Analysis of Mobility and Traffic Safety with Respect to Changes in Volumes; Case Study: Stockholm, Sweden

Full text

(1)LiU-ITN-TEK-A--21/053-SE. Analysis of Mobility and Traffic Safety with Respect to Changes in Volumes; Case Study: Stockholm, Sweden Sofia Johansson Sri Vasireddy 2021-06-21. Department of Science and Technology Linköping University SE-601 74 Norrköping , Sw eden. Institutionen för teknik och naturvetenskap Linköpings universitet 601 74 Norrköping.

(2) LiU-ITN-TEK-A--21/053-SE. Analysis of Mobility and Traffic Safety with Respect to Changes in Volumes; Case Study: Stockholm, Sweden The thesis work carried out in Transportsystem at Tekniska högskolan at Linköpings universitet. Sofia Johansson Sri Vasireddy Norrköping 2021-06-21. Department of Science and Technology Linköping University SE-601 74 Norrköping , Sw eden. Institutionen för teknik och naturvetenskap Linköpings universitet 601 74 Norrköping.

(3) Upphovsrätt Detta dokument hålls tillgängligt på Internet – eller dess framtida ersättare – under en längre tid från publiceringsdatum under förutsättning att inga extraordinära omständigheter uppstår. Tillgång till dokumentet innebär tillstånd för var och en att läsa, ladda ner, skriva ut enstaka kopior för enskilt bruk och att använda det oförändrat för ickekommersiell forskning och för undervisning. Överföring av upphovsrätten vid en senare tidpunkt kan inte upphäva detta tillstånd. All annan användning av dokumentet kräver upphovsmannens medgivande. För att garantera äktheten, säkerheten och tillgängligheten finns det lösningar av teknisk och administrativ art. Upphovsmannens ideella rätt innefattar rätt att bli nämnd som upphovsman i den omfattning som god sed kräver vid användning av dokumentet på ovan beskrivna sätt samt skydd mot att dokumentet ändras eller presenteras i sådan form eller i sådant sammanhang som är kränkande för upphovsmannens litterära eller konstnärliga anseende eller egenart. För ytterligare information om Linköping University Electronic Press se förlagets hemsida http://www.ep.liu.se/ Copyright The publishers will keep this document online on the Internet - or its possible replacement - for a considerable time from the date of publication barring exceptional circumstances. The online availability of the document implies a permanent permission for anyone to read, to download, to print out single copies for your own use and to use it unchanged for any non-commercial research and educational purpose. Subsequent transfers of copyright cannot revoke this permission. All other uses of the document are conditional on the consent of the copyright owner. The publisher has taken technical and administrative measures to assure authenticity, security and accessibility. According to intellectual property law the author has the right to be mentioned when his/her work is accessed as described above and to be protected against infringement. For additional information about the Linköping University Electronic Press and its procedures for publication and for assurance of document integrity, please refer to its WWW home page: http://www.ep.liu.se/. © Sofia Johansson, Sri Vasireddy.

