reduce the effect of adverse weather; - WG 3. Innovative Multisensor DF of Traffic and Weather Data: - Development of effective methods for data processing and filtering, - Development of optimal traffic state estimation. Below, a functional diagram of TU0702 action work program, describing different tasks in the scope of the Action.

10 © Les collections de l’INRETS Introduction

This report aims at summarizing activities by the different Working Groups concerning the state of the art about adverse weather impacts. It is structured in three main chapters. The impacts will be presented with respect to all these components. More precisely, the report addresses the following topics: - Definition of what is meant by “adverse weather”: a clarification is presented of the term “adverse” along with some other terms to avoid misleading or contrasting interpretations. - Impact of adverse weather on drivers’ behaviours and traffic operations: drivers will react and change their driving behaviour as adverse weather conditions affect the road surface with consequences on safety. These microscopic effects will reflect on a more aggregated level (macroscopic) with a significant impact of adverse weather on traffic capacity, speed, and density. The goal of this part is to present at all levels of analysis the existing studies that quantify those effects. This review represents the first step of the further activities of WG1/WG3 dealing with weather impact modelling and simulation. - Impact of adverse weather on safety: this part consists in assessing effects of inclement weather on safety. It provides a systematic review of the literature.

© Les collections de l’INRETS 11 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

- Impact of adverse weather on pavement: this chapter synthesized the key aspects about the changes in road surface conditions and pavement under adverse weather conditions. - Road Weather Information Systems (RWIS): A road weather information system is a combination of technologies that collects, transmits, models, and disseminates weather and road condition information. The current RWIS technologies (Road Weather Monitoring Sites, Parameters, Data requirements, specification) will be presented. - National Projects: The projects carried out within European, national and regional projects will be introduced. The purpose is to gather relevant results and practices dealing with a better comprehension of the weather effects on traffic and its integration into traffic management strategies. This report is the result of an effort of many researchers who have devoted much effort to compile a large body of scientific production in order to obtain a comprehensive synthesis on the State of Art. It is important to notice that COST does not support project activities. Therefore all contributions concerning individual countries are provided according to available time and engagements of the participants. Hence the structure and details of each contribution mar vary, and the reader will not necessarily find the same information for all countries. A lot of references are given, however, where further information can be retrieved.

12 © Les collections de l’INRETS

1. Definition and terminology associated with the adverse weather

The need for mitigation against harmful or even disastrous consequences caused by adverse weather (e.g. hurricanes, snow storms) is intuitively obvious. However, the conception or the meaning of the word “adverse” itself, which is explicitly adopted in the title of Action TU0702, is not necessarily obvious or generally and uniquely understood. Differences in perceptions and definitions have emerged over the course of Action TU0702; it therefore calls for initial definition and interpretation. Since Action TU0702 covers interdisciplinary research activities and joins together researchers from different disciplines such definitions are even more essential to be able to “speak in the same language”. It is appropriate to extend weather related terminology definitions to relevant terms of “adverse”, which occasionally have a somewhat similar meaning but in cases a notably different objective. Such terms can address the severity, extremity, harshness, rarity, or impacts of weather events quite differently and from different perspectives. Some of the most common weather parameters relating to adverse weather were introduced in the previous chapter. The majority of studies discussing the adverse effects of weather on traffic focus mainly on rain, snow and fog. Some papers listed in the bibliography, mostly from the traffic or transportation engineering discipline, contain the term “adverse” or “inclement” in the title and deal essentially with rainy or snowy conditions. A more recent paper (see Khattak and De Palma, 1997) includes sun glare and darkness as “adverse”. An analogy from the construction industry2 links together as synonymous “adverse and “inclement” weather. In the meteorological community, however, the word “inclement” is scarcely used. In this introductory context, relating to traffic, “adverse” could be generally defined as “atmospheric conditions at a specific time and place that are unfavourable to optimal road traffic operations“. A clear differentiation must then be made between common, recurrent weather events (rain, snow, wind etc.) and severe, extreme, rare weather events (e.g. snow storms, hurricanes, tornadoes), which can all have high impacts on road traffic and transportation, but not necessarily always and by definition. A strong meteorological multi-dimensionality is characteristic of all of the following terms that are introduced and depicted here:

2 http://www.usasoft.com/whtpaper.htm

© Les collections de l’INRETS 13 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

- severe, extreme, high-impact, rare. “Severe” is defined in dictionaries/thesauruses as an adjective meaning “rigorous; violent; very strict; unsparing; hard to endure; and inflicting (physical) discomfort”. Severe weather events can cause large losses of human lives or money or have dramatic effects on the economy and environment. Severe events and their severity can in principal be estimated by analysing expected long-term loss (risk). The probability of the event would need to be known as well as the level or extent of exposure (e.g. number of exposed people) to the event. Severe event (severity) is a function of both the meteorological event itself and the state of human affairs affected by the event. In the traffic sector, increased traffic volumes naturally lead to increased exposure to meteorological features affecting the traffic. “Extreme” is defined in dictionaries/thesauruses as an adjective “exceeding the ordinary, usual or expected; having highest limit or degree; or being outermost; greatest; stringent; or very violent”. Extreme events typically cause large damage or consequences to the society. Flash flooding may be due to extremely heavy local rainfall, whereas flooding with large spatial extent often relates to large amounts of rain during an extended time. Malfunctioning within various transport sectors and falling down of electricity lines may result from heavy snowfall and/or storm force winds. Notorious health effects (especially for the elderly) are due to extremely high temperatures, whereas very low temperatures will increase energy consumption dramatically. An extreme event can be generalized as having an all-time maximum value and/or exceeding a pre- existing, measured high (or low) threshold. By definition, extreme events are generally also rare events, e.g. having less than 5 % probability of occurrence during a given year at a given location. “High-impact” is defined in dictionaries/thesauruses as a noun “having shocking or striking effect or influence”. High-impact meteorological events are typically also severe events. It is quite customary to make a separation based on time scales, between short-duration and long-duration weather events. Examples of high-impact events with a relatively short life cycle are rapidly moving strong cyclones, or convection-induced (strong vertical motion of air) heavy precipitation cells, or overnight freezing of the road surface due to the cooling of the atmospheric surface layer as a result of outgoing long-wave radiation. An example of a long-duration weather event is a spatially large blocking high pressure area, which can give rise to a prolonged heat wave and resulting draught. The monsoon circulation is another example of a high-impact event with a long life cycle. It may be interesting to note that the World Meteorological Organization (WMO) prefers the usage of the term “high-impact” against “severe” to cover events of all time scales. “Rare” is defined in dictionaries/thesauruses as an adjective meaning “uncommon; unusual; infrequent; seldom occurring or found”. Rare events have a low probability of occurrence (e.g. 0.1 %). Because the society and the environment are not typically adapted to rare events they may cause large damages when occurring. Hence, and despite rarity, high vulnerability can lead to large losses.

14 © Les collections de l’INRETS Effects of weather on traffic and safety

Meteorological parameters potentially causing adverse effects on road traffic have already been listed: temperature, rainfall, snowfall, wind speed, fog, etc. Although being continuous in nature (like temperature) these parameters, when analysed, take often only binary or dichotomous values, i.e. the events are taken either to happen or not; there is precipitation or there isn’t, there is fog or there isn’t. However, they all can also be considered as multi-category events: e.g. rainfall can be stratified in several mutually exhaustive categories, where one of the categories might be a “rare event category”, and likewise for wind speed, temperature etc. The following set of figures show, in this case for temperature, how one can consider and interpret the terminology defined above in the context of real meteorological data. Figure 1. Monthly average temperature at Davos, Switzerland (1866 to 2009) 20 Hot Scarce Events Hot Rare Events 15 Hot Extreme Events

10

5

0 Highgh ImpImpactImmpampacactct eveevent "Freezing RoadRoRoaoadadd SurfaSurface"ce"

-5 Temperature (°C) Temperature -10

-15 Cold Extreme Events Cold Rare Events -20 Cold Scarce Events 1860 1880 1900 1920 1940 1960 1980 2000

Year Figure 1 shows time-series of observed temperatures at a weather station at Davos, Switzerland from 1866 to 2009. The temperature varies from -15 °C in winter to +18 °C in summer. The graphs illustrate the different events (rare, extreme and scarce). Figure 2 presents the temperature trend in Switzerland in terms of annual variation of temperature from the longtime average (1961-1990 standard). The years when it was too hot are shown in red and those when it was too cold in blue. This is as an impressive example of climate change. The linear temperature trend between 1864 and 2004 is +1.1 degrees Celsius per 100 years, so the global warming 1864-2005 amounts to +1.5 degrees Celsius. The linear

© Les collections de l’INRETS 15 Real-time monitoring, surveillance and control of road networks under adverse weather conditions temperature trend for the 20th century (1901-2000) is +1.4 degrees Celsius per 100 years, and this corresponds to the global warming of the 20th century. Figure 2 is then another way of visualizing the features discussed above. The histogram shows the number of observed cases in relation to the temperature scale. Figure 2. The mean annual temperature anomalies in Switzerland

Source: www.meteoswiss.admin.ch

The approach presented here to interpret the meaning and conception of terminology associated with adverse weather events could quite easily be extended from temperature to other meteorological variables as long as there are extensive weather observation datasets available. It would also be interesting to study similar distributions and implications to interpretation in other environments and countries. These might be considered as future activities of the Action. A comprehensive essay on definitions of weather related terminology is given by Stephenson (2008).

16 © Les collections de l’INRETS Effects of weather on traffic and safety

Figure 3 shows a 50-year time-series of observed temperatures at a weather station in Northern Finland. This station is quite notorious for its extremely cold wintertime conditions. Figure 3. 50-year time-series of temperature observations at a climatologically cold weather station in Northern Finland 3.a. Temperature distribution at a cold station in Finland (50-year time-serie, c. 55000 observations)

3.b. Temperature distribution at a cold station in Finland (50-year time-serie, c. 55000 observations)

© Les collections de l’INRETS 17 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

3.c. Temperature distribution at a cold station in Finland (50-year time-serie, c. 55000 observations)

3.d. Temperature distribution at a cold station in Finland (50-year time-serie, c. 55000 observations)

The Figure 3.a. shows the entire dataset with each dot indicating a single observed value. Each vertical line of dots represents one calendar year with three temperature observations per day, summing up to c. 55 000 observations. The Figure 3.a. shows the dataset from which one can immediately realize the large temperature variations at this location; summertime temperatures can exceed +30 oC, and during winter the temperature can fall below -45 oC. There are a few

18 © Les collections de l’INRETS Effects of weather on traffic and safety extreme cases during the 50 years when these high and low values have been reached (Figure 3.b.). One can conclude that here “extreme” and “rare” can be taken as synonyms. Pre-defined high/low temperature values of +30/-45 °C are chosen as representing hot/cold extreme events in the Figure 3.b. – to highlight these extreme cases the rest of the dataset is “dimmed”. Extremity can also be defined statistically like in the case of the Figure 3.c. Here the 5 % quantiles at the warm and cold ends of the distribution are seen to represent temperatures of +18 oC and -23 oC, respectively. “Scarce” would probably be a more appropriate term than “rare” here. The cases are by no means severe or extreme but then again e.g. to a Southern European -23 oC might seem (and feel) as extreme and severe. The lower right Figure demonstrates how “high-impact” weather events may be present without necessarily having anything to do with severity, extremity or rarity of the event. The temperature interval of plus/minus one degree centigrade can be highly notorious in leading to freezing of a road surface. However, it is a very common feature in the wintertime weather of Finland and other Nordic countries. In the Figure 3.c. the extremity is statistically (rather than subjectively) pre- defined by assuming the hot/cold thresholds at 5 % probabilities, 90 % of the observations are “dimmed” here. The Figure 3.d. demonstrates how common the plus/minus one degree temperature interval, prone to freezing of the road surface, is at this particular station, with thousands of cases throughout the years. Figure 4. Histogram of temperature observations at a climatologically cold weather station in Northern Finland

Figure 4 provides a histogram of the same dataset. The extreme high/low temperature values of +30/-45 °C and the 90 % quantile are highlighted, like is the plus/minus one degree temperature interval.

© Les collections de l’INRETS 19

2. Effects of weather on traffic and safety

2.1. Impact of weather on traffic operations and Drivers’ behaviours

Introduction As early as the 50s, researches about the effects of weather on traffic were launched by Tanner (see Tanner, 1952). Next, further to the seminal study of Jones et al. (1970), a certain amount of articles enables the elaboration of a state of the art about the effects of weather on traffic. However, the comparison between empirical studies is not so easy because of the heterogeneity of the data sources (type of traffic and weather data), sections (e.g. urban, inter-urban freeways, highways) and of course analysed parameters (e.g. occupancy, density etc.). In spite of these constraints, we will see that a common trend exists and the main effects of adverse weather on drivers’ behaviours are well known even if there is a need to deepen the studies (wide range of precipitations intensities, consensus about effects quantification etc.). The associated state of the art was divided into three parts according to the type of data and the levels of analysis. First, the majority of the analyses use data provided by inductive double loop detectors (ILDS) or cameras. Then, traffic data is crossed with weather data provided by weather station or RWIS. The goal here is to quantify the impact of adverse weather on traffic through data analysis from these two paired data sources. Within this task, one could highlight two approaches according to the level of analysis. Thus, the impact of weather can be analysed in two ways: - Disaggregated level: at this level, the impact of weather is carried out on microscopic data and effect of weather on individual speeds, time and spacing headways, is addressed. One could extend this level of analysis to a mesoscopic one: the impact of adverse weather on platoon phenomenon (see Billot et al., 2009). - Aggregated level: at this level, also called macroscopic level, the effect of (adverse) weather is performed on traffic at a more aggregated level. First, the changes in traffic demand under adverse weather conditions are evaluated. Second, the effects of weather on the main macroscopic traffic variables, which form the fundamental diagram, are quantified (speed, capacity, density..). This is the first step toward the integration of weather effects into traffic modelling. On the other hand, relevant experiments have been carried out through driving simulators. A review of some projects and studies related to the impact of weather on driver using driving simulators can be found in intro, 2005.

© Les collections de l’INRETS 21 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

2.1.1. Impact on macroscopic traffic characteristics Weather conditions significantly affect the travel decisions as well as mode choice. It is quite evident that during adverse weather, for example rain, road users may choose to change means of transport, delay or even postpone their trips. Hence, the impact of weather conditions on traffic characteristics is multidimensional where the three main dimensions are: - Traffic demand – in case of adverse weather conditions part on planed trips may be postponed, deferred or even cancelled, see Khattak and de Palma (1997) and Button and El Faouzi, (2010); - Traffic level of service (LOS) characterised by speed change, capacity change and flow-density of flow-speed relationships; - And traffic safety-crash rate and severity can increase dramatically during adverse weather, lighting and visibility. These three main dimensions are examined in the next sections. 2.1.1.1. Traffic demand Adverse weather has a direct impact on the traffic demand. It affects the behaviour of drivers with respect to the route choice, mode choice and departure time. The effects are partially dependent on the type of trip (e.g., professional/constrained trips versus recreational trips). Indeed, travellers are more likely to postpone long-distance trips and professional/mandatory trips are least likely to be deferred. Several studies found traffic demand to decrease during adverse weather. Chung et al. (2005) aims at investigating the effects of rainfall on travel demand in Japan. The analyses show that travel demand decreases for rainy days, and especially during the weekend, when the sensitivity to rainfall is higher (Figure 5). In Khattak and de Palma (1997) the study aims at understanding traveller behaviour under normal and unexpected travel conditions in real-life situations. The results of a comprehensive behavioural survey conducted in Brussels are reported. Among those who changed their travel decisions in adverse weather (about 50 %), more than one-quarter reported that adverse weather was either very important or important in changing their mode; and 60 % changed their departure time due to adverse weather whereas 35 % diverted to alternate routes. Furthermore, close to 75 % kept themselves informed about weather through secondary information sources such as radio and television.

22 © Les collections de l’INRETS Effects of weather on traffic and safety

Figure 5. Percentage decrease in daily trips during rainy days by day group

Hanbali and Kuemmel (1992) have observed that volume reductions range from 7 % to 56 % depending on the intensity of snow fall, time of the day, day of the week, and roadway type. On Interstate highway (I-35) in northern rural Iowa Maze et al. (2005) found that during snowstorms, commercial vehicles became a higher percentage of the traffic stream (by as much as 38 % to 70 %) than their typical proportion during clear weather. This is likely due to the fact that commercial vehicle operators were much less likely to divert trips due to inclement conditions than motorists. In general, empirical studies carried out in the USA (see Unrau and Andrey, 2006) showed that adverse weather may: - Reduce demand for travelling, as drivers postpone discretionary trips or activities get cancelled; - Increase demand for travelling through mode shift (those who travel on foot will have to change temporally their travelling mode in the occurrence of bad weather conditions); - Have more complex effects on demand (for example shifts in peak-hour demand as drivers reschedule travels to avoid driving in dangerous conditions). In Finland, Sihvola (2008) has studied the driver assessment of road weather conditions and road weather information. The investigations covered following road weather condition related issues: - Do the drivers receive/search for weather information forecasts? - Do the drivers change their travel plans or driving behaviour because of road weather conditions? - How do the drivers assess different road weather conditions?

© Les collections de l’INRETS 23 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

- and how do these assessments relate to weather forecasts? The data was collected during the winter 2007-2008 via interviews at service stations (76 %) and on the roadside (24 %). The interviews were aimed at being carried out mostly during bad or very bad driving conditions. The results showed, that in total 62 % of the drivers had received or searched for information on road conditions and weather during their journey or beforehand. Every fifth respondent indicated that they had changed or considered changing their travel plans due to road weather conditions. The drivers were more likely to have acquired information on road weather if they were less experienced, had driven for a longer time before the interview or were on a trip they did not make frequently. The results also showed that in general the drivers’ assessment on the road slipperiness did not match with the information from road weather stations. The most dangerous situations were those in which the automatic road weather stations indicated that the road was slippery, but the respondent had not detected it. The results highlight the importance of informing the drivers about road weather conditions with real-time warning services. Moreover, the safety on roads can be increased by using weather-controlled speed limits and displays. 2.1.1.2. Speed and travel time The effects of adverse weather on flow speeds have been reported in many studies. According to the Highway Capacity Manual, (Hall and Barow, 1988), on freeways, light rain or snow can reduce average speed by 3 to 13 %. Heavy rain can decrease average speed by 3 to 16 %. In heavy snow, average freeway speeds can decline by 5 to 40 %. Low visibility can cause speed reductions of 10 to 12 %. Free-flow speed can be reduced by 2 to 13 % in light rain and by 6 to 17 % in heavy rain. Snow can cause free-flow speed to decrease by 5 to 64 %. Speed variance can fall by 25 % during rain. Light rain can decrease freeway capacity by 4 to 11 % and heavy rain can cause capacity reductions of 10 to 30 % (see Dinkel et al., 2008). Studies from the EU Intelligent Roads project (INTRO3, 2005) have highlighted that the main factor which is most affected by weather is speed. The next two figures show speed changes for Swiss and Polish data. For Switzerland, data comes from a typical interurban highway with 3 or 4 whereas Polish data come from a 2 lanes around Warsaw (narrower lanes, more curves). The average speed under fine weather condition is higher than under rain or snow conditions. In reality, snow makes people more careful while controlling their vehicles and the reduction of speed in this condition is thus greater.

3 http://intro.fehrl.org

24 © Les collections de l’INRETS Effects of weather on traffic and safety

Figure 6. Speed reduction quantification - 1, site 149

Figure 7. Speed reduction quantification - lane 2, site 56

© Les collections de l’INRETS 25 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Ibrahim and Hall (1984) have studied the weather impact on highway traffic in Canada. Their conclusions were the following: - A light rain leads to a 2 km/h speed drop; - Light snow fall leads to a 3 km/h speed drop; - Heavy rain: 5-10 km/h speed drop; - Heavy snow: 38-50 km/h speed. May (1998) highlights a free speed reduction from 120 km/h to 110 km/h in highway under light rainy weather conditions, 100 km/h with heavy rain, 70 km/h with heavy snow. Perrin and Martin (2002) have obtained the following results: Table 1. Influence of weather on speed Weather Speed reduction Dry weather 0 % Rain 10 % Snowfall and slippery road 13 % Snow and slippery road 25 % Snowy road 30 % Heavy snowfall 36 %

Regarding speed and safety, most studies underline the fact that the state of the pavement plays a key role. Thus, Brilon and Ponzlet (1996) studied the speed fluctuations on German highways. A slippery road goes together with a 9.5 km/h speed drop on a two-lane section and with a 12 km/h drop on a three-lane section. This trend is confirmed by another study (see Hogema, 1996). As a conclusion, one can say that there is a clear consensus of around 9 -10 km/h speed drop under light rainy conditions (Table 2).

26 © Les collections de l’INRETS Effects of weather on traffic and safety

Table 2. Main findings on speed reduction under inclement weather conditions Type of Speed Country Section reduction

France Interurban area 8 - 12.5 %

France Urban ring road 9 %

Canada Urban freeway 10 %

USA 3 metropolitan areas 6 - 9 %

Japan Metropolitan expressway 5 %

Switzerland Expressway 2 - 6 %

Poland Expressway 5 - 13 %

Moreover, other studies deal with the importance of the wind speed, which is a key parameter to be taken into account in such studies (Edwards, 1999; Kyte et al., 2007). The effects of adverse weather on speeds, capacities, will directly reflect on a key traffic performance indicator travel time. In Chung et al. (2005), analysis of travel time under similar traffic density revealed that travel time is higher for high density (i.e. high traffic flow) and not significant at low density (i.e. low traffic flow) for rainy period. Taking other factors into account such as crash rate, free flow speed and lane capacity during rainy days, higher travel time can be expected especially for higher traffic flow condition. According to Chung et al. (2005), in off-peak hours, travel time increases by 12 % under inclement weather conditions. In case of slippery road and snow, it increases by respectively 5 % and 23 %. Under winter storm conditions (heavy snow), travel times can increase by 50 %. Other studies reveal that travel times can increase by 24 % in average due to precipitations. 2.1.1.3. Flow, Density and Capacity Several studies have demonstrated a reduction of flow and a capacity under adverse weather conditions. This drop in the capacity can be highly context- sensitive, as the next table shows:

© Les collections de l’INRETS 27 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Table 3. Capacity reduction under light rain Type of Capacity Country Section reduction

France Interurban area 18.5 - 21 %

France Urban ring road 15.5 %

USA 3 metropolitan areas 10 – 11 %

Japan Metropolitan expressway 6 – 9 %

Thus, the types of the sections and the regional differences need to be taken into account in further studies. However, the key idea is that there is a consensus about both speed and capacity reductions under adverse weather conditions, which parameters are of paramount importance for real-time monitoring. Among other studies, Hanbali and Kuemmel (1992) showed that freeway volume reductions increased with total snowfall, but that the reductions were smaller during peak travel hours and on weekdays, likely due to the nondiscretionary nature of most weekday trips. Heavy snow was also found to decrease capacity by 30 %, and light snow was found to decrease flows by five to 10 % (see. Ibrahim and Hall, 1994). A recent study on different types of highways provided evidence that cold and snow impacts on volume vary with day of the week, hour of the day and the type of highway; commuter roads experience lower reductions in traffic volume due to cold when compared to recreational roads (Datla and Sharma, 2008; Hassan and Baker, 1999). In the same study it was found that cold on off-peak hours affects volume more greatly than on peak hours, whereas the opposite applies to recreational roads. Moreover, snow is always determinant of volume reduction regardless of highway type. Strong correlation between the percentage reduction in traffic volume and, wind speed and visibility during snowy days has been also reported in freeway interstates (Maze et al., 2006). Weather effects on mobility include increased delay and congestion, lower traffic volumes and speeds, increased speed variance, and reduced roadway capacity. Interestingly, no impacts were found on traffic stream jam density, but both rain and snow did impact traffic free-flow speed, speed-at-capacity and capacity and parameters varied with precipitation intensity (Rakha et al., 2008). As Billot et al. (2009) advocate, this fact is consistent with physical considerations since the maximum number of vehicles to be accommodated by a roadway section (Jam density) is not weather-sensitive. Knapp and al. (2000) demonstrated a large variability in winter storm traffic volume impacts, ranging from 16 % to 47 % reduction. Ibrahim and Hall (1994) showed that flow reduces by 10 to 20 % during conditions of heavy rain.

28 © Les collections de l’INRETS Effects of weather on traffic and safety

The Figure 8 presents the speed flow diagram from real data from Swiss motorways under three weather conditions: fine weather, rain and snow. The diagram clearly indicates the capacity drop due to rain and snow. Figure 8. Speed-Flow curve for 3 weather conditions

The Table 4 summarizes the average impact of weather on freeway capacity and speed observed at Interstate highway (I-35) in northern rural Iowa by Maze et al. (2006). Around 2 to 22 % drop in capacity and 2 to 13 % in speed reduction is observed.

© Les collections de l’INRETS 29 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Table 4. The average impact of weather on freeways capacity and speed

2.1.1.4. Weather effects on arterial traffic and signal timing Urban traffic activity may be influenced by unseasonable or extreme weather; a traffic reduction up to 5 % under extreme weather conditions but a reduction of 10–15 % in traffic activity has been reported. Keay and Simmonds (2005) reported traffic reductions of traffic volume on urban arterials in Melbourne during wet days by 1.3 % in winter and 2.1 % in spring. Adverse weather can also have an impact on the effectiveness of traffic signal timing plans. Goodwin (2002) and Pisano and Goodwin (2004) reviewed literature concerning the impacts on adverse weather to signalized arterials. Results include the following: - Arterial flow rates were found to be between 6 % and 30 % lower than those under normal conditions, depending on road weather conditions and time of day. - Speed reductions under inclement conditions ranged from 10 % to 25 % in rainy, wet pavement conditions and from 30 % to 40% with snowfall and snowy/slushy pavement.

30 © Les collections de l’INRETS Effects of weather on traffic and safety

- Travel time delay on arterial roads increased by 11 % in wet pavement conditions, and by more than 12 % in the presence of precipitation, high winds, low visibility, or slick pavement. - Under coordinated traffic signal conditions, start-up delay was 5 % higher in wet pavement conditions, 23 % higher in snowy pavement conditions, and 50 % higher during snowstorms. The number of vehicle stops in adverse weather was 14 % higher than stops in normal conditions. - Traffic operations on arterial roadways can be improved in inclement conditions by utilizing weather-related signal timing plans, which accommodate changes in driver behaviour. Pisano and Goodwin (2004) underline that weather-related delay can be mitigated by implementing special signal timing plans that account for slower travel speeds and lower traffic flow rates or volumes. The signal timing plans for inclement weather conditions can easily be developed by modifying the data used to create standard “dry” signal timing plans and by changing a small number of other parameters. Martin et al. (2000) has recommended that signal timing plans be manually “switched on” by a trained operator or engineer once an alarm identifies the potential need. The trigger for the alarm may include observed reductions in travel speeds by mid-block detectors or reduced saturation flows from stop bar detectors. Four general criteria must be considered to ensure that such a plan gives the maximum benefit. First, the storm must be sufficiently severe to cause “inclement” road surface conditions. Second, the storm duration must be predicted to continue to cause inclement surface conditions for at least 20 minutes, to allow for the negative effects of transitioning from one signal timing plan to another to be minimized. Third, the storm must affect a sufficient length of corridor to benefit the majority. Thus, many microclimate events that only affect a small geographical area should not initiate the new timing plans. Finally, the traffic volumes must be substantial enough to warrant the time to switch a plan. A.m. and p.m. peak hours will be the most likely candidates, while off-peak plans on specific corridors should be individually considered. Special attention should be given to individual intersections that have high speed or steeply graded approaches, as they will cause additional problems in low traction conditions.

2.1.2. Impacts on microscopic traffic characteristics 2.1.2.1. Individual speeds It is widely accepted that inclement weather can impact the traffic by reducing speeds. However, according to the intensity of rain or the type of lane, the speed decrease can be more or less significant. In El Faouzi et al. (2008), the speed distributions on the slow lane have been compared. The data was collected on the A9 Motorway (2x3 lanes) between Orange and Montpellier cities in the south of France. It comes out that for passengers’ cars the frequencies of speeds under 120 km/h (75 mph) are higher under adverse weather conditions. Indeed, the frequencies move from 55 % to 62 %. These results show that passengers’ car drivers reduce in majority their speed under rainy weather conditions which is

© Les collections de l’INRETS 31 Real-time monitoring, surveillance and control of road networks under adverse weather conditions consistent with the French legal speed limit regulation on motorways: 130 km/h (81 mph) under dry conditions and 110 km/h (68 mph) under degraded conditions (see Figure 9). Figure 9. Legal speed limit on French motorways according to weather conditions

Source: www.news-assurances.com These results were supported by another recent study (see Billot et al., 2009) carried out on an interurban freeway section (Figure 10). Figure 10. Speed distribution on the slow lane (French data)

Thus, on the slow lane, a clear decrease of the frequencies of speeds > 110 km/h under rainy conditions was observed whereas the frequency of speeds between 70 and 90 km/h is higher under light rain and medium rain conditions (e.g. 53 % under medium rain conditions versus 35 % under dry conditions).

32 © Les collections de l’INRETS Effects of weather on traffic and safety

2.1.2.2. Time headways and spacing In 2005, the EU INTRO (Intelligent Roads) project had undertaken a task about the integration of the weather effects for traffic indicator forecasting (Karkowski et al., 2006). Data from Swiss and polish data were used to quantify the relationship between weather and traffic parameters. Regarding Time headways, the conclusion was the following: During fine weather conditions, more drivers maintain short headway partly due to better visibility and shorter breaking distance of vehicle (i.e. higher skid resistance). However during inclement weather, drivers maintain slightly longer headways, taking into account the visibility and road surface condition (Figure 11). Figure 11. Headway distribution of free flow under three weather conditions in 2005, lane 8, and site 226 (Switzerland)

Source: INTRO project.

This trend was confirmed in a more recent study (Billot et al., 2008). There is a drop of the short time headway under rainy conditions. The higher the intensity of rain is, the higher the drop is. In terms of frequency and regarding the short time headway on the slow lane, a drop of 12.1 % of the time headway less than 2 seconds is observed under rainy conditions (light rain). In medium rain conditions, a sharper decrease of more than 18 % is observed. This drop is reported on a rise of the time headway between 2 sec and 10 sec. Alternatively, the effect of rain on time headway distribution was assessed using a density histogram (Billot et al., 2008). Figure 12 shows in simultaneously the histogram of the time headway under fine weather conditions versus medium rain conditions. One can notice that the increase of short time headways under adverse weather conditions goes together with a distance headway increase, as previous studies demonstrated. Perrin and Martin. (2002) underline that longer stopping

© Les collections de l’INRETS 33 Real-time monitoring, surveillance and control of road networks under adverse weather conditions distances are needed during the inclement weather because of changes in traction, whether real or perceived. Figure 12. Time headway histogram (France)

In El Faouzi et al. (2008), a significant decrease of the frequency of short spacing under rainy conditions has been shown. Generally speaking, a fall of more than 6 % of the spacing less than 50 meters (54.6 yards) was noted. This trend hardly comes as a surprise. Without extrapolating, we can infer that drivers feel less secure under adverse weather conditions, so they reduce their speeds and de facto increase time and distance headways. 2.1.2.3. Lane selection and other effects Comparison of vehicles’ distribution over different types of lanes under fine weather and rainy conditions was also investigated. Most of the studies were carried out on interurban motorways or freeways. It comes out that there is no consensus about the impact of adverse weather on the vehicles distribution between the different lanes. For instance, the following figures represent the lane distribution on a two-lane Swiss motorway for different weather conditions (fine, rain and snow). It can be noted that, when flow is low then most of the flow in on slow lane, and with increase in flow the proportion of the flow on fast lane increases.

34 © Les collections de l’INRETS Effects of weather on traffic and safety

Figure 13. Lane distribution on two lanes Swiss Motorway

Lane distribution under rain conditions 100 Fast Lane Slow Lane 90

80

70

60

50

40 Percentage (%) Percentage

30

20

10

0 0 500 1000 1500 2000 2500 3000 3500 4000 4500 Flow over two lanes (vph)

© Les collections de l’INRETS 35 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Lane distribution under snow conditions 100 Fast Lane Slow Lane 90

80

70

60

50

40 Percentage (%) Percentage

30

20

10

0 0 500 1000 1500 2000 2500 3000 3500 4000 Flow over two lanes (vph)

By differentiating at each time the type of vehicles (Passenger car vs. Heavy Good Vehicle - HGV), El Faouzi et al. (2008) has shown that medium rain has an impact on lane selection. A significant proportion of HGVs (4.2 %) uses the medium lane during dry weather conditions transferred onto the slow lane under adverse weather conditions. This fact is rather logical: HGVs are more sensitive to a slippery road and HGV drivers (more experienced than the other class of drivers) are aware about their driving performances and necessary adjustments to the weather conditions. There is a need to deepen the research in order to highlight some trends. Another subject of interest is the influence of weather on platooning phenomenon. This topic was investigated by Billot et al. (2009). The authors underlined the fact that the definition of a platoon under adverse weather condition is itself weather-dependent. Indeed, the definition of platooning vehicles is usually based on time headway or spacing thresholding, both are weather sensitive (see Figure 12). Finally, one can mention the study of Knoblauch et al. (1996) regarding characteristics during inclement weather. The study finds that as the severity of the weather increases, the walking speed of the pedestrians also increases. 2.2. Impact of weather on road safety Adverse weather has a well-known impact on safety with an increase of crash rates and severity. Several studies included or focused on the effects of weather conditions on road accidents occurrence and severity, attempting to capture these often complex effects. A thorough review of mostly earlier studies on weather effects can be found in Eisenberg (2004). Weather condition data relevant to road accidents are typically recorded at the accident scene as the prevailing conditions during the accident. In several studies, weather conditions during the accident are associated with the accident

36 © Les collections de l’INRETS Effects of weather on traffic and safety outcomes usually through the calculation of casualty risk ratios using a control group (Ivey et al., 1981; Majdzadeh et al., 2008), and the results indicate increased casualty risks in adverse weather conditions. In several cases, particular groups of drivers such as older drivers (Baker et al., 2003), motorcyclists (Pai and Saleh, 2008) and heavy goods vehicles drivers (Young and Liesman, 2007) are examined. Golob and Recker (2004) found unique profiles in terms of the type of accidents that are most likely to occur in different weather and traffic conditions. Weather conditions are believed to influence various aspects that can affect road safety such as travel decisions, transport mode choice, visibility and braking and vehicle controlling, driving styles and other behavioural aspects and so on (see. Bijleveld and Churchill, 2009). A critical road user group whose decisions to travel are significantly affected by weather conditions are the powered two wheelers (PTWs). Weather has been reported to be a less influential factor in 98 % of motorcycle accidents in a research conducted in California (Hurt et al., 1981). In MAIDS report dealing with PTWs safety, weather made no contribution to accident causation in 92.7 % of MAIDS cases (854 cases) and was the precipitating event in 7 cases (0.8 % of all cases); weather conditions at the time of the accident were most frequently dry (90 %). Rain at the time of the accident was noted in 8 % of all cases, whereas dry and free of contamination in 85 % of all accidents. During 1999-2003, more than 80 % of crashes involving the death of a motorcyclist in Australian roads were reported under fine weather conditions (Johnston et al., 2008). Riding under fine weather also appears to result in more severe injuries regardless of what control measure was employed (Pai and Saleh, 2007). Moreover, the driving styles and riders/drivers behaviour are directly related to weather conditions as reflected in sudden speed drops and gap acceptance limits, especially during rainy weather (Edwards, 1998), (Hogema, 1996). Regarding road users’ behaviour, Edwards (1998) underlines that most people perceive the weather to be less determinant to driving unless conditions become so severe that the journey becomes impractical, rather than cancelling the trip merely because it is ‘less safe’. As a reaction to the perceived risk due to weather conditions, riders/drivers adopt lower speeds and increased following distances. The relation between weather and road safety is not always straightforward especially due to recent ADAS devices, such as anti-lock brakes, EPS and traction control that enhance vehicle’s behaviour and control under extreme weather phenomena. Enhanced vehicle behaviour may reduce accident risk in adverse weather conditions due to vehicle’s improved operational characteristics, but may increase risky behaviour, as users may feel more confident when driving vehicles equipped with these safety features. Weather conditions, such as air temperature and precipitation, are associated with considerable impacts on road safety, mainly through their influence on both the exposure and the behaviour of road users. The interaction between weather effects and the effects of other road safety factors, including roadway, driver, vehicle and intervention variables on road accident frequency is certainly a complex phenomenon that attracts increasing attention by researchers. Stipdonk

© Les collections de l’INRETS 37 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

(2008) underlines that weather effects need to be controlled for in any multivariate analysis aiming to explain changes in road safety outcomes. Koetse and Rietveld (2009) further emphasize this need within the climate change context. In the first case, the spatial distribution of road accident counts is associated with meteorological phenomena. Edwards (1996) and Khan et al. (2008) showed that the occurrence of road accidents in hazardous weather conditions (rainfall, fog, snowfall and wind) broadly follows the regional weather patterns for those conditions. Geurts et al. (2005) reported a significant spatial association of road accidents at hazardous locations with rainfall. Regarding accident severity, a positive spatial effect of fog in rural areas and a negative overall spatial effect of rainfall were identified (see Edwards, 1998). Valverde and Jovanis (2006) introduced hierarchical models with random spatial and time effects and found that rainfall may increase road accident frequencies. However, a large part of existing research has involved time series data that may capture both global and seasonal effects. These studies are summarized in Table 5. They range from yearly to daily analyses and from national to local level, while they use approaches ranging from generalized linear modeling techniques (i.e. Poisson-family models) to advanced, dedicated time series analysis techniques. Moreover, several additional variables are often controlled for, such as exposure, roadway design, demographics and interventions. Higher temperatures appear to have a decreasing effect on accident frequencies and severity both at daily, weekly and monthly bases (Scott, 1986); (Brijs et al., 2008). The hours of sunlight appear to increase road accidents (Hermans et al., 2006) ; (Brijs et al., 2007), while deviations from mean daily or monthly temperatures were found to increase road accidents (Brijs et al., 2008); (Stipdonk, 2008). Malyshkina et al. (2008) found that extreme temperatures (both low during winter and high during summer) are positively correlated with road accidents; on the other hand, when the monthly number of days with temperature below zero increases, road accidents are reduced possibly due to reduced exposure (see Hermans et al., 2006; Stipdonk, 2008). Findings regarding rainfall are extensive and quite consistent. Increased daily, monthly or even yearly rainfall appears to increase accident frequencies (Fridstrom and Ingebrigtsen, 1991); (Fridstrom et al., 2005); (Chang and Chen, 2005) ; (Caliendo et al., 2007). A similar effect is obtained when examining the monthly number of days with rainfall (Shankar et al., 1995); (Keay and Simmonds, 2006); (Hermans et al., 2006). Brijs et al. (2007) proposed a rainfall intensity indicator, defined as the centimetres of rainfall divided by its duration, which was found to increase the daily number of accidents. Further, lagged effects of rainfall (and precipitation in general) are often investigated. Eisenberg (2004) showed that the impact of precipitation on a given day is reduced when precipitation was observed in the previous days. Similar to this, Brijs et al. (2008) found that, the longer a “dry spell” (i.e. days from the previous rainfall), the higher the number of accidents in rainfall. In several studies, it was possible to interpret the positive effect of rainfall on road accidents. Keay and Simmonds (2006) showed that increased rainfall in centimetres results in decreased daily traffic volume, both at daytime and night

38 © Les collections de l’INRETS Effects of weather on traffic and safety time, winter and spring. Bergel and Depire (2004) decomposed the global effect of monthly rainfall in two components: a direct effect on the number of injury accidents and fatalities, and an indirect effect on traffic volume. In Stipdonk (2008), the indirect effect was confirmed, leading to a recommendation for estimating weather effects on road accidents under constant traffic conditions. Further, they also suggest that reduced traffic may lead to increased travel speeds that result in increased accident risk. In Table 5 on the next page, one can notice that the variables used to express each meteorological factor are quite diverse and in a few cases different results are obtained. For example, temperature may express either heat or frost conditions, whereas precipitation mainly refers to rainfall. Depending on the specification of the variables in each case, a correlation between temperature and precipitation variables may be more or less pronounced. To sum up, it can be stressed out that results from previous studies are rather consistent with regards to rainfall effects, but somewhat less consistent with regards to temperature effects. It is also important to note that, although most existing studies control for exposure, either through traffic measurements or through a proxy measures (e.g. petrol sales, vehicle fleet, and so on), only in a few studies are the weather effects interpreted through their effects on exposure.

© Les collections de l’INRETS 39 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Table 5. Summary of existing research on time series analysis of weather effects on road accidents 2001 Country Period

Indiana, USA 1995-1999 France 1975-2000 3 cities, Netherlands Italy 1999-2003 Melbourne, Australia 1987-2002 Belgium 1974-1999 Taiwan 2001-2002 Melbourne, Australia 1989-1996 France 1975-1999 USA 1990-1999 DK, SE, FI, NO 1975-1987 Norway 1974-1986 UK 1970-1978 Interventions

! Traffic

! ! ! ! ! Economic

!! Demographic

!!!! !!!! Road design Road

!! !! !! !! !

Number of days below 0 below days of Number

Dry spell (days from previous rainfall) previous from (days spell Dry

Rainfall intensity (mm/duration) intensity Rainfall Number of days with rainfall with days of Number

+ + Washington 1988-1993 Rainfall (mm) Rainfall

+ + + + + + + Sunlight (hours) Sunlight

+ + - Deviation from mean temperature mean from Deviation

+ + - Temperature (degrees) Temperature

- + - + + + + + - +

Yearly Monthly

● ● ● ● Weekly

! Daily

!

Other GLM (Poisson - Negative Binomial) Negative - (Poisson GLM

* ** ARMA

● ●

State-space

Traffic volume Traffic

Accident Severity Accident Accident Frequency Accident

!!! Local

!!!! !!!! Regional !!! !!!! !!!

Level Dependent Method Time Weather variables effect accidents on Other variables National !!!!! !!!!! !!!! !!! !!! !!! !!! ● ● ● ● 2008 2008 2008 2007 2006 2005 2006 2005 2004 2004 1995 1995 1991 1986 Author Year Stipdonk * with lagged variables ** Markovian Malyshkina et al. Brijs etBrijs al. Caliendo et al. Keay & Simmonds Keay & Simmonds Hermans et al. Chang & Chen Bergel-Hayat & Depire Eisenberg Shankar et al. Fridstrom etFridstrom al. Fridstrom &Fridstrom Ingebrigtsen Scott

40 © Les collections de l’INRETS Effects of weather on traffic and safety

In France, the impact of rain on road safety has been well analysed (Le Breton, 1990; ONISR4, 2007) and this theme was also investigated through national projects (see last chapter of this report). In France, adverse weather conditions causing numerous injury accidents are mainly rain, whereas adverse weather conditions causing severe accidents are mainly fog. Every year ONISR publishes injury accident statistics with respect to meteorological conditions and to roadway conditions; the results for 2007 show that 13% of injury accidents occur during light rain or strong rain. The following Figure 14 shows the distribution of the 81 272 injury accidents with respect to weather conditions. Figure 14. Injury accidents in France in 2007, according to weather conditions

The percentage of injury accidents occurring during rain tends to decrease slightly from 1990-2000 to 2007; it was 14 % in 1990-2000, 15 % in 2001, and 13 % in 2007. After the rainfall the roadway remains wet for a period, accidents occurring during this period are associated to rain. The following Figure 15 shows the distribution of accidents with respect to road surface conditions

4 Observatoire National Interministériel de la Sécurité Routière – The French National Observatory for Road Safety.

© Les collections de l’INRETS 41 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Figure 15. Injury accidents in France in 2007, according to road surface conditions

normal (78,1 %)

The distributions of injury accidents according to weather or road conditions must be analysed with respect to the traffic distributions during the same conditions. The added risk due to rain, defined as the ratio of the number of injury accidents per vehicle-kilometre during rain, and the number of accidents per vehicle-kilometre during normal weather conditions, is often considered. Le Breton (1990) indicated that in 1990 in France the risk for a driver is multiplied by two during rain (the value of the added risk being around 2). Some research has been developed (and some research is still needed) about the added risk, either to break down the number of accidents (the numerator of the risk) in different categories, or to improve the knowledge of the exposure to adverse weather or road conditions (the denominator of the risk). In the greater Athens area (Greece), a study was conducted to investigate the effect of temperature (daily average) and precipitation (total precipitation) on the number of total accidents and fatalities recorded for the period 1985-2005. For the daily accident data it is found that the increase in temperature results in an increase in the number of accidents. An increase in precipitation, on the other hand, is associated with a decrease in the number of accidents, possibly due to extra care exercised by the drivers. When comparing the months of the year, smaller numbers of accidents are observed in the month of August, reflecting the fact that smaller traffic counts are observed during that month in the Athens region. For the daily data on fatalities higher temperatures are also associated with larger numbers of fatalities while very high precipitation values are associated with a decrease in the number of fatalities. When monthly data are considered – including the effects of exposure- it is found that very low temperatures (that is,

42 © Les collections de l’INRETS Effects of weather on traffic and safety months with minimum daily temperatures below 5º Celsius) are associated with a reduction in the number of accidents, while higher monthly total rain precipitation results in smaller numbers of accident. Furthermore, higher passenger car traffic results in a lower accident number (presumably due to the lower speeds that are allowed due to higher congestion), while larger numbers of heavy trucks leads to an increase in accidents. Similar relations are found in the analysis of road fatalities at the monthly level. In the United States of America, according to the FHWA (2009), on the basis of the statistics of the National Highway Traffic Safety Administration (NHTSA), there are on average over 6 400 000 vehicle crashes each year. 24 % of these crashes – approximately 1 561 000 – are weather-related. Weather- related crashes are defined as those crashes that occur in adverse weather (i.e., rain, sleet, snow, and/or fog) or on slick pavement (i.e., wet pavement, snowy/slushy pavement, or icy pavement). Nearly 7 400 people are killed and over 673 000 people are injured in weather-related crashes each year. Most weather-related crashes happen on wet pavement and during rainfall. 75 % of weather-related crashes occur on wet pavement. 47 % happen during rainfall. 15 % of weather-related crashes happen during snow or sleet. 13 % occur on icy pavement. 11 % of weather-related crashes take place on snowy or slushy pavement. Only 2 % happen in the presence of fog5. Each year 17 % of fatal crashes, 22 % of injury crashes, and 25 % of property- damage-only crashes occur in the presence of adverse weather and/or slick pavement. That is, nearly 6 600 fatal crashes, nearly 450 000 injury crashes and nearly 1 104 900 property-damage-only crashes in adverse weather or on slick pavement annually.