(4) Abstract The growing population and motorization generate more movements. In many cities, the increase of population and motorization is much greater than the development of the capacity of the transportation network. For unprotected road users, the risk of getting in a traffic accident increases and the risk of being more severely injured in an accident. In March 2020, a pandemic was declared because of a Coronavirus. More people started to work/study from home to prevent the virus from spreading by avoiding unnecessary trips, gatherings, and crowded areas. Therefore, travel behaviours have shifted during the pandemic compared to previous years. This project aims to get knowledge of how mobility and traffic accidents are affected by significant shifts of travel flow, predict the effect of traffic accidents based on mobility, and evaluate the risk of travelling on a particular road segment. Mobility data has been collected from Google Mobility, Apple Mobility, the Environmental Barometer, Trafikkontoret, and traffic accident data collected from STRADA. Mobility and traffic accident data have been analysed using Excel, QGIS, and PTV Visum Safety. The accident rates have been calculated to determine if the accident rate has changed during the pandemic, and three scenarios for 2021 have been predicted. A risk analysis model has also been used to calculate the risk of being involved in an accident on particular streets using a car, cycle, or walking. It was found that mobility has decreased, and the usage of transportation modes has shifted. During the pandemic, it has been more popular to cycle, which is also reflected in the traffic accident data, where the percentages of cyclists being involved in traffic accidents have increased. No matter the degree of injury or transportation mode, the total number of traffic accidents had decreased in 2020. However, the number of severe accidents is almost the same as in previous years. Males are overrepresented in traffic accidents, and the differences are even more significant for 2020. In 2020, the travel speed on the roads increased, which might be due to decreased traffic volume, making it possible to drive faster. The percentage of accidents involving alcohol also increased. The results from the risk analysis show whether the risk of getting into a traffic accident on a particular street using a specific transportation mode has increased or decreased depending on the street and transportation mode. Three scenarios (better, same, worse) of the risk of travelling on the particular road stretches in 2021 have also been calculated. To make better predictions, additional years should be considered. For future work, it would also be interesting to consider the weather since it greatly impacts the transportation mode used and the risk of accidents. The infrastructure was not considered in this project, which would be interesting since the transportation mode, route used, and speed might depend on the infrastructure and current constructions. Keywords: Accident Rate; Mobility; Pandemic; Risk Analysis; Traffic Accidents; Traffic Flow; Traffic Safety; Unprotected Road Users. i.

(5) Acknowledgement First of all, we want to sincerely thank our supervisor Ghazwan Al-Haji at Linköping University, for giving us the opportunity to conduct this project and for all the support and interesting discussions during the project. We also want to thank our examiner Clas Rydergren at Linköping University, for the support and valuable input during the project. Furthermore, we want to thank our contact persons at STRADA and Trafikkontoret providing us with data and quickly responding to our questions. We also want to thank contact persons at PTV Group for guidance and for helping us with issues regarding software. We also want to thank the Technical Staff Support at Linköping University guidance and help during the project.. for the the for. Finally, we want to thank our families, classmates, and everybody involved for the support and motivation during this project, which is the grand finale of the Master of Science program in Intelligent Transport Systems and Logistics at Linköping University.. Sofia Johansson and Sri Vishwanth Chowdary Vasireddy Norrköping, 2021-06-24. ii.

(6) Table of Contents 1. INTRODUCTION. 1. 1.1 BACKGROUND 1.2 AIM & PURPOSE 1.3 RESEARCH QUESTIONS 1.4 LIMITATIONS 1.5 OUTLINE. 1 3 3 3 4. 2. METHOD. 5. 2.1 COLLECTING DATA 2.1.1 Traffic Accident Data 2.1.2 Mobility Trends 2.2 ANALYZING DATA 2.2.1 Excel 2.2.2 QGIS 2.2.3 PTV Visum Safety. 5 6 6 7 7 8 8. 3. CURRENT AND PREVIOUS STUDIES ON MOBILITY AND TRAFFIC SAFETY 10 3.1 MOBILITY 3.2 TRAFFIC SAFETY 3.2.1 Improving Traffic Safety 3.2.2 Traffic Safety in Stockholm, Sweden 3.2.3 Risk Compensation 3.2.4 Accident Prediction Models 3.2.5 Loss of Statistics 3.3 TRAFFIC ACCIDENTS DURING THE PANDEMIC. 10 11 11 12 12 13 16 17. 4. STATISTICS REGARDING MOBILITY AND TRAFFIC ACCIDENTS 4.1 MOBILITY CHANGES DURING 2020 4.1.1 Purpose of Trips 4.1.2 Transportation Mode 4.1.3 Traffic Volume 4.1.4 Violations in Traffic 4.2 TRAFFIC ACCIDENTS IN STOCKHOLM 2018-2021 4.2.1 Degrees of the Traffic Accidents 4.2.2 Years of Life Lost & Years of Life with Disability 4.2.3 Type of Traffic Accidents 4.2.4 Transportation Mode Involved in the Traffic Accidents 4.2.5 Age of the People Involved in the Traffic Accidents 4.2.6 Gender of the People Involved in the Traffic Accidents 4.3 TRAFFIC ACCIDENTS VISUALIZATION IN PTV VISUM SAFETY. 5. RISK ANALYSIS AND ACCIDENT PREDICTIONS 5.1 METHOD FOR RISK ANALYSIS 5.2 ACCIDENT RATES AND PREDICTIONS 5.3 MODEL ACCURACY VERIFICATION. 18 18 18 19 20 22 26 27 28 30 39 43 48 53. 57 57 58 65. iii.