5 Source: Eleven-year averages from 1995 to 2005 analyzed by Noblis, based on NHTSA data.

© Les collections de l’INRETS 43 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Table 6. Weather related crash statistics in the USA

Source: FHWA (2009)

A comprehensive study about weather-related crashes in the USA for the period 1995-2001 can be found in Goodwin (2002). The author explains that among all the factors, snow and ice are the most critical regarding safety risks. In the Table 6 above and in the figures 16, 17 and 18 below “Slick Pavement” includes wet, slushy, or icy pavement. “Weather” or “Adverse Weather” includes rain, sleet, snow, fog, (rain & fog), or (sleet & fog). “Non-Adverse Conditions” include dry pavement with no weather; other pavement conditions (i.e., “sand, dirt, oil”, “other”, and “unknown”); and other weather conditions (i.e., “other” or “unknown”).

44 © Les collections de l’INRETS Effects of weather on traffic and safety

Figure 16. Average Weather-Related Crashes 1995-2001.

In Canada, Andrey (2003), after reviewing 14 studies addressing the relative risk due to rain or snow, found a 70 % increase in the risk of injury due to rainfall. In Australia, Keay and Simmonds (2005) found that, in Melbourne, the rain has an effect on the accident frequency. This effect varies with the period in the day. They observed that the amount of rainfall has a significant impact on accident occurrence. Some contributory factors, such as the duration of the dry spell before the rainfall and the season, have been identified in Eisenberg (2004) and Edwards (1999a). Indeed, there is a lagged effect of precipitation across days: that is, the effect of rain is higher if many days have passed since the last precipitation. Regarding other studies, increased accident risk has been also highlighted as a major impact of precipitation in the form of rainfall and snowfall (Edwards, 1999b), (Eisenberg, 2004). A study on urban freeways pointed out that differences in certain aspects of lighting and weather are related to the mean volume and variation-of-volume in the right lane under accident conditions, which in turn influence the locations of the collisions, while hit-object collisions and collisions involving multiple vehicles that are associated with lane-change manoeuvres are more likely to occur on wet roads, while rear-end collisions are more likely to occur on dry roads during daylight (Golob and Recker, 2003). In Japan, Chung et al. (2005) study the effect of rain on travel accident measured on the Tokyo Metropolitan Expressway (MEX). Rainfall data monitored by the Japan Meteorological Agency’s meso-scale network of weather stations were used. The study investigated whether there are more accidents on rainy days. Using accident records from 10 selected routes and hourly rainfall from the corresponding weather stations, accident levels on wet and dry days were compared. It was clear that the average frequency of accidents during rainy hours (1.5 accident/hour) was significantly different from the average frequency at other times (0.85 accidents/hour). T-test also showed that the difference in average frequency of accidents during hours with no rain and with rain is significant at 95 % confidence interval.

© Les collections de l’INRETS 45 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Recently, Lin and Nixon (2008) have carried out a systematic review and a meta-analysis of the literature about the effects of adverse weather on safety. A literature search for relevant studies published from 1970 to 2005 has enabled the extraction of 34 relevant reports that provided 78 result records selected for meta-analysis. The generalized results from studies that compared daily crash rates during adverse weather and those during non-adverse weather indicate the following: most precipitation events are associated with a considerable increase in crash rate and injury rate. Snow has a greater effect than rain. It can increase the crash rate by 84 % and the injury rate by 75 % while rain can increase the crash rate by 71 % and the injury rate by 49 %. As precipitation intensity increases, the crash risk also increases. Undesirable road surface conditions (icy, slush, etc.) have an even more significant impact on crash risk. Figure 17. Effect of snow and rain on crash rates P Percent change crashof rate

Crash Category

In Figure 18, the impact of rain and snow on injury and crashes are shown for U.S.A., U.K. and Canada. When studies were evaluated by countries, there was considerable difference in the crash rate change, but there is no a clear pattern. Different transportation policies, climate and whether drivers accustomed to a specific weather driving conditions might be an explanation for the observed differences.

46 © Les collections de l’INRETS Effects of weather on traffic and safety

Figure 18. Effect of snow and rain on crash rate (regional differences) Percent change of crash

Weather conditions Synthesis To synthesise this part devoted to the weather effects on traffic, one can produce the following Table 7 provided by the Road Weather Program of the FHWA (2009), summarizing the main effects of some meteorological parameters on traffic (speed, travel time, crash risk)..

© Les collections de l’INRETS 47 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Table 7. Weather parameters effects on traffic performances

48 © Les collections de l’INRETS

3. Effects of weather on pavement

3.1. Effects of wetness on pavement skid resistance The friction between tyre and road and the skid resistance are one of the major characteristics or performance affected by adverse weather conditions. After some basic considerations and definitions, the contact area between tyre and road will be discussed before putting an emphasis on the effects of the water film thickness. Friction and skid resistance are two terms that are often used interchangeably even if they describe different concepts (see. Tyrosafe, 2009).

3.1.1. Frictions Friction represents the grip developed by a tyre on a road surface. It is measured by the coefficient of friction, which is divided into a longitudinal part (braking force coefficient) and a transversal part (sideway force coefficient). The coefficient of friction is defined as the ratio of the load (the force applied in the vertical direction) to the traction (the force resisting movement in the horizontal direction) of a vehicle, and depends on several factors (see next section).

3.1.2. Skid resistance Skid resistance describes the contribution of the road to the tyre/road friction. Basically, it is a measurement of friction obtained under specific standardised conditions. These measurements are usually made on wet road surfaces. The interaction between tyre and road surface provides grip to allow vehicles to manoeuvre. The grip is obtained by friction. This friction generates forces that are responsible for the of the vehicles (accelerating, rolling and braking) and for the stability of the vehicles in curves. In case of low tyre/road friction, the tyre will start to slide over the road surface (slipping). The tyre/road contact also generates noise emissions that will not be further considered in this document.

3.1.3. Tyre/road contact area As mentioned above, the friction represents the grip developed by a tyre on a road surface and depends on several characteristics (e.g. pavement texture, temperature, film of water, etc.). It is measured by the coefficient of friction - ratio of the load (the force applied in the vertical direction) to the traction (the force resisting movement in the horizontal direction) of a vehicle - and can be divided into two parts: a longitudinal FL (braking force coefficient) parallel to rolling direction and a transversal FT part (sideway force coefficient). These two friction

© Les collections de l’INRETS 49 Real-time monitoring, surveillance and control of road networks under adverse weather conditions coefficients can be expressed as a function of the variable vertical load FZ induced by the tyre on the pavement surface. Figure 19. Simplified diagram of forces acting on a rotating wheel F F = L = T fL and fT FZ FZ

Pavement friction also influences directly the geometry design of the roads. It is used to determine the minimum stopping sight distance, the minimum horizontal radius, the minimum radius of crest vertical curves, etc.

3.1.4. Friction mechanisms The physical rules describing the mechanisms and the characteristics of the contact between synthetic rubber tyres and the pavement surface are really complex (molecular temporary adhesion, viscoelastic deformation, local deformation of the materials, etc.). Therefore, the friction coefficients are basically analysed in an experimental manner, with specific testing vehicles. The two principal frictional force components are the adhesion and hysteresis mechanisms (Figure 19). Adhesion is the friction that results from the small-scale bonding/interlocking of the vehicle tyre rubber and the pavement surface as they come into contact with each other. It depends mostly on the micro-level surface roughness and is a function of the interface shear strength and contact area. The hysteresis component of frictional forces results from the energy loss due to bulk deformation of the vehicle tyre. The deformation is commonly referred to as enveloping of the tyre around the texture. When a tyre compresses against the pavement surface, the stress distribution causes the deformation energy to be stored within the rubber. As the tyre relaxes, part of the stored energy is recovered, while the other part is lost in the form of heat (hysteresis), which is irreversible. That loss leaves a net frictional force to help stop the forward motion (NCHRP, 2009).

50 © Les collections de l’INRETS Effects of weather on pavement

Figure 20. Key mechanisms of tyre/road friction

Although there are other components of pavement friction (e.g., tyre rubber shear), they are insignificant when compared to the adhesion and hysteresis force components. Thus, friction can be viewed as the sum of the adhesion and hysteresis frictional forces. Both components depend largely on pavement surface characteristics, the contact between tyre and pavement, and the properties of the tyre. Also, because tyre rubber is a visco-elastic material, temperature and sliding speed affect both components (NCHRP, 2009). 3.1.4.1. Longitudinal friction fL The longitudinal friction generates forces that are responsible for the accelerating, rolling and braking of the vehicle. It is measured by the braking force coefficient, which depends on several factors: Pavement texture and pavement characteristics (Jacot et al., 2007): the texture of the pavement surface is crucial for the braking force coefficient, particularly when the road surface is wet. A distinction has to be made between the following textures whose behaviour is not similar: - Microtexture: the microtexture corresponds to small indentations below 0.5 mm length (horizontally) and below 0.2 mm depth. The microtexture is mainly responsible for the tyre/road contact and the pavement friction at low speed. It results from the surface roughness of the pavement aggregates. The microtexture has a particular influence on the skidding resistance on dry road sections but also on wet pavements by breaking the film of water and ensuring a direct and "dry" contact between tyre and pavement. - Macrotexture: the macrotexture corresponds to the visible part of a road surface. It can be divided in megatexture and macrotexture. The so-called megatexture corresponds to surface aggregates with indentations from 50 to 500 mm length and 1 to 50 mm vertical depth. The megatexture is commonly associated to the rolling noise of pneumatic tyres and has no

© Les collections de l’INRETS 51 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

effect on the tyre/road friction. The macrotexture corresponds to smaller indentations that range from 50 to 0.5 mm length. The macrotexture is related to the mix design and laying parameters. The aggregates characteristics have an especially important effect on the macrotexture. The macrotexture allows the surface water drainage and is mainly responsible for reducing the risk at high speed. It results from the mix of aggregates with different friction coefficients. Figure 21. Micro- and macrotexture of road pavements

The Figure 22 below illustrates the link between tyre/road friction and pavement roughness. Figure 22. Link between the braking force coefficient and the pavement roughness

The pavement characteristics help to define the surface texture. The properties of the pavement also affect the long-term durability of the surface texture (polishing, abrasion …).

52 © Les collections de l’INRETS Effects of weather on pavement

Pneumatic tyre properties: The braking force coefficient depends on the tyre wear, as shown in Figure 23. Figure 23. Link between the braking force coefficient and the tyre wear

Several other tyre properties have an influence on the coefficient of friction. One can mention the tyre pressure, properties of the rubber, etc. - Atmospheric condition: the effect of the atmospheric condition on the tyre/road friction is important. Among these conditions one can mention the temperature, the presence of water on the road, the presence of snow or ice on the road, the presence of contaminants on the road, etc. As the tyres are visco-elastic materials, temperature has a direct impact on their properties. Researches show that tyre/road friction generally decreases with increasing tyre temperature, though this is difficult to quantify. The effect of water on the tyre/road friction is discussed in the next section, whereas the effect of snow and ice is discussed in the next section. Contaminants as sand, oil, dirt, etc. normally have an adverse effect on the tyre/road friction. - Speed of the vehicle: the braking force coefficient depends on the speed of the vehicles. A higher speed corresponds to a lower friction coefficient. The range of the decrease of this coefficient is depending on the specific tyre characteristics, the pavement quality and the atmospheric conditions (pavement surface conditions).

© Les collections de l’INRETS 53 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

- Slipping rate of the wheel: during braking or skidding manoeuvres, a part of the vehicle movement occurs by slipping of the wheel on the road surface. This phenomenon is measured by the slipping rate of the wheel, which fluctuates between 0 % (free riding) and 100 % (locked wheel). The braking force coefficient is maximum for a slipping rate around 10 and 20 % as shown in Figure 24. Figure 24. Link between the braking force coefficient and the slipping rate of the wheel Max. Friction coefficient

Slipping friction coefficient

Critical slipping Locked wheel

Free rolling 100 % slipping

The figure above shows that in case of an emergency braking a locked wheel needs a longer distance to stop the vehicle. That is why anti-lock braking systems (ABS) are usually installed in modern vehicles. 3.1.4.2. Transversal friction The transversal friction is responsible for the stability of the vehicle in curves. It is measured by the sideway force coefficient, which depends on the same physical rules as the braking force coefficient. Its value is generally smaller than the braking force coefficient and depends highly on the design of the tyres (rigidity of the edge, hardness of the rubber, shape of the tyre tread pattern).

There is a direct link between the longitudinal (fL) and the transversal friction (fT) as illustrated by the Figure 25.

54 © Les collections de l’INRETS Effects of weather on pavement

Figure 25. Link between the braking force coefficient and the sideway force coefficient

In case of braking in curves, both longitudinal and transversal friction coefficients are used simultaneously. The variation of the longitudinal and transversal friction coefficient depends on various factors. An example of both friction coefficients evolution is illustrated in Figure 26. Figure 26. Variation of transversal and longitudinal friction coefficients

© Les collections de l’INRETS 55 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

As mentioned above in this section, the physical rules describing the characteristics of the contact between synthetic rubber tyres and the pavement surface are really complex. This complex analysis has however been analysed in various publications. One can highlight (Golden, 1980) who developed a mathematical theory of wet road-tyre friction process. This theory incorporates various parameters such as the frictional properties of the rubber, the effects of road micro-texture and macro-texture, the water depth, the tyre stiffness, the wheel load and inflation pressure. The predictions obtained with this model concerning the linear braking and cornering friction forces as function of the degree of slip and velocity are qualitatively and quantitatively correct.

3.1.5. Effect of water film thickness on tyre/road friction The effect of the atmospheric condition on the tyre/road friction is significant. For instance, rain or snowfalls have a huge impact on the friction coefficient. The interaction forces induce vertical and tangent (horizontal) contact efforts. The vertical component equilibrates the vertical load in function of the circulation conditions. The horizontal efforts are in opposition to the tyre advancement and it can be separated in two different components already discussed in previous section: the mechanical friction linked to the rubber deformation (hysteresis) and on the other hand the creation of molecular connections between tyre rubber and pavement. This effort is called adhesion phenomenon and it disappears with the wheel rolling movement.

3.1.6. Skid resistance on dry versus wet pavements In general, the conditions on dry pavements are good for normal driving conditions and a focus is rather put on adverse climatic conditions. Note that the polishing of the aggregates could "improve" the skid resistance in dry conditions. In case of wet pavement, a film of water is created on top of the pavement that modifies the conditions and physical phenomenon in comparison with dry conditions. Indeed, before having a contact with the pavement, the tyre has to break the water film and to evacuate the water laterally, in order to maintain the contact between the tyre and the road surface. The evacuation of the water film is possible with the help of the shape of the tyre tread pattern and the macro- roughness of the road surface. However, the adhesion forces between tyre rubber and pavement are weakened or even dismissed and in that case only the friction related to the rubber deformation on the pavement asperities (hysteresis force) remains. Thus, for the same effort, the skid resistance decreases. In addition to the above mentioned frictions, hydro-dynamical effects can modify the contact surface between tyre and pavement. The interface between tyre and pavement can be separated into three different areas (Figure 27). In the first zone, at the front of the tyre, the water level increases and we have a loss of pressure and Fz decreases. In the second zone, the water is evacuated and the film of water discontinuous. Finally, the contact between tyre and pavement is performed (zone three). If the speed of the vehicle increases, the water located in

56 © Les collections de l’INRETS Effects of weather on pavement the first zone increases the pressure that can be higher than the contact pressure tyre-pavement. In that case, the contact is lost and the tyres are rolling on a continuous film of water (aquaplaning) Figure 27. Skid resistance on wet pavement – tyre-road interface

In following Figure 28, one can observe the braking force coefficient under various conditions: dry, wet and icy pavement. Note that the contact with an icy surface is even more difficult to reach. Figure 28. Link between the braking force coefficient, the speed of the vehicle and the road surface condition

In order to keep the water film thickness as thin as possible, the road design must be adapted to possible atmospheric conditions. This can be reached by choosing road pavement materials permitting a good evacuation of the water from the road surface as for instance porous asphalt. Note the crucial effect of the communicating voids in the pavement.

© Les collections de l’INRETS 57 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Figure 29. Braking force coefficient for different pavements and water film thicknesses

The usual values of the friction coefficient are heavily variable, as shown in the Table 8 on the next page.

58 © Les collections de l’INRETS Effects of weather on pavement

Table 8. Usual values of the braking force coefficients (longitudinal coefficient)

Dry Wet

≤ 50km/h > 50km/h ≤ 50km/h > 50km/h Pavement type

from to from to from to from to

Portland cement unused 0.80 1.20 0.70 1.00 0.50 0.80 0.40 0.75 normal wear 0.60 0.80 0.60 0.75 0.45 0.70 0.45 0.65 polished by traffic 0.55 0.75 0.50 0.65 0.45 0.65 0.45 0.60

Bituminous asphalt unused 0.80 1.20 0.65 1.00 0.50 0.80 0.45 0.75 normal wear 0.60 0.80 0.55 0.70 0.45 0.70 0.40 0.65 polished by traffic 0.55 0.75 0.45 0.65 0.45 0.65 0.40 0.60 excess of bitumen 0.50 0.60 0.35 0.60 0.30 0.60 0.25 0.55

Gravel compact, oiled 0.55 0.85 0.50 0.80 0.40 0.80 0.40 0.60 non-compact 0.40 0.70 0.40 0.70 0.45 0.75 0.45 0.75

Ashes compacted 0.50 0.70 0.50 0.70 0.65 0.75 0.65 0.75

Stone Crushed 0.55 0.75 0.55 0.75 0.55 0.75 0.55 0.75

Ice Smooth 0.10 0.25 0.07 0.20 0.05 0.10 0.05 0.10

Snow compacted 0.30 0.55 0.35 0.55 0.30 0.60 0.30 0.60 non-compacted 0.10 0.25 0.10 0.20 0.30 0.60 0.30 0.60

Open graded pavement types permit to reduce considerably the water film thickness on the road surface. For these pavement types high rain quantities can be evacuated in the sub-surface of the pavement. Various researches have demonstrated that the friction coefficients and skid resistance characteristics are in general satisfactory under dry conditions. Unfortunately this is not the case under wet conditions and the number of accidents can increase drastically. As discussed in this section, one of the reasons is the loss of skid resistance linked to the presence of water. This is however not the only reason and the geometrical characteristics of the road, the functional and structural pavement conditions, vehicle characteristics as well as

© Les collections de l’INRETS 59 Real-time monitoring, surveillance and control of road networks under adverse weather conditions human factors should also be taken into account for a global analysis of wet conditions effects. 3.2. Crash risks related to wet road surface Road traffic crashes are the most prolific cause of death and injury to which developed countries are exposed. Many efforts have been devoted to quantify the influence of various factors on accident. From this knowledge policies and actions aiming at reducing accident numbers (or at least their severity) have been defined and implemented. Among influencing factors it has been recorded that the weather is one of the most sensitive. Some conclusions related to weather-induced crashes were already reported in the previous chapter and the emphasis here is put on the couple weather and pavement related crashes or associated risk. A literature survey of published papers concerned with this problem was conducted in 1997. More than 35 papers were selected and analysed. A common point between these papers is the conclusion that wet road accident risk is significantly higher than accident risk on dry road. In the literature, one indicator usually used to quantify weather-related risk is the Excess Risk Coefficient, which is defined as the ratio of the wet road accident rate to dry road accident rate. The “dry road accident rate” is the number of dry pavement accidents during the study period divided by section length and cumulated traffic during this period. The “wet road accident rate” is the number of wet pavement accidents during the study period divided by section length and cumulated traffic during this period. This “excess risk” is generally explained by the following reasons related to wet weather: - Influence on physiological and psychological characteristics of drivers; - Reduction in the driving visibility; - Water on roads causes glare (headlamps of oncoming vehicles reflecting on the water film); - Water on roads makes road markings not clearly visible; - Reduction of the road skid resistance. On the French A6 motorways, it was found that speeds of vehicle in wet conditions are slightly reduced. The reduction depends upon the traffic density. The speed reduction is hardly noticeable during rush hours. Rain has almost no influence on vehicular time gaps (TG) distribution. Analysis of both vehicle speeds and TG distribution did not exhibit any significant modification. Vehicle files with TG lower than 2 seconds were less frequent during raining periods. On road curves, specific vehicle performances measures examined were: speed 200 m in advance, speed at the entrance, lateral acceleration at critical curve location, deceleration before the entrance of the curve.

60 © Les collections de l’INRETS Effects of weather on pavement

Rain has a little effect on mean speeds but speed reduction at the entrance of the curves is not dependent upon the pavement condition. There is no “wet curve” effect meaning that the speed reduction (if any) is similar before and on the curve. Driver behaviour is different during the night. Indeed, during night time, wet pavement leads to significant speed reduction. Regarding rain, drivers seem to be most sensitive to the reduction perception. Facts corroborate this statement as drizzle or light rain does not affect drivers speed, and speeds are significantly reduced only during heavy rain. Results from questionnaires conducted in France points out how the driver actually perceived the risk and react to mitigate it. The Table 9 shows a summary of this study. Table 9. Driver perception and reaction as a function of road surface wetness and visibility Road Risk surface Visibility Speed reduction perceived wetness

Running Yes V 

Wet No V ≅ V dry weather

Moist No V ≅ V dry weather

Rainfall Heavy Yes V 

Light No V ≅ V dry weather

After rainfall No V ≅ V dry weather

The main conclusions of this comprehensive study are: - Wet road accident risk is significantly higher than accident risk on dry road. - The frictional mechanism is reduced by a lubricating film of water between the tyre and the road: - Even with light rain, the drivers’ ability to stop quickly and safely is actually affected. Indeed, a light rain can make the pavement more slippery than a heavy rain, especially if it comes after a dry period in which oil and other automotive liquids have been deposited on the roadway. The resulting mixture of oil and water caused by a light rain can be extremely hazardous. - During inclement weather, water can create a critical situation by increasing potential for hydroplaning or skidding, in particular when the skid resistance of a pavement is low.

© Les collections de l’INRETS 61 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

- Heavy rain: a patterned tyre provides grooves or channels into which the water can squeeze as the tyre rolls along the road, thus again providing a region of direct contact between tyre and road. - The existence of a water film on the road is a contributing factor to road accidents, because of the decrease of the skid resistance. There are two ways for quantifying this increase in accidents. - Global point of view: observing the number of accidents and relating it to the wet surface exposure (whatever the water depth is). This approach is used for assessment and benefits to the selection of safety actions related to wet surface. In the cost-benefit analysis of such actions, the benefit is based on that risk, weighted by the efficiency of the action and by the number of vehicles or vehicles x kilometres concerned. - Analytic and comprehensive point of view: quantifying the water film depth and its time evolution, quantifying the relationship between the water film depth and the skid resistance, then quantifying the accident risk related to the skid resistance. This approach is based on physical equations and empirical calibration and might be the base of hazard/danger messages towards drivers. 3.3. Assessment of road surface condition Different reasons for wet pavement exist and the main factors are: - Precipitation; - Flooding from adjoining areas like road or from drainage system; - Leaking vehicles; - Melting snow and ice or frost; - Dew; - Retention of humidity in the pores of porous asphalt. Most of time the wet pavement is the result of precipitation but of course the period of wet pavement can be much longer than the time of precipitation. In this case the wetness on the road surface is not so intensive. Hence, in many cases it is possible to assess the wetness of the road surface by measuring the precipitation. Especially higher water film thickness is most of the time the direct result of precipitation. However, to get exact information about the wetness of the road surface it is necessary to measure it directly. The assessment of the road surface condition is part of winter maintenance operations and traffic management systems. It is necessary for the operation of special programs for variable message signs for wet road surfaces. In Germany

62 © Les collections de l’INRETS Effects of weather on pavement standards for the assessment of wet road surfaces are formulated in the MNS6 (2002). Regarding to the German standards it is necessary to measure at least: - The intensity of precipitation [mm/h]; - And the type of precipitation. [rain, snow]. However, for accurate results and the check of the measured data it is recommended to measure additionally the following variables: - Water film thickness on the road surface [mm]; - Condition of the road surface [dry, wettish, wet]; - Air temperature [°C] ; - Air humidity [%]. For the measurement of the intensity and the type of precipitation as well as for the measurement of the water film thickness and the condition of the road surface different measuring methods exist. In the Tables 10 to 13 given below, methods are described which are usable regarding the German standards.

6 “Merkblatt für die Nässerfassung in Streckenbeeinflussungsanlagen”

© Les collections de l’INRETS 63 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Table 10. Measuring methods for measuring the intensity of precipitation

Source: MNS (2007)

64 © Les collections de l’INRETS Effects of weather on pavement

© Les collections de l’INRETS 65 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Table 11. Measuring methods for measuring the type of precipitation

Source: MNS (2007)

66 © Les collections de l’INRETS Effects of weather on pavement

© Les collections de l’INRETS 67 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Table 12. Measuring methods for measuring the water film thickness

Source: MNS (2007)

68 © Les collections de l’INRETS Effects of weather on pavement

Table 13. Measuring methods for measuring the condition of the road surface

Source: MNS (2007)

© Les collections de l’INRETS 69 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

3.4. Effect of snow and ice on pavement skid resistance Generally speaking, the major interest in most of the research projects is typically focused on the tyre/pavement friction in wet conditions, where between vehicle tyre and road surface a water film exists. Such conditions are typically reported as “bad road conditions”. Most of the field friction test methods also simulate such conditions by the artificial creation of a control water film thickness between the surface and tested device (car tyre, in most of the cases). However, not typically tested but much more sever for the traffic safety, especially in some Northern-Europe countries, are the road conditions where snow or ice exists on the pavement surface.

3.4.1. Tyre and wintry road friction Tyre/wintry road friction is, in some countries, one of the major pavement characteristics affected by adverse weather conditions. Wintry road conditions may exist for a significant part of the year in mountainous areas or countries located at high latitudes. Effect of snow (or ice) on the tyre/road friction is briefly discussed in this subchapter.

3.4.2. Tyre and road contact area In general, the physical phenomena related to friction in wintry conditions are similar to those described previously, for wet conditions. However, icy and snowy road conditions reduce the friction markedly in cases of freezing rain or snowfall. In addition, wet road surface may become icy if temperature cools below zero degrees. Ice can be considered, for the purpose of friction, as an untypical substance, with a friction significantly lower than other crystalline solids (see Bowden, 1953). In addition to the low ice/snow friction, the total friction of pavement is lower than it would take place for wet condition because of the differences in the texture. When snow/ice contaminates pavement surface, both micro- and macro textures are influenced, resulting in lower tyre/road contact. According to the study conducted in Finland (see Hippi et al., 2010), the amount of snow and ice affects friction much more severely than the amount of water (see Figure 30.a.). In case of snow/ice, the friction values decrease when the snow and ice amount increases, however the data distribution is relatively wide. Relation between measured friction and amount of water on the surface is shown in Figure 30.b. In this case, friction also decreases when the amount of water increases, however the data distribution is narrow.

70 © Les collections de l’INRETS Effects of weather on pavement

Figure 30. Effect of snow/ice on tyre/road friction

3.4.3. Effect of snow/ice on tyre/road friction Friction means the good grip between pavement surface and tyre. In cases of dry and clear road the value of friction is typically between 0.3 -1.0. Water, snow and especially ice reduce friction, causing the drop off of the friction to as low as 0.1. According to definitions by Finnish Road Administration, road conditions can be categorized into three levels: very bad, bad and normal. The link between road surface condition and friction values is presented in Table 14. Table 14. Effect of snow/ice on tyre/road friction

© Les collections de l’INRETS 71 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

3.4.4. Accident risks related to wintry road surface The risk for traffic accidents increases during wintry weather conditions, because snow and ice on the road effectively reduce the grip (friction) between the road surface and tyres. This results in increased braking distances. On the other hand, dense snowfall reduces the visibility, thus further increasing the accident risk during slippery conditions. Severe pile-ups have occurred for example in Finland in such conditions (Juga and Hippi, 2009). In a blizzard case, the brisk wind combined with snowfall and slipperiness can cause very dangerous conditions for the road traffic and there is even a risk for vehicle collisions with fallen trees blocking the roads. In Finland the amount of car accidents was up to fourfold the daily average during a blizzard case in November 2008 (Rauhala and Juga, 2010). In the same vein, Andreescu and Frost (1998) found out, that in Montreal, Canada the number of accidents increased substantially on days with snowfall. This happened even though people in Montreal are used to driving in snowy conditions during winter. Norrman et al. (2000) have developed a method for deriving quantitative relationships between road slipperiness, traffic accident risk and maintenance activity in southern Sweden. They found out, that about 50 % of accidents during winter occurred in slippery road conditions. The highest accident risk was related to rain or sleet falling on frozen road surface. The second most hazardous road condition was when snowfall and hoarfrost formation occurred at the same time. In these conditions, accidents occurred in spite of full maintenance activity. So, to reduce the amount of crashes, public awareness must be increased. Andersson (2010) has investigated the relation between winter road conditions and traffic accidents in Sweden and UK. It appeared that when the winter was mild, there were more traffic accidents on the Swedish roads in slippery conditions caused by hoar frost formation or icing for example. When the winter was cold, snow seemed to be the cause for most of the accidents. Another issue to study was the effect of a warming climate on traffic accidents on the wintry roads. The investigations showed that due to rising temperature in the future the amount of accidents occurring in slippery conditions will probably decrease both in Sweden and UK. Also the number of days that need winter maintenance will be reduced until the end of this century. However, there will continuously be many dangerous marginal nights that will need the winter maintenance service. In Finland wintry conditions are common. Therefore efficient winter road maintenance and road weather warning services for car drivers are important. The Finnish Meteorological Institute and the Finnish Road Administration have produced an operational road weather warning service in co-operation since autumn 1997. The service was evaluated during winters 1997/98 – 2006/07 and the warnings were mostly successfully focused on days with a high accident rate (Sihvola et al., 2008). Typically high amount of accidents occurred when a low pressure with snowfall approached Finland from west or southwest. Many peak days for traffic accidents were also related to very low temperature with some snowfall, which was packed on the road surface by traffic. In low temperatures

72 © Les collections de l’INRETS Effects of weather on pavement

(below - 5°C) the effect of salting reduces, so salting is not typically carried out i n Finland when the weather is very cold. The reduction of friction increases the accident risk. If the driver reaction time is assumed to be 1 second and the friction coefficient drops from 0.8 to 0.25, the stopping distance with a speed of ca. 90 km/h is almost doubled, from 65 to 129 metres (Wallman and Åström, 2001). In addition to this, the drivers’ estimations of the friction conditions are generally very poor. This fact is also highlighted by Salli et al. (2008). They have investigated the relation between accident risk and wintertime road conditions in Finland. They found out, that the risk for accidents resulting in physical damage or injuries was ca. four times higher during snowy or icy road conditions compared to dry road conditions. In a Norwegian Road-Grip Project accident rates for different roadway conditions as well as for different friction intervals was assessed (Wallman and Åström, 2001). The highest accident rates were linked to ice or hoarfrost covered roads. The accident rate in very slippery conditions (friction below 0.15) appeared to be four times higher compared to “more normal” winter conditions (friction 0.35-0.44, see Table 15. However, Wallman and Åström (2001) pointed out, that the relation between road surface friction and accident rate is not a straightforward problem to explain. Table 15. Accident rates at different friction intervals (personal injuries per million vehicle kilometers) Friction interval Accident rate

< 0.15 0.80

0.15-0.24 0.55

0.25-0.34 0.25

0.35-0.44 0.20

Source: Wallman and Åström, 2001

The effect of snow and ice on the road surface and driving conditions has been reviewed largely in the Final report of the COST Action 3537. From that review it appears that heavy snowfalls have a great negative impact on driving conditions. Visual distances are reduced and traffic signs may become invisible. Traffic fluency is largely decreased and under wintry conditions the capacity of the roads is only approximately half of the capacity under normal conditions (Cypra, 2007). The grip between tyres and road surface is substantially lowered. Figure 31 below shows the road surface friction values in different road conditions and the lowest values are measured on ice and snow covered roads, when the friction is typically below 0.3 and can be on conditions as low as 0.1. On a moist surface the friction values are substantially higher, so the grip during slippery conditions can be largely improved with salting. This fact appears

7 Winter service strategies for increased European road safety.

© Les collections de l’INRETS 73 Real-time monitoring, surveillance and control of road networks under adverse weather conditions clearly from Figure 31, where the accident risk during slippery conditions is very high, but is notably lowered after salting. By efficient road maintenance actions the accident risk on highways can be minimized and the amount of accidents decreased. Also the use of studded tyres can improve the grip in slippery conditions. They are mainly used in the Nordic countries. It has been noticed, that the accident risk is very high if the whole road network is not covered by snow and ice. The black spots where the ice and snow surprise the driver are dangerous. The driver behaviour is also an important issue affecting the traffic safety and it should be improved with efficient road weather information and education (see Rama, 2001). Many studies show that drivers typically reduce the speed of their vehicle to some extent under rainy or snowy conditions compared to the average speeds under fair conditions. However, the speed reduction is not high enough to compensate the lower friction. Also, it has been observed that during or immediately after the snowfall drivers drive cautiously and with lower speeds, but somewhat later they drive much faster in spite of the fact that the friction level might still be low.

74 © Les collections de l’INRETS Effects of weather on pavement

Figure 31. Road surface friction during different road conditions

Source: COST Action 353 (2008)

© Les collections de l’INRETS 75 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Figure 32. The dependence of accident rate on the timing of road salting during wintry conditions

Source: COST Action 353, 2008

3.5. Assessment and prediction of road surface condition

3.5.1. Measurements of the pavement condition The assessment of road surface conditions is based on the measurement of the pavement conditions and water film thickness in wintertime is the same as in summertime. However, in winter there are additional dangerous pavement conditions caused by ice and snow. The main indicator, if there are already icy conditions, to evaluate the risk of dangerous wintry conditions is the freezing point temperature respectively the difference between pavement temperature and freezing point temperature. The freezing point temperature can be calculated (passive sensor) or measured (active sensor). That means an active measurement does not mean conductivity, capacity and polarisation measurements. The measurements are deemed active by virtue of the fact that sensors in the road surface can be cooled and warmed. The main advantage of this is that it is

76 © Les collections de l’INRETS Effects of weather on pavement possible to anticipate critical situations by simulating them artificially before they actually occur. To illustrate this principle, consider following situation: The road surface is wet and the temperature is dropping. When the temperature drops below 4 °C, a cycle is triggered to cool the probe. In this way, the probe will detect the presence of ice before there is any risk of black-ice on the road. So it is that this active measuring technique makes it possible to anticipate any risk of black- ice forming. The active measurement is independent from kind of thawing agent. One of the main characteristics of this process is that it takes account of the presence of any residual thawing agent as well as of sundry polluting agents that may be present on the road surface. By this technology, the condition of the road can be determined by simulation so that more accurate alarms can be triggered. Figure 33. Accorate FPT - prediction with active measuring

Examples are a BOSO Sensor, which uses several measurement sensors and a Peltier thermal element that heats and cools part of the surface of the probe. The active part cools down 2°C below pavement temperature and is able to use a certain cycle to detect/predict hoar frost. The next step is to measure the freezing point temperature of the liquid solution by artificially cooling a small area on the surface of the sensor up to 15°C below the current pavement temperature. This measurement takes implicitly into account all parameters influencing the freezing point temperature (type and remaining quantity of chemicals, dust, tyre particles etc.), thus delivering the most accurate information.

© Les collections de l’INRETS 77 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Figure 34. Different active pavement sensors measuring freezing point temperature

The reliability of the measuring point prognoses for road conditions is increased yet further by active sensor technologies. For this, a road surface sensor is cooled in stages at regular intervals until ice formation is determined on the sensor surface. This means that a freezing point temperature is not calculated from resistance measurements but is measured (increased accuracy). This active technology is particularly advantageous for forecasting frost. Note that the COST 309 “Road Weather Conditions” has investigated some methods of detecting, forecasting and mapping hazardous weather related road conditions (COST 309, 1992). The prEn 15518-38 lays down the requirements for the recommended components of a stationary equipment of a RWIS. In Table 16, requirements are shown for freezing point temperature, which is divided into measured (active) and calculated (passive) freezing point temperature.

8 Winter maintenance equipment - Road weather information systems — Part 3: Requirements on measured values of stationary equipments)

78 © Les collections de l’INRETS Effects of weather on pavement

Table 16. Requirement for freezing point temperature (prEN 15518-3)

Parameter Requirements

Measured:

Measuring range: –30 °C to 0 °C Resolution: 0,1 °C Accuracy: 0 °C to –15 °C, (± 0,5 °C) –15 °C to –30 °C, (± 1,5 °C) This requirement is independent of the de-icing agent being used. These accuracies are obtained under the following conditions: Aqueous solution film thickness: 0,05 mm to 0,5 mm; Measured from ≤ 4 °C pavement surface temperature. Freezing

point temperature Calculated:

1. Measuring range: –30 °C to 0 °C 2. Resolution: 0,1 °C 3. Accuracy: 0 °C to –2,5 °C, (± 0,5 °C) –2,5 °C to –30 °C, (± 20 %) This requirement depends on the de-icing agent being used. These accuracies are obtained under the following conditions: Aqueous solution film thickness: 0,05 mm to 0,5 mm; Under defined and constant de-icing agent Measured from ≤ 4 °C pavement surface temperature

3.5.2. Prediction of pavement condition Predictions of pavement conditions based on numerical weather prediction (NWP) models might deliver valuable information in order to prepare for difficult weather conditions. Until recently, resolution and physics of NWP models were not detailed enough to deliver accurate information on icing. The quality of weather forecasts strongly improved in the last few years. One reason is growing computing resources that allow smaller grid sizes and, thus, a better representation of reality. Another reason is that improvement of quantitative precipitation forecasts was one research focus in the last few years and led to an accelerated development in this field. One important step was that a new generation of NWP models became operational in the last few years. Those use horizontal grid sizes in the range of 2 – 3 km and concentrate on the forecast range from 2 – 24 hours ahead. Forecasts are run several times a day and the newest measurements are used in order to stay close to reality. Further increase of horizontal resolution is planned for the next years.

© Les collections de l’INRETS 79 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Predictions of pavement conditions are similar to forecasts of icing on structure. Similar variables are relevant: temperature, humidity, clouds and rain water. Thus, results of case studies with atmospheric icing on structures are used to show the capability of high-resolution NWP to predict the relevant variables. The mesoscale, non-hydrostatic Weather Research and Forecasting (WRF) model is used to forecast icing events measured at three sites in Switzerland during the winter 2008 / 2009. The simulations are performed at a high resolution of 800 m grid size. A sophisticated cloud microphysics scheme is used in order to get a good description of cloud and rain development. Simulated temperature, wind speed, cloud and rain water drive an accretion model (Makkonen, 2000) that calculates the temporal development of ice load on a cylindric structure. The results of a case study at Schwyberg, in the pre-alpine region, are shown in Figure 35. The region is characterized by complex terrain (Figure 35.a.). The simulated and measured temporal development of ice shows a pretty good agreement (Figure 35.b.). The icing event is simulated about 4 hours too early and the duration of the event is overestimated by about 25 %. The comparison of several case studies regarding the duration of icing events shows that most of the icing events are captured by the model. An uncertainty of 10 hours for the start of the icing event was allowed during this evaluation. The mean absolute error of the duration of icing events is 20 hours compared to an average duration of about 50 hours, which is in the order of 40 %. Figure 35. Orography at the Schwyberg site and the comparison of measured and simulated ice load Figure 35.a. Schwyberg site

80 © Les collections de l’INRETS Effects of weather on pavement

Figure 35.b. Ice load

For the Jura and the Pre-Alps region, grid sizes of 2 – 3 km are sufficient to get a reasonable result, while in the Inner-Alps grid sizes around a few hundred meters are necessary. The prediction of the maximum ice load is less reliable than the prediction of frequency and duration of the icing event. Especially because it strongly depends on the cloud droplet number concentration which is a parameter that is not well known. Figure 36. Measured and simulated duration of icing events for several case studies at Gütsch, Schwyberg and Matzendörfer stierenberg in Switzerland

© Les collections de l’INRETS 81 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

The results show that nowadays high-resolution weather models are able to capture weather situations with icing conditions in a satisfying way. The ability to predict temperature and cloud and rain water is an important basis for the prediction of pavement conditions with icing. Still, operational forecasts at 800 m grid size are still out of reach due to limitations in computing resources. Thus, operational forecasts at 2-3 km grid size will provide a good basis for pavement condition forecasts. In order to improve local predictions, NWP model output is post-processed by using statistical methods like MOS, Kalman filter or neural networks. These measurements are efficient tools if measurements are available at the position.

82 © Les collections de l’INRETS

4. Operational state of practice and best practices

As mentioned in the previous chapters, data on weather and road surface conditions are required by a number of real-time applications and user services such as on-trip and pre-trip information services, weather-related traffic information, tunnel and bridge control and rerouting. Real-time and forecasted weather and road surface condition data are also an essential prerequisite for winter maintenance operations (Giloppe et al., 2002). In addition, such data is collected for the purpose of planning ITS, winter maintenance and other road operator services. A Road Weather Information System (RWIS) is comprised of Environmental Sensor Stations (ESS) in the field, a communication system for data transfer, and central systems to collect field data from numerous ESS. These stations measure atmospheric, pavement and/or water level conditions. Centralised RWIS hardware and software are used to process observations from ESS to develop nowcasts or forecasts, and display or disseminate road weather information in a format that can be easily interpreted by a manager. RWIS data are used by road operators and maintainers to support decision making and associated decision support systems – DSS, (see, TEMPO, 2007). 4.1. Weather sensing and RWIS

Introduction Road weather and road surface condition information is today primarily collected by fixed road weather monitoring outstations (see Figure 37). In addition, CCTV9 cameras are being increasingly used for measuring or/and verifying road weather and road surface conditions. In-vehicle systems for monitoring road weather and road surface information are under development, but the development has not yet reached a stage where quality requirements could be agreed upon. The same applies to local climate models, which are an important part of road weather monitoring enabling the estimation of the road surface status on the basis monitoring data at a specific location (see COST 344, 2002).

9 Closed Circuit Television.

© Les collections de l’INRETS 83 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Figure 37. Road weather outstation

Several sensors have been developed to measure and monitor road surface and weather conditions (see COST 727, 2006). Individual sensors are usually attached to a recording device; and usually a number of sensors are mounted to a tower to create an outstation. RWIS networks support in a number of significant ways: - They improve the accuracy of decision-making by providing an understanding of actual road temperatures, trends and forecast accuracy. - They provide a monitor of road temperature, wet/dry status, freezing point of the solution on the road, the presence of chemicals and concentration as well as subsurface temperatures. - By installing atmospheric sensors on a tower, they can provide real-time localised information about the atmospheric conditions; such as precipitation, relative humidity, dew point, air temperature and wind speed and direction. - Weather forecast providers can use the information to assist in the provision of localised road weather forecasts to help the highway authorities’ decision making. The data can also be used to verify the quality of weather forecasts. - Anti-icing and de-icing chemical usage can be optimised through more accurate deployment of equipment and application of chemicals. - Additional sensors can be added to RWIS to further support the highway authority in maintaining the road network; e.g. devices to measure road friction and snow cover, and automated liquid de-icer application systems (RWIS Web Guide, 2009).

84 © Les collections de l’INRETS Operational state of practice and best practices

4.1.1. RWIS Infrastructure An Outstation is usually composed of different modules, which include some or all of the following (FGSV AK 3.2.1, 2009): - Road sensors in travel lanes: This type of sensors allow the following measurements: - surface temperature, - sub-surface temperature, - surface condition (at least dry, moist, wet, frozen), - amount of deicing chemical on the road, - freezing point of the road surface. - Atmospheric sensors adjacent to the road: The available measurements are: - air temperature, - relative humidity, - wind speed and direction, - precipitation (quantity and type). - A data logger to which all the sensors are connected and which translates and records the signals received from the sensors into a format that can be communicated to another computer. - A communication device, such as a modem, to allow remote collection of data and transfer of new data logger programs and other software updates without visiting the site. - And a power source. It is very important to find suitable locations for road weather monitoring outstations to acquire the best data for assessing the road weather conditions. Thermal mapping is often used to find optimal locations to represent a certain stretch of road and to get the earliest possible warning for conditions that affect the traffic. Most of the outstations are located along the road (see subsection 4.2.3.) but sometimes stations may be placed far from the road to get representative measurements and calculations. Local vegetation and topography must be taken into account when installing a station, and the sensors also have to be placed according to technical guidelines to measure the parameters correctly (RWIS Web Guide, 2009).