(7) 6. CONCLUSION. 66. 6.1 FUTURE STUDIES. 67. REFERENCES. 68. iv.

(8) List of Figures FIGURE 1.1, COMPARISON OF ROAD FATALITIES IN SWEDEN, TARGET VS ACTUAL (TRANSPORTSTYRELSEN, 2021) 2 FIGURE 1.2, COMPARISON OF ROAD TRAFFIC VOLUMES IN MILLION PASSENGER-KILOMETRES IN SWEDEN (STATISTA, 2021) 2 FIGURE 2.1, BLUEPRINT OF THE PROJECT 5 FIGURE 3.1, MOBILITY TRIANGLE IN 2020 (STOCKHOLMS STAD, 2020B) 11 FIGURE 4.1, THE PURPOSE OF TRIPS IN STOCKHOLM MUNICIPALITY 19 FIGURE 4.2, TRANSPORTATION MODE USED IN STOCKHOLM MUNICIPALITY 20 FIGURE 4.3, ROAD SEGMENTS (PURPLE AND BLUE LINES) WITH DATA REGARDING TRAFFIC VOLUME. THE LEFT MAP PRESENTS DATA FOR 2018 AND 2019 (PURPLE LINES), AND IN THE RIGHT MAP, DATA FOR 2020 IS INCLUDED (CYAN LINES). 21 FIGURE 4.4, TRAFFIC VOLUME ON PARTICULAR STREETS IN STOCKHOLM 22 FIGURE 4.5, CHANGE OF AVERAGE SPEED 2019-2020 23 FIGURE 4.6, PERCENTAGES OF THE ROAD SEGMENTS WHERE THE AVERAGE MEASURED SPEED IS HIGHER THAN THE SPEED LIMIT 23 FIGURE 4.7, ROAD SEGMENTS WHERE 15 % OF THE VEHICLES DRIVE AT LEAST 20 KM/H FASTER THAN THE SPEED LIMIT (BLUE: 2018/2019, RED: 2020) 25 FIGURE 4.8, THE SHARE OF PERSONS INVOLVED IN TRAFFIC ACCIDENTS THAT WERE UNDER THE INFLUENCE OF ALCOHOL AT THE TIME OF THE ACCIDENT 26 FIGURE 4.9, NUMBER OF ACCIDENTS IN STOCKHOLM 27 FIGURE 4.10, MONTHLY STATISTICS OF THE DEGREES OF THE ACCIDENTS FOR 2018,2019,2020,2021 (JAN & FEB) 28 FIGURE 4.11, YEARS OF LIFE LOST AND YEARS OF LIFE WITH DISABILITY 30 FIGURE 4.12, TOTAL ACCIDENTS CATEGORISED BASED ON “TYPE OF ACCIDENT.” 31 FIGURE 4.13, TOTAL ACCIDENTS BASED ON SINGLE VEHICLE ACCIDENTS. 32 FIGURE 4.14, TOTAL ACCIDENTS BASED ON HEAD-TO-HEAD, REAR - END COLLISIONS 33 FIGURE 4.15, TOTAL ACCIDENTS BASED ON MOTOR VEHICLE: INTERSECTION, PASSING, OVERTAKING 34 FIGURE 4.16, TOTAL ACCIDENTS BASED ON ANIMALS, PARKED & OTHER VEHICLE ACCIDENTS. 