4.1.2. Weather data granularity and applications Data is usually collected from the stations at specific intervals, which may vary from 0.5 to 5 minutes for traffic control, management and other VMS-based information applications and between 20 and 60 minutes for collective traffic information service and winter maintenance applications. Data needs to be processed at the latter types of applications within 3 minutes of data retrieval.

© Les collections de l’INRETS 85 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Road weather data can be distributed to end users in a number of forms – via VMS showing in text form temperatures, road surface condition, visibility or wind, via radio broadcasts either in spoken or coded (RDS-TMC) form, TV broadcasts, TeleText and also via subscribed services to end user terminals, e.g. mobile phones. Internet-based direct end-user access can also be set up, with users finding data by site description and location or by interaction with a web-based Geographical Information System. The negative effects of adverse weather conditions on traffic can be met by displaying adapted speed limits and warning signs on VMS.

4.1.3. Quality Needs Assessment Assessment and verification of existing data is scarce, unless visual observations or camera information or both are available at the station. According to the RWIS Guide (RWIS Web Guide, 2009), weather data quality can be assesses via the following metrics: - Consistency: The data needs to be consistent to enable ‘like for like’ comparisons to be usefully made and to achieve the user acceptance essential for required impacts of the user services. - Coverage: In countries and regions with critical weather problems on a regular basis, the coverage needs to be high covering all existing critical climate areas. In other countries and regions, all major links on the TERN10 need to be covered. The data should be available for more than 99 % of time. - Volume: The number of monitoring sites/units needs to be sufficient to cover all critical climate areas. The site selection should be made based on either long-time analysis of the location of road-weather related traffic problems and/or inventory measurements such as thermal mapping. In countries and regions, where critical road weather problems are frequent, this means a high number of stations (one station for each 50 km or less). - Present experiences in Germany (Dinkel et al., 2008) show, that in section- based control systems, an interdistance of 2 kms (max. 4 kms) lead to an optimized acquisition of the longitudinally inhomogeneous environmental data and thus to more feasible traffic control. - Accuracy: The accuracy and other quality level requirements for road weather monitoring parameters are given in Table 17.

10 The Trans-European road network.

86 © Les collections de l’INRETS Operational state of practice and best practices

Table 17. Available quality levels for the various monitoring parameters. Calculated data is determined on basis of measured data as a temporally aggregated value

range/ Temporal spatial temporal spatial Parameter Accuracy parameter accuracy resolution resolution resolution resolution resolution

Surface Avg. surf. -40…+50/0.1°C ± 0.3 °C 1 min. point ± 0.2 °C 10 min. point Temperature temp.

Subsurface Avg. subsurf. -40…+50/0.1°C ± 0.3 °C 1 min. point ± 0.2 °C 10 min. point Temperature temp.

Air -35…+40/0.1°C ± 0.3 °C 1 min. point Avg. air temp. ± 0.2 °C 10 min. point temperature

Dew point ± 0.6 °C 1 min. point Avg. dew point ± 0.3 °C 10 min. point

Surface Avg. freezing ± 1 °C 1 min. point ± 0.5 °C 10 min. point freezing pt. pt.

Relative 0…100/1 % ± 3 % 1 min. point Avg. rel. hum. ± 2 % 10 min. point Humidity

Road surface classes 90 % 1 min. point 90 % 10 min. point Condition

Road surface Average or 0.05…1.0/0.01 ± 0.1 5 m ± 0.05 0.2 - 1km Friction minimum fr.

Wind direction 0…360/1° ± 10° 1 min. point Avg. wind dir. ± 5° 10 min. point

Avg. wind Wind speed 0…40/1 m/s ± 1 m/s 1 min. point ± 0.5 m/s 10 min. point speed

Gust 20…80/1 m/s ± 1 m/s 5 sec. point Max. gust ± 1 m/s 10 min. point

Precipitation 0…300/ 0.1 Cumulative ± 35 % 1 min. point ± 30 % 15 min. point Intensity (*) mm/h precipitation

Precipitation ± 30 % 1 min. point Avg. amount ± 30 % 15 min. point Amount

Precipitation rain/ 90 % 1 min. point 90 % 15 min. point Type wet/snow/snow

Cloudiness 2 - 5 ± 1/8 (1/8 of sky) times/day

10…2000/ Visibility ± 50m 10 sec. up to 4 km Avg. visibility ± 30 m 5 min. up to 4 km 10 m

Global 0…1200/ ± 0.3 % 1 min. point Avg. Radiation ± 0.3 % 15 min. point radiation 4 W/m2

+/- 2000/ Net radiation ± 1 % point Avg. radiation ± 1 % 15 min. point 4 W/m2

(*) accuracy of precipitation intensity is linked to the type of precipitation (for rain accuracy is ±10 %, for wet snow ±30 % and for dry snow ±50 %).

© Les collections de l’INRETS 87 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

4.1.4. Data quality As discussed in the previous chapters, critical weather conditions in road traffic affect directly road capacity and traffic safety. To adjust those constraints, network management systems have to consider environmental data. For sustaining a high level of traffic safety for road users, traffic control algorithms rely on sufficient, precise and fast detection of meteorological data, especially precipitation intensity, visibility and water film thickness. At present, the operators of traffic management systems often face poor quality or even malfunctioning of road weather sensors. Environmental data-based traffic control can only have positive effects on network level of service and safety if weather information is sufficiently accurate to allow its integration into control strategies. Therefore the stationary detected environmental data should be of good quality and the actual meteorological conditions should be displayed promptly in the control system. These are preconditions for developing weather-sensitive control strategies. The exact value of atmospheric conditions is hard to check because of their inhomogeneous and unsteady characteristics. Validation and quality management is a major task of data processing to detect malfunctioning data sources and exclude them from further usage for traffic control. Several plausibility checks have been developed, tested and optimized (e.g. Ben Aissa, 2007, Dinkel et al., 2008). In the future they can be used in day-to-day operations. Meteorological consistency (e. g. between fog and relative humidity) can be used to reduce false alarm rates and improve acceptance of traffic control systems. These plausibility checks were developed and tested (Goodwin, 2003) based on the following: - plausibility checks for single measurements; - checking logical-physical coherences; - long term plausibility checks. 4.2. National projects and initiatives As mentioned earlier, the stations are nowadays used for extreme weather situations, for winter maintenance and for cross-wind warning. This section is based on the input from different countries and covers experimental projects, pilot implementation and related initiatives.

4.2.1. US Department of Transport The U.S. Department of Transportation (DOT) Federal Highway Administration (FHWA) Road Weather Management Program, in conjunction with the Intelligent Transportation Systems (ITS) Joint Program Office has

88 © Les collections de l’INRETS Operational state of practice and best practices established the Clarus11 Initiative in 2004 to reduce the impact of adverse weather conditions on surface transportation users. Clarus is a research and development initiative to demonstrate and evaluate the value of “Anytime, Anywhere Road Weather Information” that is provided by both public agencies and the private weather enterprise to the breadth of transportation users and operators. The goal of the initiative is to create a robust data assimilation, quality checking, and data dissemination system that can provide near real-time atmospheric and pavement observations from the collective state's investments in road weather information system, environmental sensor stations (ESS) as well as mobile observations from Automated Vehicle Location (AVL) equipped trucks and eventually passenger vehicles equipped with transceivers that will participate in the Vehicle Infrastructure Integration (VII) Initiative. During 2004 to 2005, the Clarus System was designed and constructed. During 2006, the system was extensively tested during a Proof of Concept Demonstration utilizing ESS data from three States. The system is now ready to be populated and for its data to be utilized by the transportation community and the weather enterprise. In addition, the Federal Highway Administration of the US Department of Transport (DOT) has compiled a report for transport managers, in which important best practice cases have been described and analysed. This report contains 30 case studies in 21 US states that improve roadway operations under inclement weather conditions, listed below.

11 http://www.clarusinitiative.org/

© Les collections de l’INRETS 89 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Table 18. US road weather management program case studies in 21 US states State DOTs / Cities Road weather management tools Alabama DOT Low Visibility Warning System California DOT Motorist Warning System City of Palo Alto California Flood Warning System City of Palo Alto California Flood Warning System City of Aurora Maintenance Vehicle Management System Florida DOT Motorist Warning System City of Clearwater Florida Weather-Related Signal Timing Anti-Icing/Deicing Operations Idaho DOT Motorist Warning System Michigan DOT Maintenance Vehicle Management System Access Control Minnesota DOT Anti-Icing/Deicing System Anti-Icing/Deicing Operations Montana DOT High Wind Warning System Nebraska DOT Road Weather Information for Travelers Nevada DOT High Wind Warning System New Jersey Turnpike Authority Speed Management City of New York Anti-Icing/Deicing System City of Charlotte North Carolina Weather-Related Signal Timing Oklahoma Environmental Monitoring System Hurricane Evacuation Operations South Carolina DOT Low Visibility Warning System Tennessee Low Visibility Warning System City of Dallas Texas Flood Warning System Houston Texas Environmental Monitoring System Fog Dispersal Operations Utah DOT Low Visibility Warning System Virginia DOT Weather-Related Incident Detection Road Weather Information for Travelers Washington State DOT Speed Management Wyoming DOT Avalanche Warning System

90 © Les collections de l’INRETS Operational state of practice and best practices

One group of these case studies describe motorist information and warning systems (including warning systems due to a low visibility, high wind warning systems, flood warning systems, access control systems for winterly road conditions), line control systems for speed reduction under adverse weather conditions and weather-related signal timing. The other group of tools support road operators’ maintenance tasks (mainly winter maintenance as e.g. automatic de-icing systems), avalanche warning systems and evacuation planning. Many of these systems have been implemented after inclement weather conditions had caused a major traffic accident with fatalities and injuries. Before- and-after studies have been conducted for most systems, and most systems have been shown to improve road safety (e.g. reduction of accident rates) or to reduce total network delay and maintenance costs. Each case study has six sections including a general description of the system, system components, operational procedures, resulting transportation outcomes, implementation issues, as well as contact information and references. More details on each case study can be found in Goodwin and Pisano (2003).

4.2.2. Road Pilot - Austria 4.2.2.1. Objectives of implementing RWIS The traffic information system on Austrian motorways is operated by the Austrian Road Operator ASFINAG, the Austrian Railway Operator ÖBB and the Austrian Air Management Control Agency Austrocontrol, who is providing weather information. Weather information is offered in addition to traffic information as e.g. travel time and information on traffic-related events (e.g. road construction sites). Motorway users can access information via Internet (see www.asfinag.at, category Services) and mobile telephones with integrated internet browser. 4.2.2.2. Spatial and temporal coverage About 220 measurement stations are situated alongside the Austrian motorway network, which has a length of about 4000 km carriageway.

© Les collections de l’INRETS 91 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Figure 38. Meteorological measurement network of Austrian motorways

Weather parameters and associated indicators Weather parameters which are contained in the information for road users are precipitation, air temperature, wind speed and direction and visibility (incl. predicted values for 3, 6 12 and 24 hours). Table 19. Road pilot (Austria) characteristics

Range/ Temporal Spatial Measured parameter Accuracy resolution resolution resolution

Air temperature NA NA NA NA

Wind direction NA NA NA NA

Wind speed NA NA NA NA

Precipitation type NA NA NA NA

Visibility NA NA NA NA

92 © Les collections de l’INRETS Operational state of practice and best practices

4.2.3. Danish road directorate weather information system 4.2.3.1. Objectives of implementing RWIS There is in total 320 weather stations in Denmark. 150 are owned by the 100 municipalities, 150 are owned by The Danish Road Directorate and the remaining part is owned by Sund and Bælt – owner of the great Danish bridges - and Danish airports. The applications up to now for these stations are primarily winter maintenance and warning for cross-wind on the large bridges. 4.2.3.2. Spatial and temporal coverage On the roads administrated by the Danish road Directorate (approximate 3 600 kms), there is in average one measuring station per each 30 km. On the major roads administrated by the 98 Danish municipalities (approximately 10 000 kms), there is in average one measuring station per each 60 km. The measurement stations used in Denmark have been manufactured by Malling (the original Danish system), by Vaisala and by Boschung Mecatronic. There are about 120 Malling stations, 100 Vaisala stations and 100 Boschung stations. The stations have been installed in Denmark since 1985. Today they are collected as a total set of data from all stations to each database for every 5 minutes. In parallel to this each terminal is updated – as standard – for each 3 minutes. Collection of a complete data set from all stations to the database will have duration of less than 1 minute if no problems occur, and never longer than 2 minutes. This means that a complete dataset in average will be 4 minutes old, when received at the terminal, and never older than 8 minutes. 4.2.3.3. White versus blue spots Due to the use of stations for different applications, there are different methods of positioning a weather station. Positioning a cross-wind station is rather simple. One can actually choose the large bridges, and then there is an ideal cross-wind position. However, for positioning of measuring stations at white spots (parts of the road which first become icy) a technical sound approach is needed and the Danish Road Directorate has developed a know-how, which can be refer to as a white spot methodology. Regarding extreme rain situations, the Danish Road Directorate have started an investigation on how to develop a, associated RWIS where weather stations have to be place at a so called blue spots. - White spots. A white spot can be defined, at least in Denmark, as the coldest spot on a road within a limited distance, e.g. 30 km. The advantage of this cold spot definition is that it can be determined via an objective measurement done with a thermographic vehicle. When several road locations have the same low surface temperature, the final selection of the position for the new measuring station is then based on the probability of humidity and availability of supporting infrastructure (communication and power supply). If there is an ideal cold spot without infrastructure, the station

© Les collections de l’INRETS 93 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

will be established with solar cells for power supply and wireless internet via GPRS or 3G. - Blue spots. During the last 3 years there have been observed several situations with very heavy rainfall. This has resulted in increased risk of incidents and accidents, because the drains along the roads cannot drain off the water from the road surface. A blue spot refers to such a spot where there is an increased risk of dammed up water on the road. Similarly to the white spot positioning, there is at this moment an internal study at the Danish Road Directorate, with the purpose of designing an objective criteria for finding the blue spots. The three-phase process is undertaken: (i) phase 1 focuses on solving problems caused by extreme rain situation by improving drain systems in areas with high probability of damming-up of water. This is a pure mathematical task with use of GIS; (ii) phase 2 deals with the use sensors on the existing measuring stations to detect how much water is on the surface of the road; (iii) In phase 3, it is planned to develop prognosis models based on weather radar pictures and measurements from sensors. The forecast from the models shall then via different information systems (e.g. Vintertrafik) gives appropriate information to the drivers. The Figure 39 illustrates the density and position of all Danish measuring stations. Figure 39. Positioning of Danish measuring stations

94 © Les collections de l’INRETS Operational state of practice and best practices

3.2.3.4. Weather parameters and associated indicators The instrumentation of a measuring station is dependent on the manufacturer and specifications on the end-user. For this reason, there is a minimal configuration for a measuring station in the Danish RWIS. The minimal configuration is the set of parameters given by the first 6 rows in the Table 20 below. In this table there is an overview over different sensors, which are used in the Danish RWIS. Table 20. Danish RWIS characteristics

Measured parameter Availability Accuracy Comments

Pavement Temperature Available ± 0.1 °C Standard

Subsurface Temperature Available ± 0.1 °C Standard

Air temperature Available ± 0.1 °C Standard

Dew point Available ± 0,5 °C Standard

Relative Humidity Available ± 3 % Standard

Road surface Condition Available Standard

Surface freezing point Partly available

Water level Partly available

Road surface Friction Partly available

Wind direction Partly available ± 1 minute

Wind speed Partly available ± 0.1 m/s

Precipitation Intensity Partly available

Precipitation Amount Partly available

Precipitation Type Partly available

Visibility Partly available

As a supplement to measured data from stations, there is a need to be able to watch prevailing traffic conditions at a measuring position, mainly during snow situation. Due to this advantage and affordable high speed broadband connections, there has been an increase in the amount of webcams along the major Danish roads for the last 5 years. Today, webcams are installed on around 25 % of all Danish stations. Images from webcams are collected and distributed in the same manner as regular measurements. All data from weather stations are collected by the Danish meteorological Institute, which stores all observations and creates prognoses data. All stations are communicating via public communication lines. Until 2005 the standard

© Les collections de l’INRETS 95 Real-time monitoring, surveillance and control of road networks under adverse weather conditions communication line was a dial-up connection via ISDN telephone lines or via a dedicated network. Primarily for technical reasons the Danish Road Institute made a solution for converting transmission for all stations to Internet based communication lines. This was a very attractive solution from technical and financial standpoint. It is also important to recognize that all weather stations are genuine standard stations and all conversion of data formats are done in the central database. 4.2.3.5. Weather information dissemination and applications The collected information from weather stations and webcams has the general purpose to inform the drivers and professionals about (winter) weather incident on the roads. The general RWIS system is shown in Figure 40. Figure 40. The complete RWIS in Denmark DR D DMI Traffic WE B‐P ortal Danish Meteorological Institute

Interchange of weather and video information

Videostations RWIS stations Approx. 320

Internet

S imple weather and Detailed observations, video informations prognoses and video informations

P rofessionel Public user RWIS user

For public users, information is an easily understandable subset of data and webcam pictures. For professionals all measured data and webcam pictures are supplemented with additional weather radar precipitation data and cloud information from satellites. All information to drivers about traffic weather is given via the Danish Road Directorates internet portal http://www.trafikken.dk. The structure for the system is

96 © Les collections de l’INRETS Operational state of practice and best practices shown as part of RWIS system. This portal has a lot of subsystems, each with a different purpose. Of special interest for this report is Vintertrafik (Winter trafic) and webcams. Vintertrafik gives information to public users about the situation on the roads (temperatures, description of the surface state in form of warning signs and eventually in text messages what is done to prevent hazardous situations. In extreme situations there are about 100 000 hits on the homepage in 24 hours. In ordinary winter situations there are about 1 500 to 3 000 hits. A sample of a screen picture from Vintertrafik is shown in Figure 41. Figure 41. Vintertrafik screen dump

Webcams shows pictures from about 100 different positions in Denmark. Each image is updated each 30 sec. The Figure 42 shows where webcams are positioned around Copenhagen.

© Les collections de l’INRETS 97 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Figure 42. Webcams around and in Copenhagen

Information to professionals is more comprehensive, since all operators have been specially trained in interpreting weather information. The presentation system VejVejr (Road Weather) have the following type of information: - observations from measuring station, - prognoses based on observations for each measuring station, - precipitation radar from the five Danish weather radars, - cloud information from satellites, - images (updated each 30 sec) from webcams.

4.2.4. Finish Road Weather Information System 4.2.4.1. Objectives of implementing RWIS Weather and driving conditions have a significant effect on road safety in Fin- land in the wintertime. The Finnish Road Administration’s weather and driving condition monitoring system has been used to observe existing driving conditions for the purposes of wintertime road maintenance, providing traffic information and traffic management. This road weather information system (RWIS) has been in

98 © Les collections de l’INRETS Operational state of practice and best practices development since 1975. It became countrywide in the beginning of 1990's. Now it consists of about 400 road weather outstations and over 350 cameras. Since winter 2009 it has included also about 100 new optical remote road surface state and friction sensors. Nowadays real-time road weather information, forecasts and weather radar and satellite images provide information for the Finnish Road Administration’s traffic information centres (TICs). This data is also provided to managers at the contractors’ road weather centres who direct road maintenance subcontractors for winter maintenance. The most difficult road condition factor to measure has been road surface friction (skid-resistance) which is essential to maintain safe control of a vehicle. The Traffic Information Service warns road users of bad conditions by issuing traffic announcements and traffic weather forecasts. Traffic announcements give road condition estimates for the next 0 to 6 hours, and any possible decrease in road surface friction is considered a significant negative factor for driving conditions. The traffic weather forecasts, on the other hand, provide an estimate of the driving conditions for the next 0 to 24 hours. Friction and its changes during this time period, play a key role in these forecasts. The forecasts classify expected driving conditions as normal, bad or extremely bad. Road surface condition is a very significant factor affecting traffic management. Variable speed limits and warning signs are changed according to road conditions. Friction is an important factor affecting road usability and therefore also speed limits. The responsible organisation for traffic management including traffic information as well as snow and ice control is the Finnish Road Administration (Finnra), which is a governmental office. Finnra gets its financing from state budget. Finnra's road regions order snow and ice control from private companies. Finnra supervises the realization of the snow and ice control. Finnra has four Traffic Information Centres (TICs), which take care of traffic management and information. TIC's task is to operate the dynamic traffic management and control systems as well as to serve road users via telephone, local radio stations, text TV, Internet and the like. Traffic information consists of real time information about road conditions and traffic. 4.2.4.2. Spatial and temporal coverage Finnra has approximately 400 road weather stations, and the average density of the road weather stations is 22-25 km on the main road network. The stations collect data on air, road surface, road body (-5 cm) temperatures, wind speed and direction, humidity and dew point, precipitation, visibility, and road conditions. Some stations, about 100, can measure friction of the road surface with a new optical sensor. The Road Weather Information System (RWIS) operated by the TICs retrieves data from the stations every 5 - 60 minutes. In addition to the road weather stations, Finnra also has 350 driving condition CCTV cameras located along the main roads. These cameras are normally used to verify the road weather conditions, but increasingly also the traffic conditions at the camera locations (see Figure 43).

© Les collections de l’INRETS 99 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Figure 43. Road weather stations and CCTV cameras, and traffic counting stations in Finland in 2009

In addition to road weather data, the RWIS includes image products, such as rain radar and cloud satellite images. Finnra also purchases map, table and text- based road weather forecasts. The general RWIS system is shown in Figure 44.

100 © Les collections de l’INRETS Operational state of practice and best practices

Figure 44. Finnish Road weather information system in 2009

Road Weather Web based Users’ Information System workstations -RWC's in Finland, RWIS -Maintenance operators -TIC's -Finnra's Road weather Database personnel stations, 400 pcs. server Optical friction sensors, 100 pcs

Internet pages Road weather -local radios cameras, 350 -service providers pcs. File server -road users

Internet Teletext server -road users

Radar and satellite images, FMI 80 ILMA +2 o C Intranet TIE - 3 o C server

Road weather forecasts RDS-TMC VMS 0-24 h, 2-5 days (FORECA)

All road weather data is centrally stored in a common, national database. New weather and road condition data is written in the database at a rate of approximately nine rows a second around the clock. Database queries are processed about every other second. Road weather data is viewed using the RWIS workstation’s dedicated Windows application and/or web based user interface. The user can view road weather data in map, table and graphic windows. Finnra takes care of road weather data detection and collection. Image products, such as rain radar and cloud satellite images are purchased from Finnish Meteorological Institute (FMI). Weather forecasts are purchased either from a private weather service supplier Foreca Ltd or FMI, depending on the winner of the tendering process. 4.2.4.3. Weather parameters and associated indicators The stations collect data on air, road surface, road body (-5 cm) temperatures, wind speed and direction, humidity and dew point, precipitation, visibility, and road conditions. Some stations, about 100, can also measure the friction of the road surface by the new optical sensor.

© Les collections de l’INRETS 101 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Figure 45. Road weather station with optical road surface sensors and technical building of the weather controlled road on E 18 highway in Kotka

Table 21. Finnish RWIS characteristics

Temporal Spatial Measured parameter Range/ resolution Accuracy resolution resolution

Surface temperature NA NA NA NA

Subsurface temperature NA NA NA NA

Air temperature NA NA NA NA

Dew point NA NA NA NA

Relative humidity NA NA NA NA

Road surface friction NA NA NA NA

Wind direction NA NA NA NA

Wind speed NA NA NA NA

Precipitation Intensity NA NA NA NA

Precipitation amount NA NA NA NA

Precipitation type NA NA NA NA

Visibility NA NA NA NA

102 © Les collections de l’INRETS Operational state of practice and best practices

4.2.4.4. Weather information dissemination and applications Finnra's Road Weather Information System (RWIS) takes care of data processing and distribution. The main users of the road weather information system are located in the administration's traffic information centres and the on- call/road weather centres of winter maintenance contractors. Road weather information system information is also used by production and the public by means of intranet, Internet and Text TV pages and applications. The road weather monitoring data is also used to produce forecasts of road weather conditions for the following 24-hour period. This can be done for each station several times per day. Traffic monitoring is carried out with the help of more than 250 loop detectors based on fixed stations since the 1990’s. Continuous traffic monitoring is carried out in the capital area, where traffic related problems occur daily. In other parts of the network, where traffic related problems are usually seasonal (summer weekends), traffic monitoring is carried out when problems are anticipated or reported, e.g. in connection with an incident. In addition, since 2009, Finnra has purchased a travel time service from a private company. The service covers approximately 3 200 km of main road network. The system is based on infrared video cameras and the recognition of licence plates. The system measures the traffic on the road links, 3-5 km in urban areas and 20-40 km elsewhere. The emphasis of Finnra’s traffic management is one core functions that are most effective for ensuring traffic safety, efficiency and the effective handling of traffic demand. These core functions include traffic information via the mass media on traffic efficiency, incidents, roadworks, weather and road surface conditions, and incident management. The principle effort is on providing the core services and on the real time monitoring needed to do so. The real time traffic and road condition monitoring and the relevant data management systems are also important for the providers of commercial and value-added services. Figure 46. Example of the travel time system user interface on Helsinki metropolitan area in 2009

© Les collections de l’INRETS 103 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

The main user services operated by Finnra today are driver information, incident management and traffic control. The RWIS includes a control and computing application that produces recommendations used to control variable speed limit and information signs on the basis of set computing and deducing rules. Thus, the application makes it possible to automatically control and change the information displayed on the signs. In Finland there are about 350 kilometres of roads, which are equipped with variable speed limit system. The system can also be replenished with variable information and warning signs. About 50 kilometres of the roads are automatically controlled and the rest of the signs are manually controlled. In both cases the control information comes from RWIS. Providers of commercial and value-added services will provide customised and individualised information services. Finnra will make real-time traffic data available on its data systems for service providers. Service providers will also be allowed to collect real-time monitoring data with their own means on public roads, provided this does not interfere with traffic. Driver information services linked to data produced by the RWIS can be roughly divided into two different categories: (i) information provided before a trip (planning purposes) and (ii) information provided during a trip (en-route traffic information). Information provided before a trip can be used to plan the timing of the trip (possibly even postponing the trip to a better time), to possibly select an alternate route, and to reserve enough extra time for the trip. Information channels include TV, radio, Internet, special info signs etc.

4.2.5. French Road Weather Information System 4.2.5.1 Objectives of implementing RWIS The road weather information systems are mainly designed by the toll-based motorways, operated by three major groups (Vinci, Eiffage, Macquarie), which are well equipped by weather stations in addition to those owned by the national meteorological Agency (Météo-France). The road weather information system on these motorways is provided by the traffic operators (ASF, APRR, AREA, COFIROUTE, SENEF,…) and Météo-France, who provides weather forecasts. Weather information is offered in addition to traffic information and users can access this information via Internet and mobile phones. 4.2.5.2. Spatial and temporal coverage Weather information is provided by either the traffic operator’s RWIS or by Météo-France weather stations. For example, ASF (Autoroutes du Sud de la France) who own the largest motorway network has more than 150 weather stations (Vaisala or Boschung Mecatronic). A number of weather stations are operated by Météo-France which are categorized into 6 classes according to data availability. In principle, one can have one or two major weather stations (categories 0 or 1) for each of the 100 French departments; various stations of category 2 or 3, equipped with rain gauges, anemometers, thermometers.

104 © Les collections de l’INRETS Operational state of practice and best practices

Table 22. French weather stations categories

Type Definition/data availability

Professional station with local human observation of meteorological events 0 • hourly data, available at H+1 • daily data available the next day, at 8 H AM

Non Professional station with local or distant human observation of meteorological events 1 • hourly data, available at H+1 • daily data available the next day, at 8 H AM

On line automatic station 2 • hourly and daily data, available next day, at 8 PM

Off line automatic station 3 • hourly and daily data available 45 days after the end of the current month

Non automatic station 4 • only daily data, available 45 days after the end of the current month

automatic station with occasional call 5 • variable data availability.

In addition, there is also a national coverage by Doppler radars. One single radar analyses nebulosity on an average radius of 100 km; however, this spatial coverage depends on the height of the clouds, low clouds far from the radars are not detected; a radar image is taken every 5 minutes, which gives the possibility of estimating the evolution of clouds; the spatial resolution of the radar is 1 km².

© Les collections de l’INRETS 105 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Figure 47. Map of the Météo-France weather stations (Est part of France)

As an example, in the department “Côtes d’Armor” (located in the West region), Météo-France operates 55 weather stations of categories 0 to 5.

Figure 48. Map of the 55 weather stations (categories 0 to 5) in Côtes d’Armor

Two types of information can be expected according to the category of each station:

106 © Les collections de l’INRETS Operational state of practice and best practices

- Human-based observations: Someone - a meteorologist in weather stations of category 0, or a non-professional person in weather stations of category 1, records the meteorological events during a ivn time period (typically one hour). A large range of events are coded: different kinds of rainfall in intensity and duration (light rain, heavy rain, rainstorm, thunderstorm,.), freezing rain, ice, drizzle, mist, fog, hail, different kinds of snow. - Automatic observations: In this case the configuration is the set of parameters given by the Table 23 below: Table 23. French Météo-France weather stations characteristics Measured Temporal Availability Accuracy Comments parameter resolution Surface Average and available ± 0.1 °C 1 hour temperature maximum Subsurface available ± 0.1 °C 1 hour temperature Average and Air temperature available ± 0.1 °C 1 hour maximum Dew point available Surface freezing unknown 1 hour point Relative Humidity available Road surface non available Condition Road surface non available Friction Average and ± 1 Wind direction available 1 hour direction of the minute maximum wind Wind speed available ± 0.1 m/s 1 hour Average maximum wind Gust available ± 1 m/s 1 hour during the hour Precipitation non available Intensity Precipitation 6 min./ 1 available Amount hour available by Precipitation Type human 1 hour observer available by Cloudiness (1/8 of human 1 hour sky) observer Visibility available 1 hour Global radiation available Net radiation Available

© Les collections de l’INRETS 107 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

In addition, Météo-France can provide an estimation of rainfall every hour, in every area of 1 km², every hour, after mixing radar data and pluviometer (the Antilope system). Weather forecasts can also be provided. 4.2.5.3. Weather information dissemination and applications The road weather information is traditionally disseminated via two different channels: - Traffic information and meteorological websites such as traffic information (Bison futé) combined with a road weather bulletin12, weather and traffic on the motorways, traffic information around Paris (Sytadin - http://www.sytadin.fr/), weather and traffic information service of TV broadcaster France 2 (Pointroute - http://pointroute.france2.fr), websites Climathèque of Météo-France, meteorological alerts (Carte de vigilance - http://www.vigimeteo.com/). - Météo-France phone service and customized information to professionals. More recently, some motorways companies and traffic operators, has developed specific systems with weather information at section levels. Figure 49. gives an illustration of ASF system resulting in a joint effort with Météo France. Figure 49. ASF Traffic and Weather screen dump

12 http://www.bison-fute.equipement.gouv.fr/fr/rubrique.php3?id_rubrique=33

108 © Les collections de l’INRETS Operational state of practice and best practices

4.2.6. Greek Road Weather Information System 4.2.6.1. Objectives of implementing RWIS An example of applying RWIS systems in Greece is the case of Egnatia . Egnatia Street is built on the traces of the ancient Roman Road and has a length of 670 Km starting from Igoumenitsa in the Prefecture of Thesprotia in Western Greece to Kipi in the Prefecture of Evros in the Eastern Greece, connecting Europe with Little Asia. 4.2.6.2. Spatial and temporal coverage The road sections where weather stations operate are indicated on the following map. Figure 50. Map of road weather stations

On certain Egnatia Motorway sections, modern weather stations operate and collect weather data (wind, air temperature, frost, etc). Operating weather stations fall under three categories: - stations measuring wind data (anemometres), - stations measuring ambient temperature, - road Weather Information Systems stations (RWIS).

© Les collections de l’INRETS 109 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Figure 51. Egnatia weather stations

- Anemometer stations On certain Egnatia Motorway sections, mainly during winter, severe weather phenomena and strong winds often cause either the accumulation of large volumes of snow at specific locations on the road (in case of snowfall) or run-off road incidents that mainly involve high vehicles. In order to evaluate the problems and find possible solutions, anemometric data were collected on wind direction and speed at specific locations. These anemometric stations had wind speed and direction sensors based on ultrasounds technology, connected to a data logger. They store the average, minimum, maximum and standard deviation of values every 10’. Communication with the data collection system is performed telemetrically through a GSM modem. - Ambient temperature stations At the East Region of the Egnatia Motorway, from Ag.Syllas IC through Kipi, ambient temperature measurement sensors were installed. Data are collected automatically through wireless communication technologies (GSM, VHF), and they are accessible by maintenance crews through Web, WAP and SMS services. The information provided helps in optimising winter maintenance works (salt spreading, timely notification of snow ploughs, etc.). - Road weather information stations Weather data collection/ recording systems and systems that forecast local road weather phenomena, such as frost, fog, wind etc contribute greatly to the prompt provision of information to users, as well as to the effectiveness of winter maintenance works. On the Egnatia Motorway, Road Weather Information Systems operate on various road sections. More weather stations are to be installed on Egnatia Motorway sections that are under construction, and specifically on sections with tunnels. Weather data transfer and processing is performed telemetrically at the respective tunnel Traffic Control Centres on the above road sections.

110 © Les collections de l’INRETS Operational state of practice and best practices

4.2.6.3. Traffic data and forecasting model EGNATIA ODOS A.E. (EOAE) provides traffic data, forecasts and analyses concerning the Egnatia Mainline and its Vertical Axes, but also the greater “corridor” of the Egnatia Motorway. The “tools” developed to fulfil these needs are described below. Since 1997, EGNATIA ODOS A.E. has launched a traffic count program along the Egnatia Motorway "corridor" consisting of systematic traffic counts performed on the Egnatia Motorway and its Vertical Axes, specifically on road sections ready to be opened to traffic. The traffic count collection and processing system that has been developed is an integrated system that will ultimately comprise 65 stations in total. The systems applied at the traffic count stations involve the use of inductive loops and Remote Traffic Microwave Sensors. The collected data are transferred from all remote locations on the road axis to the EOAE headquarters at Thessaloniki with the use of special telematic equipment. In 1997, due to increased demand for reliable traffic forecasts that would contribute to the motorway design, EOAE developed a traffic forecasting model using all available data on the existing transport networks and traffic demand. This model is updated on a regular basis with new data collected by the company. The EOAE traffic model has been and is currently being used for the provision of an abundance of traffic data, forecasts and analyses necessary in the decision- making process for the design of the road, the necessary E/M installations and telematic applications, the toll collection system, the Service Areas, the feasibility studies, the calculation of environmental parameters, the planning of pavement maintenance works, etc.

4.2.7. Polish general directorate weather information system and motorways 4.2.7.1. Objectives of implementing RWIS The System of weather stations on the Polish road network governed by the General Directorate of Roads and Motorways is complementary with the traffic measuring system. Its tasks are measurement and registration of meteorological parameters on the area with local microclimate on road sections menaced by the possibility of icy roads as well as sending current data to the staff directing winter maintenance action on roads. Weather stations should be placed near the road on menaced sections of the road.

© Les collections de l’INRETS 111 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

4.2.7.2. Spatial and temporal coverage There are three sources of meteorological information, which are relevant for RWIS: 1) Traffic and Weather monitoring system on A4 Highway: - Over 80 traffic and weather monitoring stations on A4 Highway between Wroclaw and Katowice. - Parameters collected: - temperature 3 m, - temperature 20 cm, - temperature 0 cm, - temperature -5 cm, - dew point, - precipitation kind, - precipitation intensity, - wind direction, - wind speed, - number of cars on each lane, - cars number and speed in 10 classes (every 10 km); - All parameters collected every 10 minutes; - Full access to all data recorded in a database; - Access to information presented on VMS (over 40 VMS signs). 2) Traffic and Weather monitoring system on Warsaw suburbs: - Over 20 traffic and weather monitoring stations on S7 and S8 expressway between Warsaw and Radom; - Parameters collected: - temperature 3 m, - temperature 20 cm, - temperature 0 cm, - temperature -5 cm, - dew point, - precipitation kind, - precipitation intensity, - wind direction, - wind speed, - number of cars on each lane, - cars number and speed in 10 classes (every 10 km), - individual car data (if needed); - All parameters collected every 10 minutes; - Full access to all data recorded in a database; - Access to information presented on VMS (5 VMS signs).

112 © Les collections de l’INRETS Operational state of practice and best practices

3) RWIS system for Polish road network: - Over 250 weather measuring stations; - Parameters collected: - temperature 3 m, - temperature 20 cm, - temperature 0 cm, - temperature -5 cm, - dew point, - precipitation kind, - precipitation intensity, - wind direction, - wind speed; - All parameters collected every 10 minutes; - Full internet access to all database.

Figure 52. Polish road network

© Les collections de l’INRETS 113 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

The central unit RC12 is the basic element of the road measuring station. The station has the set of measuring sensors as well as indispensable equipment. The system is adapted to automatic sending information by different transmission means (by telephone, radio or cellular phone). Figure 53. Weather-traffic measuring station schema

The central unit is a multichannel measuring instrument enabling connecting: - 15 analogical sensors, - 8 string sensors, - 16 impulse sensors, - 4 cameras, - 4 detectors of traffic volume, - variables of informative board content.

114 © Les collections de l’INRETS Operational state of practice and best practices

Data collected by measurement stations are transmitted by GPRS to the database. The information is stored in the database for every station assembled in periods of 10 minutes. Every record contains all parameters of the measured station, date and hour of measurement as well as the name of the station. By use of suitable software, it is possible to read the current data from a chosen station. 4.2.7.3. Weather parameters and associated indicators Table 24. Polish RWIS characteristics

Range/ Temporal Spatial Accuracy Measured parameter resolution resolution resolution

Surface Temperature NA NA 10 min point

Subsurface Temperature NA NA 10 min point

Air temperature NA NA 10 min point

Dew point NA NA 10 min point

Wind direction NA NA 10 min point

Wind speed NA NA 10 min point

Precipitation Intensity NA NA 10 min point

Precipitation Type NA NA 10 min point

4.2.8. Ongoing projects 4.2.8.1. Austrian Wetter und Verkehr The project ‘Wetter und Verkehr’ (Weather and Traffic) aims at incorporating traffic information into traffic models, whereby a very focussed and pragmatic approach is to provide a model which primarily adapts the input values and parameters for the traffic management system to the weather situation. Parameters are estimated by means of statistical methods from historical traffic and weather data as well as from the observations and measurement of traffic flow under different weather conditions. The project is funded by the Austrian Ministry of Transport and started in February 2008 with duration of 18 months. In the following all participants and their role in the project are listed: - ITS Vienna Region – Project management and traffic data provider; - Prisma Solutions – Compilation of intermodal graph; - PTV – Setup snd calibration of traffic model for the city of Vienna; - Arsenal research – Estimation of parameters for traffic model; - Meteomedia – Weather data provider; - Mentz Datenverarbeitung – weather sensitive routing.

© Les collections de l’INRETS 115 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

To consider adverse weather situations in the traffic model two parameters are estited for different conditions: - Free flow speed: Desired speed drivers are choosing, without any interference by other vehicles; - Capacity: Maximum possible traffic flow. Both parameters represent characteristic points in the fundamental diagram, which allow reconstructing a general form for the diagram. In the traffic model developed by PTV, these two parameters have a crucial role for estimating traffic states for the network. Fundamental diagrams are related to road segments, thus both investigated parameters have to be estimated for each road segment. To estimate free flow speed, velocity measurements of several locations from fixed sensors were selected, where it could be assumed the speed has been free chosen. This is done by looking at the headway of vehicles or (if no single measurements are available) selecting values with very low traffic flow. Weather data are separated by areas, where some areas contain a measurement station and others are interpolated by a weather model. All traffic data are combined with weather information. An estimation of free flow speed is done by separating all data into weather classes and calculation statistical values of velocities within one class. The results showed no significant differences for adverse weather condition in urban areas. Thus this parameter is estimated once, but remains unchanged inside the model during adverse weather situations. Regarding capacity estimation on the basis of historical data, congested traffic situations have to be found. This is necessary because in these situations capacity of a road has been reached (assuming capacity drop is due to traffic load) and therefore measured traffic flow immediately before congestion represents road capacity. By applying a Kaplan-Meier estimator, known from survival analysis, observations with no congestion are considered as well in the estimation of empirical cumulative distribution function for capacity. This empirical function is approximated by a Weibull distribution. Median value is calculated from the theoretical distribution and provided to the traffic model for calibration. The procedure described above has been applied as well as for free speed estimation for several different weather situations (Figure 54).

116 © Les collections de l’INRETS Operational state of practice and best practices

Figure 54. Empirical and approximated theoretical distribution of capacity for different weather situations

The estimation of saturation flow has to be estimated Based on a model for signalized , saturation flow has been estimated by linear regression. The applied model represents behaviour of drivers including start and end lag to estimate an effective green time. Moreover it is considered, that trucks (resp. articulated trucks) require more space and thus reduce the number of vehicles passing an intersection. Estimations have been conducted for four different weather situations and three locations.

© Les collections de l’INRETS 117 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Figure 55. Estimation of saturation flows

Compared to the estimation of capacity, more adverse weather conditions have been investigated. To incorporate saturation flow into the traffic model of PTV it can be considered that capacity is directly influenced by saturation flow. From saturation flow and length of (effective) green time an upper limit for capacity can be estimated. As preliminary results, investigations showed significant deviations for capacity (and saturation flow) under adverse weather conditions. An influence of weather on free flow speed could not be observed for urban roads. Data of fixed located sensors have been used to estimate capacity under several weather conditions. In that case, a capacity drop up to 10 % was estimated during rain or snow situations. Because a sufficient number of observations was necessary for this method, only frequent weather situations could be investigated. Thus rarely occurring but extreme weather conditions (heavy rain, snow covered road) have been estimated by applying the model for intersection. Here the reduction due to snow on the road was about 21 % compared to wet road. A clustering of locations has to be done and further research activities should reveal a method to allocate results from one location (with sensor) to several sites with same conditions but no traffic data. Moreover the investigations are planned to be conducted for rural areas as well. Especially free flow speed is expected to behave differently from urban roads. 4.2.8.2. German Test Site for road weather and road surface condition monitoring To improve reliability of environmental sensor systems under real conditions, in the year 2003 a test site was installed in “Eching Ost” near Munich / Germany. In the test site, environmental sensor systems of different manufacturers are checked and compared under similar conditions. This way, sensor systems get

118 © Les collections de l’INRETS Operational state of practice and best practices rated for their ability as detectors in lane control systems. This shall lead to a better understanding and optimization of the systems (hardware and software). The test site project is supported financially as well as organizationally by the Federal Ministry of Transport, Building and Urban Affairs (BMVBS) and the Federal Highway Research Institute (BASt). A working group of Germany’s Road and Transportation Research Association (FGSV) serves as a supervisory board for the project. The actual test site is operated by the South Bavarian Highway Authority (ABDS); monitoring and evaluation is carried out by the Chair of Traffic Engineering and Control of Technische Universität München. This project started in 2003 and will finish in 2011. The test site provides abundant equipment like cameras, measuring facility and data link and is continuously attended. Regularly updated pictures of two web cameras in the test site are available under http://www.vt.bv.tum.de/umfelddaten. Direct input data for environmental control are precipitation intensity, water film thickness and visibility. These are the most important environmental data, the other data are used for plausibility purposes. The sensors in the test site are mainly assessed according to the plausibility of the result delivered and the reaction time. The plausibility of the result is determined: - Relative to each other (continuously); - Fully after observation (random samples); - Fully against reference (random samples). Therefore ‘qualitative’ comparisons of time-variation curves were made. In coordination with manufacturers of road weather sensor systems, reference methods have been established in order to evaluate several environmental parameters. If available, the measured data is compared to reference measurements (e. g. sensor Pluviometer for precipitation intensity, spraying box for water film thickness and status of the road surface) respectively reference observation (analysis of web cam pictures for visibility) additionally. Identified differences are analyzed. Furthermore data acquired by different measuring methods were correlated and analyzed. Based on the above characterized checks for the plausibility of measured environmental data, an evaluation scheme was developed and applied. The sensors were classified as - “Applicable”: the sensor responds to situations promptly and represents them well; - “Appropriate with restrictions”: the sensor responds to situations and represents them; - “Not appropriate”: the sensor does not responds to situations promptly and represents them insufficiently.