35 FIGURE 4.17, TYPE OF ACCIDENTS OCCURRED BASED ON DEGREE: SEVERE. 36 FIGURE 4.18, SEVERE SINGLE VEHICLE ACCIDENTS 36 FIGURE 4.19, SEVERE HEAD-TO-HEAD, REAR-END COLLISION. 37 FIGURE 4.20, SEVERE MOTOR VEHICLE: INTERSECTION, PASSING, OVERTAKING. 38 FIGURE 4.21, FATALITIES CATEGORISED BY TYPES OF ACCIDENTS. 39 FIGURE 4.22, TRANSPORTATION MODES INVOLVED IN TRAFFIC ACCIDENTS. 39 FIGURE 4.23, STATISTICS FOR BUS, PEDESTRIANS, CYCLISTS, AND TRUCKS INVOLVED IN TRAFFIC ACCIDENTS FOR 2018,2019,2020,2021 (JAN & FEB) 40 FIGURE 4.24, STATISTICS FOR MOTORCYCLISTS, MOPEDS, CARS, AND OTHER MODES INVOLVED IN TRAFFIC ACCIDENTS FOR 2018,2019,2020,2021 (JAN & FEB) 41 FIGURE 4.25, TRANSPORTATION MODES INVOLVED IN SEVERE TRAFFIC ACCIDENTS. 42 FIGURE 4.26, STATISTICS FOR CYCLISTS, PEDESTRIANS, CAR, AND BUS, MOTORBIKE, MOPED, TRUCK & OTHER MODES FOR 2018,2019,2020,2021 (JAN & FEB) 42 FIGURE 4.27, FATALITIES CATEGORISED BY MODE OF TRANSPORT. 43 FIGURE 4.28, AGE GROUPS INVOLVED IN TRAFFIC ACCIDENTS, YEARLY. 43 FIGURE 4.29, BABY, CHILDREN, YOUTH, YOUTH WITH FRESH DRIVING LICENSE INVOLVED IN TRAFFIC ACCIDENTS, MONTHLY FOR 2018,2019,2020,2021 (JAN & FEB) 44 FIGURE 4.30, AGE GROUPS (20-24, 25-34, 35-44, 45-54) INVOLVED IN TRAFFIC ACCIDENTS, MONTHLY FOR 2018,2019,2020,2021 (JAN & FEB) 44 FIGURE 4.31, AGE GROUPS (55-64, 65-74, 75+, UNIDENTIFIED) INVOLVED IN TRAFFIC ACCIDENTS, MONTHLY FOR 2018,2019,2020,2021 (JAN & FEB) 45 FIGURE 4.32, AGE GROUPS INVOLVED IN SEVERE TRAFFIC ACCIDENTS, YEARLY. 46 FIGURE 4.33, AGE GROUPS (7-14, 15-17, 18-19, 20-24) INVOLVED IN SEVERE TRAFFIC ACCIDENTS, MONTHLY FOR 2018,2019,2020,2021 (JAN & FEB) 46 FIGURE 4.34, AGE GROUPS 25-34, 35-44, 45-54, 55-64 INVOLVED IN SEVERE TRAFFIC ACCIDENTS, MONTHLY FOR 2018,2019,2020,2021 (JAN & FEB) 47 FIGURE 4.35, AGE GROUPS 65-74 AND 75+ INVOLVED IN SEVERE TRAFFIC ACCIDENTS, MONTHLY 47. v.

(9) FIGURE 4.36, FATALITIES CATEGORISED BY AGE 48 FIGURE 4.37, GENDER OF THE PERSONS INVOLVED IN TRAFFIC ACCIDENTS - YEARLY. 49 FIGURE 4.38, GENDER OF THE PERSONS INVOLVED IN TRAFFIC ACCIDENTS - MONTHLY 50 FIGURE 4.39, GENDER OF THE PERSONS INVOLVED IN SEVERE TRAFFIC ACCIDENTS - YEARLY 51 FIGURE 4.40, GENDER OF THE PERSONS INVOLVED IN SEVERE TRAFFIC ACCIDENTS MONTHLY. 52 FIGURE 4.41, FATALITIES CATEGORISED BY GENDER. 53 FIGURE 4.42, VISUALIZATION OF PEDESTRIAN ACCIDENTS (LEFT) AND TYPE OF COLLISIONS INVOLVED PEDESTRIAN (RIGHT) IN PTV 54 FIGURE 4.43, VISUALIZATION OF BICYCLE ACCIDENTS (LEFT) AND TYPE OF COLLISIONS INVOLVED BICYCLE (RIGHT) IN PTV 54 FIGURE 4.44, VISUALIZATION OF HEATMAP FOR CYCLE ACCIDENTS (LEFT) AND PEDESTRIAN ACCIDENTS (RIGHT) IN PTV 55 FIGURE 4.45, VISUALIZATION OF CAR ACCIDENTS (LEFT) AND TYPE OF COLLISIONS INVOLVED CARS (RIGHT) IN PTV 55 FIGURE 4.46, VISUALIZATION OF OTHER VEHICLE ACCIDENTS (LEFT) AND TYPE OF COLLISIONS INVOLVED OTHER VEHICLES (RIGHT) IN PTV 56 FIGURE 5.1, RISK OF TRAVELLING ON VASAGATAN 2018-2020 AND PREDICTION 2021 62 FIGURE 5.2, RISK OF TRAVELLING ON SVEAVÄGEN 2018-2020 AND PREDICTION 2021 62 FIGURE 5.3, RISK OF TRAVELLING ON JARLSGATAN & STUREGATAN 2018-2020 AND PREDICTION 2021 63 FIGURE 5.4, RISK OF TRAVELLING ON VALHALLAVÄGEN 2018-2020 AND PREDICTION 2021 63 FIGURE 5.5, RISK OF TRAVELLING ON TORSGATAN 2018-2020 AND PREDICTION 2021 64 FIGURE 5.6, RISK OF TRAVELLING ON KLARASTRANDSLEDEN 2018-2020 AND PREDICTION 2021 64. vi.

(10) List of Tables TABLE 3.1, SUMMARY OF THE APM’S TABLE 4.1, PERCENTAGE OF OVERLAPPING ROAD SEGMENTS WITH MEASURED SPEEDING TABLE 4.2, TRAFFIC ACCIDENTS IN STOCKHOLM (JAN 2018 - FEB 2021) TABLE 4.3, YEARS OF LIFE LOST AND YEARS OF LIFE WITH DISABILITY TABLE 5.1, DISTANCE & AADT 2018-2020 AND PREDICTION FOR 2021 TABLE 5.2, ACCIDENT RATES 2018-2020 AND PREDICTION FOR 2021 TABLE 5.3, ACCURACY RATES FOR PREDICTION MODEL BASED ON MODE OF TRANSPORT. vii. 14 24 26 29 59 61 65.