© Les collections de l’INRETS 119 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Validation and quality management is a major task of data processing to detect malfunctioning data sources and exclude them from further usage for traffic control. By now a huge database allows analysis of correlation between measured values by data mining methods to discover benchmark tests and error detection. A prototype software tool has been implemented to apply plausibility checks to the database. Several plausibility checks have been developed, tested and optimized. In future they can be used e. g. in day-to-day operations. Meteorological coherences (e. g. between fog and relative humidity) can be used to reduce false alarm rates and improve acceptance of traffic control systems. These checks were developed and tested: - Plausibility checks for single measurements; - Checking logical-physical coherences; - Long term plausibility checks. Results of the test site show, that not all tested sensors are able to fulfil the high expectations in quality. This is an important result, especially in consideration of fact, that in daily operations environmental data will not be analyzed as intensively as in the test site project. The developed plausibility checks shall help improving data quality. The effects of traffic near the test site since 2009 are under investigation (see Dinkel et al., 2008). 4.2.8.3. Finnish ColdSpots project A three-year national collaborative research project, ColdSpots was initiated to study the causes of, and distinguish regions (i.e. road stretches) susceptible for slipperiness, as well as to refine and further develop available road weather forecasting tools. Project ColdSpots is co-funded by the Ministry of Transport and Communications in Finland, the Finnish Road Administration, and the consortium of three public and private partners: Finnish Meteorological Institute, Foreca Ltd and Destia. ColdSpots was initiated in 2005 with first analyzing the available information and compiling the necessary databases. A test set of some fifty most problematic locations were selected based on accidents having occurred due to slipperiness of the road surface and, additionally, based on the human knowledge of individual road features by local road maintenance experts. During the second phase, 2006-07, this ColdSpots location data was implemented in the road condition models, and pilot studies were performed. A “ColdSpot” is a place where icy and slippery conditions and accidents due to slipperiness occur. ColdSpots places were defined by the knowledge of road maintenance personnel who have a lot of experimental knowledge of road places which need more care and control than other places. Additional ColdSpots information was collected from traffic accident database by picking up the accidents that happened during icy conditions.

120 © Les collections de l’INRETS Operational state of practice and best practices

Collected ColdSpots places have different kind of features: - An open area with large sky-view factor and radiative cooling; - A valley with cool air pooling during the night; - A coastal area near the sea or a lake with lot of moisture advection; - An elevated spot, a hill top with lower temperature and forced uplift of moving air; - A special place like bridge, curve, ramp or passing line. There are almost 500 road weather stations along the roads in Finland. Especially in southern Finland, main roads are pretty well covered by road weather stations. Most of the road weather stations are situated in places which are for some reason difficult for road maintenance activities. Road weather station measures typical weather parameters (air temperature, road surface temperature, humidity, wind, precipitation) but also state of the road and electrical conductivity. New optical measurement devices can also define prevailing value of friction. Mobile measurements were carried out by driving a car along the main roads in southern and western part of Finland. There were two different kinds of measurement devices attached to the track back of the vehicle; one device measures road surface temperature and the other measures the thickness of snow, ice or water on the surface and it also defines the value of friction. The observations were done every fifth seconds and the data was stored continuously. Spatial resolution was about 110 meters when driving 80 km/h. The vehicle was equipped with a GPS (global positioning system) receiver so the exact location of the car at each time step was possible to identify afterwards. The bottlenecks of road weather modelling were studied and analyzed. One of the aims of this project was to study possibilities to do more accurate road weather forecasts especially for those stretches or spots where slippery conditions exist more often by improving existing road weather models. Existing models are energy balance models, so the physic behind the environmental features must be solved. It became clear that some of the environmental features can be taken into account to the model. Such features are screening and slopes affected by topography. Other features are more or less complicated or impossible to take into account when modelling accurate road weather forecasts. Also, local and separate circumstances would be difficult and laborious to maintain. Traffic is partly embedded into the road weather model already now. Road maintenance activities could be possible to take into account, but activity information is not entyre collected in Finland. Mobile measurements for a vehicle driving between Helsinki and Turku on 30.1.2007 are presented in Figure 56. Road surface temperature was mainly between -10 and -20 ° C degrees (Figure 56 - left). Road salting is not possible in such cold temperatures, because salt loses its effectiveness to melt ice when temperature is below -6 ° C. Large fluctuation can be seen in the road surface temperature even within short times and distances. Car measurements as well as measurements from road weather stations are plotted (as stars and circles), and they are mainly close to each other. However, a couple of outliers can be found.

© Les collections de l’INRETS 121 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Temperature from road weather station presents road surface temperature whereas car measurements present air temperature. The reasons for the fluctuation were studied. Some of the cold places are situated in valleys, on crossroads, on rest stops or in city areas. However, there seems to be lots of fluctuation that cannot be explained easily. All the time there was at least tiny ice or snow cover on the surface (Figure 56 - middle). On the highway, ice or snow layer was mainly so tiny that a driver could not notice it. The peaks in the middle of observation part and in the beginning and in the end are from parking places or rest stops. Also, friction (Figure 56 - right) varies much along the test period. There are long periods between 9:30 and 10:00, when friction is low all the time. Surprisingly, friction varies sometimes quickly from 0.2 to 0.8 through the whole scale. Surface temperature and friction can vary drastically even within very short distances depending much on the prevailing weather situation. In addition, much of the observed variations are caused by environmental circumstances like topography, vicinity of surface water, openness of the road etc. Verification statistics often strongly reflect the origin of the observations (remote, fixed) and, hence, it is necessary to take this into consideration when depicting the verification results.

122 © Les collections de l’INRETS Operational state of practice and best practices

Figure 56. Road weather measurements presented as a function of time and distance – date: 30.01.2007 dle: dle: stars road weather station, circlescarmeasurements). Mid Up: Roadsurface temperature (linemobile measurements, Thicknessof water, ice and ice layers in mm:s. Bottom:Value of friction

© Les collections de l’INRETS 123 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Thermal mapping could be suitable a tool to improve road weather forecasts. The system is already in use in many countries. Thermal mapping means temperature measurements along the roads in different weather situations. The technique provides climatic temperature map of the road network. Using that information it would be possible to do statistical corrections to the road weather models outputs. The system knows automatically which places are usually colder in a specific weather condition and makes corrections automatically. 4.2.8.4. Carlink Project The objective of the international R&D project CARLINK (Wireless Traffic Service Platform for Linking Cars) is to develop an intelligent wireless traffic service platform between cars, supported with wireless transceivers along the roads. Over ten partner institutes and companies representing three countries, Finland, Luxembourg and Spain, participate in this three-year (2006-08) research undertaking. Each participating country has its own site-specific application. The CARLINK platform boasts a wireless ad-hoc type communication entity with connectivity to the backbone network via base stations and consists of three modules: - traffic Service Central Unit (TSCU), - traffic Service Base Stations (TSBS), - and Mobile End Users (MEU). The TSCU hosts a network of TSBSs along the roads. Our pilot platform uses the IEEE 802.11g-based WiFi network as the radio protocol in the TSBS. However, the platform can host practically any radio communication protocol available. The Mobile End User (MEU), located in a given vehicle, will transfer platform data every time when passing a TSBS (as well as with an encountering vehicle). The MEUs form a wireless network. They do not have continuous connectivity but operate in ad-hoc manner with each other, typically when two cars are passing each other. The data gathered from vehicles are delivered to specific weather and incident/emergency services beyond the TSCU. The updated service data are geographically spread by the TSCU to appropriate TSBSs. Potential critical warnings are delivered without delay via an additional GPRS-based backbone connection, directly from the TSCU to MEUs. This means that any extraordinary weather conditions (like dangerously slippery roads, black ice) or accidents, when having been observed by the first arriving vehicle, will generate a notice to be delivered to encountering vehicles over the platform. The incident/emergency warning service parameters are an airbag blast, a push of the emergency button inside the car, a tyre odometer and an engine status, all of them including the GPS location of the observed issue. The road weather service core covers a weather forecast model generating operational local road weather forecasts. The model is supplemented with temperature observations of the cars and their GPS locations. The resulting local road weather information is delivered to the TSCU which further forwards this data to the vehicles over the CARLINK platform. Similarly, the incident/emergency warning

124 © Les collections de l’INRETS Operational state of practice and best practices service collects vehicle data to build up a warnings database. Depending on the significance of the warning the TSCU selects the appropriate path for its delivery. The most critical warnings (e.g. location of an accident) are delivered through the GPRS connection as rapidly as possible, while the more informative-like warnings can be distributed through the base stations. The ultimate goal of the CARLINK concept is to build an intelligent wireless communication platform for vehicles in which they will deliver weather observations to the platform core and to the road weather analysis/forecasting system of FMI. This information is further delivered to vehicles as analyzed information about local road weather and as potential warnings against incident en route. The CARLINK solution has been extensively tested during autumn/winter 2008 on a pre-defined test route along highway E18 at the outskirts of Turku in Finland. Test cars traveled back and forth along this route and various functions of the CARLINK platform has been tested and demonstrated.

© Les collections de l’INRETS 125

Conclusion

This State of the Art report of the COST Action TU0702 “Real-time monitoring, surveillance and control of road networks under adverse weather conditions” has summarized in a concentrated form the existing experiences and knowledge on the effects of inclement weather conditions on road traffic operations and road safety. Together with the State of the practice as well as best practices which are available in various countries it gives a comprehensive view on the traffic, pavement and weather interactions. From scientific standpoint, the main purpose of the report was especially to show that there are questions not yet answered and where need for further research and development has to be done. All the questions and open topics for research and development also shall be a guide for the responsible officials within the European Commission to provide additional and sufficient financial support for future research projects within the member countries of the European Community. This overview, even though incomplete, also shows that a number of relevant projects are on-going. They are widely spread among countries, although the research is fragmented and, in most cases, publication and distribution of most findings is at the national level only. Coordination at the European level has only recently emerged, through initiatives such as COST Action TU0702. However, the importance of the coordination of research and development has been recognized and collaborative projects at European or even at international levels (mainly with Australia and Japan) are being set up. While this process is still ongoing, a next step could logically be to search for more collaboration in the field of weather effects on traffic operations with other parts of the world. Provided that sufficient support is given, the high capability for commitment to innovation and development of more comprehensive approaches accounting for those effects into traffic management to achieve weather-sensitive traffic management strategies and their large scale deployment. We all in the COST Action TU0702 hence can look with great expectations into future.

© Les collections de l’INRETS 127

List of figures and tables

Figures Figure 1. Monthly average temperature at Davos, Switzerland (1866 to 2009) ...... 15 Figure 2. The mean annual temperature anomalies in Switzerland ...... 16 Figures 3. 50-year time-series of temperature observations at a climatologically cold weather station in Northern Finland ...... 17 Figure 4. Histogram of temperature observations at a climatologically cold weather station in Northern Finland ...... 19 Figure 5. Percentage decrease in daily trips during rainy days by day group ...... 23 Figure 6. Speed reduction quantification – lane 1, site 149 ...... 25 Figure 7. Speed reduction quantification – lane 2, site 56 ...... 25 Figure 8. Speed-Flow curve for 3 weather conditions ...... 29 Figure 9. Legal speed limit on French motorways according to weather conditions ...... 32 Figure 10. Speed distribution on the slow lane (French data) ...... 32 Figure 11. Headway distribution of free flow under three weather conditions in 2005, lane 8, and site 226 (Switzerland) ...... 33 Figure 12. Time headway histogram (France) ...... 34 Figure 13. Lane distribution on two lanes Swiss Motorway ...... 35 Figure 14. Injury accidents in France in 2007, according to weather conditions ...... 41 Figure 15. Injury accidents in France in 2007, according to road surface conditions ...... 42 Figure 16. Average Weather-Related Crashes 1995-2001 ...... 45 Figure 17. Effect of snow and rain on crash rates ...... 46 Figure 18. Effect of snow and rain on crash rate (regional differences) .... 47 Figure 19. Simplified diagram of forces acting on a rotating wheel ...... 50 Figure 20. Key mechanisms of tyre/road friction ...... 51 Figure 21. Micro- and macrotexture of road pavements ...... 52 Figure 22. Link between the braking force coefficient and the pavement roughness ...... 52 Figure 23. Link between the braking force coefficient and the tyre wear .. 53 Figure 24. Link between the braking force coefficient and the slipping rate of the wheel ...... 54 Figure 25. Link between the braking force coefficient and the sideway force coefficient ...... 55

© Les collections de l’INRETS 129 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Figure 26. Variation of transversal and longitudinal friction coefficients .... 55 Figure 27. Skid resistance on wet pavement – tyre-road interface ...... 57 Figure 28. Link between the braking force coefficient, the speed of the vehicle and the road surface condition ...... 57 Figure 29. Braking force coefficient for different pavements and water film thicknesses ...... 58 Figures 30. Effect of snow/ice on tyre/road friction ...... 71 Figure 31. Road surface friction during different road conditions ...... 75 Figure 32. The dependence of accident rate on the timing of road salting during wintry conditions ...... 76 Figure 33. Accorate FPT-prediction with active measuring ...... 77 Figure 34. Different active pavement sensors measuring freezing point temperature ...... 78 Figures 35. Orography at the Schwyberg site (left) and the comparison of measured and simulated ice load (right) ...... 80 Figure 36. Measured and simulated duration of icing events for several case studies at Gütsch, Schwyberg and Matzendörfer stierenberg in Switzerland ...... 81 Figure 37. Road weather outstation ...... 84 Figure 38. Meteorological measurement network of Austrian motorways . 92 Figure 39. Positioning of Danish measuring stations ...... 94 Figure 40. The complete RWIS in Denmark ...... 96 Figure 41. Vintertrafik screen dump ...... 97 Figure 42. Webcams around and in Copenhagen ...... 98 Figure 43. Road weather stations and CCTV cameras, and traffic counting stations in Finland in 2009 ...... 100 Figure 44. Finnish Road Weather Information System in 2009 ...... 101 Figure 45. Road weather station with optical road surface sensors and technical building of the weather controlled road on E 18 highway in Kotka ...... 102 Figure 46. Example of the travel time system user interface on Helsinki area in 2009 ...... 103 Figure 47. Map of the Météo-France weather stations (Est part of France) ...... 106 Figure 48. Map of the 55 weather stations (categories 0 to 5) in Côtes d’Armor ...... 106 Figure 49. ASF Traffic and Weather screen dump ...... 108 Figure 50. Map of road weather stations ...... 109 Figure 51. Egnatia weather stations ...... 110 Figure 52. Polish road network ...... 113 Figure 53. Weather-traffic measuring station schema ...... 114

130 © Les collections de l’INRETS References

Figure 54. Empirical and approximated theoretical distribution of capacity for different weather situations ...... 117 Figure 55. Estimation of saturation flows ...... 118 Figure 56. Road weather measurements presented as a function of time and distance – date: 30.01.2007 ...... 123

Tables Table 1. Influence of weather on speed ...... 26 Table 2. Main findings on speed reduction under inclement weather conditions ...... 27 Table 3. Capacity reduction under light rain ...... 28 Table 4. The average impact of weather on freeways capacity and speed ...... 30 Table 5. Summary of existing research on time series analysis of weather effects on road accidents ...... 40 Table 6. Weather related crash statistics in the USA ...... 44 Table 7. Weather parameters effects on traffic performances ...... 48 Table 8. Usual values of the braking force cœfficients (longitudinal cœfficient) ...... 59 Table 9. Driver perception and reaction as a function of road surface wetness and visibility ...... 61 Table 10. Measuring methods for measuring the intensity of precipitation ...... 64 Table 11. Measuring method for measuring the type of precipitation ...... 66 Table 12. Measuring method for measuring the water film thickness ..... 68 Table 13. Measuring method for measuring the condition of the road surface ...... 69 Table 14. Effect of snow/ice on tyre/road friction ...... 71 Table 15. Accident rates at different intervals ...... 73 Table 16. Requirement for freezing point temperatures ...... 79 Table 17. Available quality levels for the various monitoring paramaters ...... 87 Table 18. US road weather management program case studies in 21 US states ...... 90 Table 19. Road pilot (Austria) characteristics ...... 92 Table 20. Danish RWIS characteristics ...... 95 Table 21. Finnish RWIS characteristics ...... 102 Table 22. French weather stations categories ...... 105 Table 23. French Météo-France weather stations characteristics ...... 107 Table 24. Polish RWIS characteristics ...... 115

© Les collections de l’INRETS 131

References

AASHTO, (2004). A Policy on Geometric Design of Highways and . American Association of State Highway and Transportation Officials (AASHTO). 5th edition, Washington, USA. Abdel-Aty M.A. and R. Pemmanaboina, (2007). Calibrating a Real-Time Traffic Crash Prediction Model Using Archived Weather and ITS Traffic Data. IEEE Transactions on ITS, Vol. 7, n°2, pp. 167-174. Aguero-Valverde J., Jovanis P.P., (2006). Spatial analysis of fatal and injury crashes in Pennsylvania. Accident Analysis and Prevention 38, pp. 618-625. Al-Ghamdi A.-S., (2004). « Experimental Evaluation of Fog Warning System ». TRB 2004, Annual Meeting, CD-ROM. Al-Kaisy A., Freedman Z., (2006). Weather-Responsive Signal Timing. In Transportation Research Record: Journal of the Transportation Research Board, No. 1978, Transportation Research Board of the National Academies, Washington, D.C, pp. 49-60. Andersson Anna, (2010). Winter Road Conditions and Traffic Accidents in Sweden and UK. Present and Future Climate Scenarios. Doctoral thesis A131, the University of Gothenburg. Andreescu M-P. and Frost D. B., (1998). Weather and traffic accidents in Montreal, Canada. Clim., Res., 9, pp. 225-230. Andrey J., B. Mills, M. Leahy and J. Suggett., (2003). Weather as a Chronic Hazard for Road Transportation in Canadian Cities. Natural Hazards, Vol. 28, pp. 319-343. Aron M., Seidowsky R., (2008). “Using traffic counts to estimate lane changes and safety indicators”, in CD Rom Transport Research Arena TRA08 session 13.4.3 paper 0273, 7 pages, Ljubjana, Slovenia. Baker T.K., Falb, T., Voas R., Lacey J., (2003). Older women drivers: Fatal crashes in good conditions. Journal of Safety Research 34, pp. 399-405. Ben Aissa, (2007). Ben-Aissa A., (2007). Estimation et prévision des temps de parcours sur autoroutes. Phd Thesis Lyon University. Bijleveld F., Churchill T., (2009). The influence of weather conditions on road safety, September SWOV Institute for Road Safety Research, Leidschendam. Brijs T., Karlis D., Wets G., (2008). Studying the effect of weather conditions on daily crash counts using a discrete time-series model. Accident Analysis and Prevention 40, pp. 1180-1190.

© Les collections de l’INRETS 133 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Brijs T., Vlassenroot, S., Broekx S., De Mol J., Panis L.I., Wets G., (2007). Driving with intelligent speed adaptation: Final results of the Belgian ISA-trial. Transportation Research Part A: Policy and Practice 41, pp. 267-279. Billot R., El Faouzi N.-E., De Vuyst F., (2009). Multi-level assessment of the rain impact on drivers‘ behaviours: standardized methodology and empirical analysis. In Proceedings of the 88nd annual meeting of the Transportation Research Board. CDROM. Transportation Research Board, Washington, D.C. Billot R., El Faouzi N.-E., Sau J. and De Vuyst F., (2008) How does Rain affect traffic indicators? Empirical study on a French interurban motorway. In Proceedings of the Lakeside Conference, Klagenfurt, Austria. Bowden F. P., (1953). Friction on Snow and Ice, Proceedings of the Royal Society of London. Series A, Mathematical and Physical Sciences, Vol. 217, No. 1131, pp. 462-478. Brilon W., and Ponzlet M., (1996). Variability of Speed-Flow Relationships on German Autobahns, Transportation Research Record 1555, pp 91-98. Buton J., N.-E. El Faouzi, (2010). Effect of weather on traffic demand. Technical report, INRETS-LICIT, 175 pages. Caliendo C., Guida M., Parisi A., (2007). A crash-prediction model for multilane roads. Accident Analysis and Prevention 39, pp. 657-670. Carlink Consortium, (2008). Intelligent Road Weather Forecasting in the “CARLINK” Platform. SIRWEC Road Weather Conference, Prague, Czech Republic, May 2008. Chang L.Y., Chen W.C., (2005). Data mining of tree-based models to analyze freeway accident frequency. Journal of Safety Research 36, pp. 365-375. Chung E., O. Ohtani H. Warita M., Kuwahara and H. Morita, (2005). Effect of Rain on Travel Demand and Traffic Accident, 8th International IEEE Conference on Intelligent Transportation Systems, 10. Chung E., Ohtani O. and Kuwahara M., (2005). Effect of rainfall on travel time and travel demand. 5th ITS European Congress, Hannover, Germany. Chung E., Ohtani O. Warita H., Kuwahara M. and Morita H., (2006). Does rainfall affect highway capacity? 5th TRB International Symposium on Highway Capacity and Quality of Service, Yokohama, Japan. Cools M., Moons, E., Wets G., (2008). Assessing the impact of weather on traffic intensity. In Proceedings of the 87nd annual meeting of the Transportation Research Board. CDROM. Transportation Research Board of the National Academies, Washington, D.C. COST-727, (2006). Atmospheric Icing on Structures: Measurements and data collection on icing: State of the Art. Publication of MeteoSwiss, 75, 110 pages.

134 © Les collections de l’INRETS References

COST 344, (2002). Improvements to Snow and Ice Control on European Roads and Bridges - http://cordis.europa.eu/cost-transport/src/cost-344.htm Cypra T., (2007). Entwicklung einer Entscheidungsmethode fur Mass-nahmen im Winterdienst auf hochbelasteten Bundes-autobahnen (Development of a decision method for winter maintenance measures on heavily travelled motorways), Doctoral thesis, published in: Veröffentlichungen des Instituts fur Strassen- und Eisenbahnwesen der Universität Karlsruhe (TH), Heft 56, Karlsruhe 2007. Daniel, Janice R., Chien, Steven I., (2009). Impact of Adverse Weather on Freeway Speeds and Flows. In Proceedings of the 88nd annual meeting of the Transportation Research Board. CDROM. Transportation Research Board of the National Academies, Washington, D.C. Datla S., Sharma S., (2008). Impact of cold and snow on temporal and spatial variations of highway traffic volumes, Journal of Transport Geography 16, pp. 358–372. Delanne Y., Travert P., (1997). Accident rates and road surface skidding properties: a literature survey – Paper 97SAF015, 30th International Symposium on Automotive Technology & Automation (ISATA), Road and vehicle safety. Delanne Y., Violette E., (2001). Wet road accidents - Driver behaviour, safety margins and road design, Traffic Safety on Three Continents, Moscow. Do M - T., (2009). SARI Project (Surveillance Automatisée de la Route pour l’information des conducteurs et des gestionnaires) Thème 3 : IRCAD, Final Report http://www.recherche-innovation.equipement.gouv.fr/ IMG/pdf/Theme_3_IRCAD_Rapport_final_cle54ec18.pdf Dinkel A., Leonhardt A.; Badelt H., (2008). Fusion of xFCD and local road weather data. WIRELESSCOM (Hrsg.), SIRWEC 2008, 14th International Road Weather Conference Prague - Abstract proceedings, ISBN 978-80-87205-01-3. Dinkel A.; Piszczek S.; Glas F.; Mutzbauer J. (2008). Test Site "Eching Ost" - evaluation of environmental sensors to improve traffic control and safety, The Lakeside Conference - Safety in Mobility. Dinkel A.; Leonhardt A.; Badelt H. (2008). Fusion of xFCD and local road weather data. WIRELESSCOM (Hrsg.), SIRWEC 2008, 14th International Road Weather Conference, Prague. Dumont A.-G., Tille M., (2008). Conception des voies de circulation, EPFL. Dumont A.-G., (2006). Réalisation des infrastructures de transport, EPFL. Edwards J.B., (1996). Weather-related road accidents in England and Wales: a spatial analysis. Journal of Transport Geography 4 (3), pp. 201-212.