(11) 1. Introduction The population and motorisation around the world grow fast, especially in the cities. According to Falcocchio and Levinson (2015), the increase of population and motorisation are much greater than the development of capacity in the transportation network, which causes congestion, especially during peak hours. Congestion affects mobility and accessibility since people make decisions, both short- and long-term, with regards to the level of congestion in the traffic system. Long term decisions could be where to live and work, and short-term decisions could be which transportation mode and route to choose if the trip should even be made. According to Ding et al. (2020), in recent years, many people have chosen to cycle in order to avoid congestion, save time and/or money, get exercise, and be more environmentally friendly. However, cyclists are more exposed to traffic and the risk of being involved in a traffic accident, being severely or fatally injured increases. According to WHO (2020), most traffic worldwide fatalities are cyclists, motorcyclists, and pedestrians, which are vulnerable and unprotected road users. Injuries from road traffic accidents are the leading cause of death for people between the age of 5-29 years worldwide. Many countries are working progressively to improve traffic safety by implementing laws and regulations regarding speeding and drunk driving, constructing safer vehicles, and evaluating and improving road infrastructure. According to Lord and Washington (2018), it is an enormous challenge to have high mobility and improve traffic safety since many accidents are connected to users. When designing a transportation network, more factors than safety have to be considered, for example, efficient throughput, congestion, heterogeneous users, and to link various transportation modes. In December 2019, traces of the SARS-CoV-2 virus were discovered in Wuhan, China, named Coronavirus and the disease covid-19. Later this virus started to spread worldwide, and finally, the World Health Organization (WHO) announced it as a world pandemic. Since then, numerous governments have imposed several rules and regulations to restrict the virus’s spread (Evelyn et al., 2020). Since the restrictions were imposed, people have started to stay at home in the beginning of the pandemic, and traffic has decreased significantly (Transport Analysis, 2021). Due to the decrease in traffic, people have the opportunity to drive at high speeds, which directly had a severe impact on traffic safety (Evelyn et al., 2020). Besides increased speed, people also began to shift from public transport to car, bicycle and walking to fulfil their needs (Jenelius & Cebecauer, 2020). The project’s final aim is to analyse the travelling fluctuation during the covid-19 period and the impact of the fluctuation on mobility and traffic safety in Stockholm.. 1.1 Background Traffic safety is an essential issue regarding the number of deaths globally, and Sweden is considered one of the countries with the highest traffic safety. According to the STRADA statistics, approximately 250-300 deaths occur every year in traffic in Sweden. The Swedish government has designed Vision Zero, the National Road traffic safety program (NRSP), to 1.

(12) reduce traffic accidents’ fatalities to zero. In 2007, the Swedish transport agency decided to set the number of road fatalities to work accordingly and take precautionary actions to regulate road accidents. The initial target was framed as 440 fatalities in 2007 to reduce the fatalities to 220 (50 % reduction) by 2020. Figure 1.1 clearly describes the trend between the target and actual fatalities. It can also be observed that in 2020, the number of fatalities is 190, which is lower than the target. 190 is the lowest value of fatalities in traffic in one year since World War 2 (European Union, 2018; Transportstyrelsen, 2020).. Figure 1.1, Comparison of road fatalities in Sweden, target vs actual (Transportstyrelsen, 2021). Figure 1.2 describes the trend of the traffic volume fluctuations in Sweden from 2007 to 2020. The figure represents the annual traffic volumes in million passenger-kilometres of people based on all road transportation modes. In 2020 the volume recorded is approximately 122.3 billion passenger kilometres which is lower than 2019 (Statista, 2021).. Figure 1.2, Comparison of road traffic volumes in million passenger-kilometres in Sweden (Statista, 2021). In Sweden, cars are the most common transportation mode, followed by public transport and walk/cycle. However, there are significant variations in the share of transportation mode depending on the region. In the capital city, Stockholm, 57 % of the kilometres travelled are by public transport, and the corresponding value for the northern part of Sweden is 6 %. (Brundell-Freij et al. 2016) Therefore, it has been assumed in this study to be misleading to 2.

(13) investigate Sweden as a country and instead focus on Stockholm. For that region, it is also possible to access further mobility data than for other cities.. 1.2 Aim & Purpose The study aims to present differences in mobility and traffic safety by comparing 2020 to previous years, as there is a significant shift in traffic demand during the year 2020, based on transport modes, age, gender, and several behavioural factors. The purpose of the study is to learn how traffic accidents are affected by significant shifts in travel flow. The purpose of the study is also to be able to predict the number of traffic accidents based on traffic volume and to evaluate the risk of travelling on a particular road segment by evaluating the accident rate.. 1.3 Research Questions In order to reach the aim, four research questions have been formulated. The first question has been developed to get knowledge of how the mobility and number of traffic accidents have been affected by the pandemic. The second question has been formulated to investigate the accidents of different age groups and if more younger persons have been involved in traffic accidents. The third question has been formulated in order to get knowledge of if the mode of transportation has shifted during the pandemic. The fourth question has been formulated to acquire knowledge of if the risk of travelling at a particular road segment has increased or decreased during the pandemic compared to previous years. 1. Do mobility and traffic accidents follow the same trend during the pandemic? 2. Is the risk increased/decreased in the younger age group regarding traffic safety during the pandemic and is there a difference between the genders? 3. What are the percentages of the shifts between transportation modes, and what might be the reasons? 4. Is the risk of travelling on a particular road stretch (For example, on the city centre’s streets) increased or decreased during the pandemic?. 1.4 Limitations In this study, the pandemic is not the focus, but rather the effects of large shifts in traffic demand, no matter the cause. The area of interest in this study has been limited to Stockholm, Sweden, due to the lack of mobility data for other regions. The share of transportation modes used differs greatly depending on the region. The main transportation mode is cars, due to traffic volume data, but other transportation modes (public transport, cycling, and walking) have been considered and analysed. In the traffic accident data, other transportation modes, such as truck, moped, and motorcycling, are included. However, it has not been investigated the types of the different vehicles, for example, the car brands, the share of SUVs on the roads might have increased, or intelligent systems in newer cars. For years of life lost (YLL), the 3.

(14) total life expectancy has been considered, not different values for males/females. The life expectancy increases by 0,18% yearly, but the value for 2018 has been used respectively. No weight factor depending on the age of the fatalities has been used. The data comparison is limited to 2018, 2019, 2020, and January to February 2021. The reason why only January and February are included for 2021 is since the data for this study was collected in March 2021, and no more recent data were available. The reason why data for 2021 is considered is mainly because of the mobility data and that no earlier data than before the beginning of 2020 is available. In order to be able to compare the mobility before and during the pandemic, January and February 2021 can be compared to January and February 2020. Every year, Trafikanalys (Traffic analysis) (TRAFA) publishes reports regarding mobility. However, the information for 2020 is published May 28, 2021, and is, therefore, not possible to use in this study. Mobility data collected from Apple and Google are not trips actually performed, but the requested data from users, for the respective maps, based on the mode and purpose of the travel.. 1.5 Outline In Chapter 2, the methods followed across the project are presented. Chapter 3 presents the theory regarding mobility and traffic safety. Chapter 4 presents the statistics of mobility data collected from Apple, Google, and the Environmental Barometer, and statistics and discussion regarding the accident data collected from STRADA; the chapter also presents a visualisation of traffic volume data and speeding collected from Trafikkontoret. In Chapter 5, the method, result, and a discussion of the risk analysis are presented, along with predictions of the accident rates in 2021 and accuracy of predictions. Chapter 6 presents the conclusion for the study, and the research questions have been answered. The last chapter is References, where the references used in this paper are listed.. 4.

(15) 2. Method In this chapter, the method used for the project is presented. The method has also been visualised; see Figure 2.1 for an overview. A literature study has been carried out to get a deeper knowledge regarding mobility, traffic safety, and model constructions for accident prediction models. Data regarding mobility and accidents have been collected and has been sorted, filtered, and analysed in MS Excel and QGIS. Based on data, the locations of the traffic accidents have been identified using PTV Visum Safety and the road segments with high speeding’s have been identified using QGIS. The collected data and an accident prediction model have been used to evaluate the risk of travelling on particular roads in Stockholm and to predict the number of accidents at the roads in 2021.. Figure 2.1, Blueprint of the project. 2.1 Collecting Data According to Eiras (2011), every modern country must have a database with reliable health information. Reliable data regarding health in a country is crucial for scientific research and making improvements. Therefore, many countries have databases regarding public health, and the most commonly used indicator for public health is the mortality rate. According to European Union (2018) traffic safety is an essential issue regarding the number of deaths globally and it is of interest to have a database of road accidents and injuries in order to work proactive. According to the Swedish Transport Agency (Transportstyrelsen) (2021), data regarding injuries or fatalities in road accidents in Sweden is registered in the Swedish Traffic Accident Data Acquisition (STRADA). Both the police and the hospitals report injuries or fatalities to STRADA, and the statistics can be found on the Swedish Transport Agency’s web page and is open for anyone to use. 5.