© Les collections de l’INRETS 135 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Edwards J.B., (1998). The Relationship Between Road Accident Severity and Recorded Weather. Journal of Safety Research 29 (4), pp. 249-262. Edwards J.-B., (1999). Speed adjustment of motorway commuter traffic to inclement weather, Transportation Research Part F 2, pp. 1-14. Edwards J.B., (1999). The temporal distribution of road accidents in adverse weather. Meteorol. Appl. 6, pp. 59–68. Eisenberg D., (2004). The mixed effects of precipitation on traffic crashes, Accident Analysis & Prevention Volume 36, Issue 4, pp. 637-647. El Faouzi N.-E., O. De Mouzon and R. Billot, (2008). Toward Weather- Responsive Traffic Management on French Motorways, In proceedings of the fourth national conference on surface transportation weather, Transportation Research E-Circular Issue Number: E-C126, Indianapolis, pp. 443-456. Euroregional Monitoring Expert Group, Data Quality Aspects in Euroregional Projects, TEMPO, DG TREN, 2006. Euroregional Monitoring Expert Group, Data Quality Aspects in Euroregional projects, V 1.0 01/06/2006, TEMPO Programme, MIP 2005, DG TREN. FGSV AK 3.2.1, (2009). “Umfelddatenerfassung in Verkehrsbeeinflussungs- anlagen”, Köln, draft version. FHWA,-(2010). http://www.ops.fhwa.dot.gov/weather/index.asp. Federal Highway Administration (FHWA) web site (consulted at February 2010). USA. FHWA. Road Weather Management Overview. http://ops.fhwa.dot.gov/ Weather/ overview.htm. Accessed March, 24th, 2009. FHWA,-(2008). Integration of Weather Information in Transportation Management Center Operations: Self-Evaluation and Planning Guide. Federal Highway Administration (FHWA), Report No. FHWA-JPO-08- 057. Washington. FHWA,-(2006). Integration of Emergency and Weather Elements into Transportation Management Centers. Federal Highway Administration (FHWA), Publication No. FHWA-HOP-06-090. Washington. FHWA,-(2005). Road Weather Information System Environmental Sensor Station Siting Guidelines. Federal Highway Administration (FHWA), Publication No. FHWA-HOP-05-026. Washington. Final Report of COST Action 353, (2008). Winter Service Strategies for Increased European Road safety.

136 © Les collections de l’INRETS References

Fridstrøm L., Ifver J., Ingebrigtsen S., Kulmala R., Thomsen L.K., (1995). Measuring the Contribution Of Randomness, Exposure, Weather, And Daylight To The Variation In Road Accident Counts. Accident Analysis & Prevention 27 (1), pp. 1-20. Fridstrom L., Ingebrigtsen S., (1991). An aggregate accident model based on pooled, regional time-series data. Accident Analysis and Prevention 23(5), pp. 363-378. Galin D., (1981). Speed on Two Lane Rural Roads – A Multiple Regression Analysis », Traffic Engineering and Control, N°.22, pp. 453-460. Geurts K., Thomas I., Wets G., (2005). Understanding spatial concentrations of road accidents using frequent item sets. Accident Analysis and Prevention 37, pp. 787-799. Gillam W. & R. Withill, (1992). UTC and Inclement Weather Conditions, Leicestershire County Council & the University of Nottingham in the United Kingdom, presented at the Institute of Electrical and Electronics Engineers Conference, pp. 85-88. Golden J.M., (1980). A Theory of wet road-tyre friction. Wear 71. Golob T. F. and Recker W. W., (2003). Relationships Among Urban Freeway Accidents, Traffic Flow, Weather, and Lighting Conditions, Journal Of Transportation Engineering, 129(4), page 342. Golob T.F., Recker, W.W., (2004). A method for relating type of crash to traffic flow characteristics on urban freeways. Transportation Research Part A 38, pp. 53-80. Goodwin L.C., Pisano P., (2003). Best Practices for Road Weather Management, Road Weather Management Program, Office of Transportation Operations, Federal Highway Administration. Goodwin L.-C., (2002). Analysis of Weather-Related Crashes on U.S. Highways, Mitretek Systems. Goodwin L.-C., (2002). Weather Impacts on Arterial Traffic Flow, Mitretek Systems. December 24, 2002. Hall F., Barrow D., (1988). The Effect of Weather on the Relationship Between Flow and Occupancy on Freeways. In Transportation Research Record, No. 1194, TRB, National Research Council, Washington, D.C., pp. 55-63. Hanbali R. M. and D. A. Kuemmel, (1992). Traffic Volume Reductions Due to Winter Storm Conditions, Transportation Research Record 1387, TRB, National Research Council, Washington D.C. Hassan Y.A., Barker J.J., (1999). The impact of unseasonable or extreme weather on traffic activity within Lothian region, Scotland. Journal of Transport Geography 7 (3), pp. 209–213.

© Les collections de l’INRETS 137 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Hippi M., Juga I., P. Nurmi, (2010). A statistical forecast model for road surface friction, 15th International Road Weather Conference, Quebec City, Canada. Hippi M., P. Nurmi and P. Saarikivi. Development Project ColdSpots: Towards More Detailed Road Condition Forecasts, SIRWEC Road Weather Conference, Prague, Czech Republic, May 2008. Hogema J.-H., (1996). Effect of Rain on Daily Traffic Volume and on Driving Behaviour, Report TM-96-B019. TNO, Netherlands, 1996. Hranac R., E. Sterzin, D. Krechmer, H. Rakha, M. Farzaneh, (2006). Empirical Studies on Traffic Flow in Inclement Weather, Publication No. FHWA-HOP-07-073 October. http://www.standards.its.dot.gov/fact_sheet.asp?f=24 Ibrahim A. and Hall, F., (1994). Effect of Adverse Weather Conditions on Speed-Flow-Occupancy Relationships. In Transportation Research Record, No. 1457, TRB, National Research Council, Washington, D.C., pp. 184-191. Intelligent Roads (INTRO), (2005). Performance Indicators needs and derivation: Single and multisource solutions., Deliverable 4.1., INTRO project, DG TREN, FP6-012344. Ivey D.L., Griffin L., Newton T.M., Lytton R., Hankin, K.C., (1981). Predicting Wet Weather Accidents. Accident Analysis and Prevention 13, pp. 83-99. Jones E. R., M. E. Goolsby and K. A. Brewer, (1970). The Environmental Influence of Rain on Freeway Capacity. In Highway Research Record, No. 321, HRB, National Research Council, Washington, D.C., pp. 74–82. Jacot A., Lindenmann H.P., Seiler L., (2007). Grundlagen zur Revision der Griffigkeitsnormen. FEDRO report n°1202. Juga Ilkka and Marjo Hippi. Snowfall induced severe pile-ups in southern Finland on 17 March 2005. EMS Annual Meeting Abstracts, Vol. 6, EMS2009-389, 2009. http://meetingorganizer.copernicus.org/EMS2009/poster_programme/1684 Karkowski M. et al. (2006). Integration of weather effects for traffic indicators forecasting. Deliverable 4.2 of INTRO (Intelligent Roads), Fehrl/IBDiM, Poland. Keay and Simmonds, (2006). Road accidents and rainfall in a large Australian city. Accident Analysis and Prevention, Vol. 38, pp. pp. 445-454. Keay K. and Simmonds I., (2005). The association of rainfall and other weather variables with road traffic volume in Melbourne, Australia. Accident Analysis and Prevention, 37 pp. 109-124.

138 © Les collections de l’INRETS References

Khan G., Qin X Noyce, D.A., (2008). Spatial Analysis of Weather Crash Patterns. Journal of Transportation Engineering, May 2008, pp. 191-202. Khattak A.-J, De Palma A., (1997). The Impact of Adverse Weather Conditions on the Propensity to Change Travel Decisions: A Survey of Brussels Commuters, Transp. Res. A, Vol. 31, N° 3, pp. 181 -203. Khattak A.-J, Kantor P., Council F.-M., (1998). Role of Adverse Weather in Key Crash types on Limited-Access Roadways – Implications for Advanced Weather Systems, Transportation Research Record 1621. Knapp Keith K., Leland D. Smithson, and A. J. Khattak, (2000). The Mobility and Safety Impacts of Winter Storm Events in a Freeway Environment. Mid-Continent Transportation Symposium Proceedings. Knapp Keith K., Leland D. Smithson, and Aemal J. Khattak, (2000). The use of multiple data sources to evaluate the volume and safety impacts of winter storm events. ITE Annual Meeting at Nashville, Tennessee, Institute of Transportation Engineers, Washington D.C., August 6-8, 2000, page 11. Koetse M.J., Rietveld P., (2009). The impact of climate change and weather on transport: An overview of empirical findings. Article in press, Transportation Research Part D. Knoblauch R, Pietrucha M, Nitzburg M., (1996). Field studies of pedestrian walking speed and start-up time. Transportation Research Record, 1438, pp. 67-73. Kulmala R. & Karhumäki T., (2006). VIKING Monitoring Guidelines 2005. Euro- regional project VIKING, MIP 2005 Deliverable, version 1.0 March 2006. Kulmala R. & Karhumäki T., (2006). VIKING Monitoring Plan 2005. Euro- regional project VIKING MIP2005 Deliverable, version 1.0 March 2006. Kyte M., Khatib Z., Shanon P., Kitchener F., (2007). Effect of Weather on Free- Flow Speed., Transportation Research Record N°1776. Lähesmaa J. & Levo J., (2003). Tiesääseurannan tavoitetila. (Objective state of road weather monitoring). Kouvola 2003. Finnish Road Administration. 39 p. + 6 apps. Lamm R., Choueirie M. and Mailaender T., (1990). Comparison of Operating Speeds on Dry and Wet Pavements of Two-Lane Rural Highways. Transportation Research Record: Journal of the Transportation Research Board, No. 1280, TRB, National Research Council, Washington, D.C., pp. 199-207. Le Breton P., (1990). Study of the rain impact on road safety (main roads and motorways). Statistical analysis. Note d’information SETRA-CSTR N° 77.

© Les collections de l’INRETS 139 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Lee C., F. Saccomanno and B. Hellinga, (2002). Analysis of Crash Precursors on Instrumented Freeways. In Transportation Research Record 1784, TRB, National Research Council, Washington D.C., Paper n°02 - 3790. Liang W.-L., Kyte M., Kitchener F., and Shannon P., (2000). Effect of Environment Factors on Driver Speed., Transportation Research Record 1635, pp 155-161. Livet J., Pitre R., (2001). Capteurs routiers équipant les stations météo- routières – Evaluation de la qualité des informations des états de surface « sec », « humide » et « mouillé », APTP project, Research report (in French). Lin Q. and Nixon W., (2008). Effects of Adverse Weather on Traffic Crashes: Systematic Review and Meta-Analysis. In Proceedings of the 87nd annual meeting of the Transportation Research Board. CDROM. Transportation Research Board of the National Academies, Washington, D.C. Majdzadeh R., Khalagi, K., Naraghi, K., Motevalian, A., Eshraghian, M.R., (2008). Determinants of traffic injuries in drivers and motorcyclists involved in an accident. Accident Analysis and Prevention 40, pp. 17-23. Maki, P., (1999). Adverse Weather Traffic Signal Timing, Short Elliott Hendrickson http://www.trafficware.com/documents/1999/00005.pdf Martin P., et al. (2000). Inclement Weather Signal Timings, University of Utah Traffic Lab, 2000, http://www.ndsu.nodak.edu/ndsu/ugpti/MPC_Pubs/ pdf/MPC01-120.pdf May A.-D., (1998). Capacity and Level of Service for Freeway Systems, Third Interim Report. Maze T. H., Agarwal M., Burchett G., (2006). Whether Weather Matters to Traffic Demand, Traffic Safety, and traffic operations and flow, In Transportation Research Record, No. 1948, TRB, National Research Council, Washington, D.C, pp. 170-176. http://www.intrans.iastate.edu/reports/whether_weather.pdf Martin P. T., Perrin, J., Hansen, B. and Quintina, I., (2000). Inclement Weather for Signal Timings, MPC Report No. 01-120., 2000. Musk L., (1991). Climate as a factor in the planning and design of new roads and motorways, Highway Meteorology, edited by A.H. Perry and L.J.Symons, E&FN Spon, London. NF, P99-320, (1998). Météorologie routière – Recueil des données météo- rologiques et routières. NCHRP, (2009). Guide for Pavement Friction. Final Report for NCHRP Project 01-43.

140 © Les collections de l’INRETS References

Norrman J., Eriksson M. and Lindqvist S., (2000). Relationships between road slipperiness, traffic accident risk and winter road maintenance activity. Clim. Res., 15, pp. 185-193. Olander J., (2009). Real-time Traffic Management Communications in Sweden. TR News, 265 November-December. ONISR, (2005). The major themes of road safety - results 1995-2004. (in French) http://www.securiteroutiere.equipement.gouv.fr/IMG/ Synthese/Pluie.pdf, accessed July 25. 2006. Pai Chih-Wei and Saleh Wafaa, (2008). Modelling motorcyclist injury severity resulting from sideswipe collisions at T-junctions in the United Kingdom: new insights into the effects of manoeuvres, International Journal of Crashworthiness, 13(1), pp. 89-98. Pai C-W., Saleh W., (2008). Modelling motorcyclist injury severity by various crash types at T-junctions in the UK. Safety Science 46, pp. 1234-1247. Perrin J., P. Martin, (2002). Modifying Signal Timing during Inclement Weather, University of Utah Traffic Lab, presented at the 2002 Institute of Transportation Engineers Annual Meeting. Pinto J. G., Brucher T., Flink A. H. and Kruger A., (2007). Extraordinary snow accumulations over parts of central Europe during the winter of 2005/06 and weather-related hazards. Weather, 62, pp. 16-21. Pirkko S., Pirkko M., Sipilä P., Nurmi M., Hippi. Project ColdSpots: A new way to improve winter oad condition forecasts, XIII International SIRWEC Road Weather Conference, Torino, Italy, March 2006. Pisano P.-A., and Goodwin L.-C., (2004). Research Needs for Weather- Responsive Traffic Management, TBR 2004 Annual Meeting. Rakha H., Farzaneh M., Arafeh M. and Sterzin E., (2008). Inclement Weather Impacts on Freeway Traffic Stream. In Proceedings of the 87nd annual meeting of the Transportation Research Board. CDROM. Transportation Research Board of the National Academies, Washington, D.C. Rauhala Jenni and Ilkka Juga, (2010). Wind and snow storm impacts on society. SIRWEC 15th International Road Weather Conference, Quebec City, Canada, 5-7 February 2010. Rizenbergs R.L., Burchett J.L., Warren A., (1977). Relation of accidents and pavement friction on rural, two-lane Roads, Transportation Research Record n° 633, Transportatio Research Boar d, Washington, pp. 21-27. RWIS Web Guide, (2009): http://www.sirwec.org/en/rwis_web_guide.pdf

© Les collections de l’INRETS 141 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Salli, Riikka, Maiju Lintusaari, Hanne Tiikkaja and Markus Pöllänen, (2008). Wintertime road conditions and accident risks in passenger car traffic. Tampere University of Technology. Transport systems. Research Report 68. ISBN 978-952-15-1965-9 (in Finnish, abstract in English). Saint Pierre G., Aron, M., Bergel R., Violette E., (2007). Rain reconstruction from various weather-related data sets using logistic regression: methodology and applications, TRB, Washington , paper 07-0916, January 2007. Schlösser L.H.M., (1976). Traffic accidents and road surface skidding resistance, Transportation Research Record n° 623, Transportation Research Board, Washington, pp. 11-20. Shankar V.N. et al. (2004). Impacts of Design, Traffic, Weather, and Related Interactions on Roadside Crashes. In Transportation Research Record 1897, TRB, National Research Council, Washington D.C., pp. 156-163. Shankar, V.N et al. (1995). Effect of roadway geometrics and environmental factors on rural freeway accidents frequencies. Accident Analysis and Prevention, 27 n°3 pp. 371-389. Sherretz L.A., Farhar B.C., (1978). An analysis of the relationship between rainfall and the occurrence of traffic accidents, Journal of Applied Meteorology, Vol 17, American Meteorological Society, pp. 711-715. Sihvola N., Rämä P. and Juga I., (2008). Determining the successfulness of the road weather information service in the winter seasons 1997-2007. Finnish Road Administration 15/2008. 87 p. + app. 18 p. 1457-9871, 3201095 (in Finnish, abstract in English). Stephenson D.B., (2008). Definition, diagnosis, and origin of extreme weather and climate events. In: “Climate Extremes and Society”, R. Murnane and H. Diaz (Eds), Cambridge University Press, pp. 348. Stern A. D., V. Shah, L. C. Goodwin, and P. A. Pisano, (2003). Analysis of Weather Impacts on Traffic Flow in Metropolitan Washington, D.C.” 19th Conf. on IIPS, Long Beach, CA, American Meteor. Soc. Tanner J.C., (1952). Effect of Weather on Traffic Flow. Nature, 4290 p. 107. TLS, (2002). Technische Lieferbedingungen für Streckenstationen, issued by Bundesanstalt für Straßenwesen, Bergisch Gladbach. TYROSAFE, (2009). Report on different parameters influencing skid resistance, rolling resistance and noise emissions. Deliverable 10. Unrau D., Andrey J., (2006). Driver response to rainfall on urban expressways. In Transportation Research Record: Journal of the Transportation Research Board, No. 1980, Transportation Research Board of the National Academies, Washington, D.C. pp. 24-30.

142 © Les collections de l’INRETS References

Violette E. et al. (2002). Accident Par Temps de Pluie. Etude PREDIT 2 APTP . Rapport de synthèse. CETE Normandie-Centre, LCPC, INRETS, SERA, METEO France, PSA. Violette E. et al. (2009). Programme PREDIT. Projet IRCAD-SARI., Evaluation des expérimentations IRCA-SARI, Deliverable 4.1, LCPC, Nantes. Wallman Carl-Gustaf and Henrik Åström, (2001). Friction measurement methods and the correlation between road friction and traffic safety. A literature review. VTI meddelande 911A, 2001. Welleman A.G., (1978). Water nuisance and road safety, OECD Symposium on road drainage, pp. 82-95. Young R.K., Liesman, J., (2007). Estimating the relationship between measured wind speed and overturning truck crashes using a binary logit model. Accident Analysis and Prevention 39, pp. 574-580. Zhang L., Prevedouros, P.-D., (2005). Motorist Perception on the Impact of Rainy Conditions on Driver Behaviour and Accident Risk, TRB 2005 Annual Meeting CD-ROM.

© Les collections de l’INRETS 143

Publication data form

Research unit Project N° INRETS RESEARCHES LICIT TU0702 Ref.: R283 Title Real-time monitoring, surveillance and control of roads networks under adverse weather conditions Subtitle Language State of the art and best practices English Editor: Nour-Eddin El Faouzi Affiliation INRETS/ENTPE Sponsor, co-editor, name and address COST, 149 Avenue Louise, 1050 Brussels, Belgium Publication date December 2010 Summary This State of the Art report summarizes the work done within the COST Action TU0702 “Real-time monitoring, surveillance and control of road networks under adverse weather conditions”. It gives an overview about the effects of inclement weather conditions on road traffic operations and road safety as well as the best practices which are available in various countries. Several projects carried out in the last years on this topic (especially in Austria, Denmark, Finland, France, Germany, Greece, Poland and USA) are described. This overview shows that a number of relevant projects are on-going and are widely spread among countries, although the research is fragmented and, in most cases, publication and distribution of most findings is at the national level only. Coordination at the European level has only recently emerged, through initiatives such as this COST Action. However, the importance of the coordination of research and development has been recognized and collaborative projects at European or even at international levels (mainly with Australia and Japan) are being set up. The next step could logically be the integration of weather forecasting and traffic management capabilities together in an integrated framework that captures the effect of weather and weather-related strategies on traffic system performance. Key Words: traffic, pavement, weather, adverse weather, RWIS, traffic modelling, skid resistance, pavement condition Nbr of pages Price Bibliography 146 45 € Yes

© Les collections de l’INRETS 145 Real-time monitoring, surveillance and control of road networks under adverse weather conditions

Fiche bibliographique

Unité de recherche Projet n° RECHERCHES INRETS LICIT TU0702 Réf. : R283 Titre Real-time monitoring, surveillance and control of roads networks under adverse weather conditions Sous-titre Langue State of the art and best practices Anglais Auteur : Nour-Eddin El Faouzi Rattachement ext. INRETS/ENTPE Nom adresse financeur, co-éditeur COST, 149 Avenue Louise, 1050 Bruxelles, Belgique Date de publication Décembre 2010 Résumé Cet ouvrage résume le travail effectué au sein de l’action COST TU0702 "Gestion temps réel et régulation des réseaux routiers lors de conditions météorologiques défavorables". Il dresse une synthèse sur les effets des conditions météorologiques défavorables sur la circulation et la sécurité routières, il rapporte les meilleures pratiques de différents pays (en particulier en Autriche, Danemark, Finlande, France, Allemagne, Grèce, Pologne et États-Unis). Cet ouvrage recense également un certain nombre de projets centrés sur ce thème. Une des particularités de ce recensement est le caractère fragmenté de l’effort de recherche ; en effet dans la plupart des cas, les publications des résultats ne dépassent pas le niveau régional, au mieux le niveau national. La coordination européenne n'est apparue que récemment grâce à des initiatives telles que l'action COST TU0702. La prochaine étape devrait logiquement être l'intégration des prévisions météorologiques dans les outils de gestion du trafic dans un cadre intégré qui prend en compte les effets de la météorologie et des stratégies associées sur les performances du système de la circulation. Mots clés : trafic, circulation, chaussée, météorologie, conditions météo- rologiques défavorables, stations météo routières, modélisation du trafic, adhérence, état de la chaussée Nbre de pages Prix Bibliographie 146 45 € Oui

Imprimé en france – SEVEN, 17 rue de Gerland – 69007 Lyon Dépôt légal décembre 2010

146 © Les collections de l’INRETS