(16) Data analysis has been conducted to investigate how the pandemic has affected mobility and traffic safety in Stockholm. Therefore, data regarding traffic accidents and mobility trends have been collected.. 2.1.1 Traffic Accident Data The Swedish Traffic Accident Data Acquisition (STRADA) has been contacted to access data regarding traffic accidents in Sweden. To access relevant data, it is essential to specify what is needed before making a request. What has to be specified is: ● ● ● ● ● ● ●. Time periods. Geographical delimitation. Number of accidents or number of persons. Transport modes. Degree of accidents. Age of the persons involved in accidents. Gender of the persons involved in accidents.. Other factors might affect traffic safety, for example, speeding and driving under the influence of alcohol. Data regarding the number of cases where the driver got the driver licence revoked due to speeding and drunk driving has been retrieved from the Swedish Transport Agency.. 2.1.2 Mobility Trends Data regarding mobility has been collected from several sources: Apple Mobility Data Mobility trends in various cities can be collected from Apple Maps (Apple Maps, 2021), which can then be categorised based on the transportation mode, i.e., percentage usage of cars, public transport, and walking analysed. However, the data is not actual trips performed but based on the user’s requests, which might affect the values since some trips might not have been performed and some trips might have been performed without the user requesting it. The baseline, set by Apple, is 2020-01-01, might also be misleading since it is a holiday. Mobility data from Apple was available from 2020-01-13 and the end date used in this study is 202102-26. Google Mobility Data Mobility data collected from Google (Google, 2021) can be used to analyse the mobility trends based on the purpose of the visit, i.e., the percentage of the population visiting different areas like grocery stores, pharmacies, parks, etc., in the city. However, the data is not based on actual trips performed but based on the user’s requests. Data was available from 2020-02-15 to 202112-31 and the baseline was set by Google to be January 18, 2020, which might make the values misleading since the data is not compared to previous years when the conditions are more similar. 6.

(17) Environmental Barometer Bicycle mobility trends in Stockholm collected from the Environmental Barometer (Miljobarometern, 2021), an environmental company under the Swedish government’s regular surveillance. The data was collected from sensors placed across various places in the city. Data was available from 2020-01-01 to 2021-02-26 and later this data was adjusted to the proportion of Apple mobility trends. Stockholms stad - Trafikkontoret Data regarding traffic volumes and measured speed in Stockholm municipality have been collected from Stockholm stads - Trafikkontoret (Stockholms stad, 2020a). In the data set, measurements for specific road segments in Stockholm are included. For each road segment, several measurements have been taken and the values represent the yearly average. Since the yearly average is considered, data is valid to 2020-12-31 in this study. In order to get knowledge of if the measured speed is higher than the speed limit, data regarding speed limit of the road segments has also been collected. Data regarding speed limits in Stockholm can also be found at Stockholms stad - Trafikkontoret (Stockholms stad, 2021). TomTom Traffic volumes of cycling, walking, and cars in 2020 for the streets which are used to calculate the risk analysis are retrieved from TomTom (TomTom, 2021).. 2.2 Analyzing Data According to Huber (2011), data has to be analysed in order to be helpful. One way of analysing data is to visualise the data set. By visualising, it might be easier for a person to understand the data set. Although, some information might get lost since it might be difficult to visualise all of the information in the data set without the result getting too blurry for the human eye. A data set can be very small or very large, and the appropriate size of the data set depends on the content and purpose. If the data set is too small, it might not contain enough data to draw any conclusions, and if the data set is too large, it might be challenging to handle, and it might be unnecessarily large, taking too much space and time. Even if a data set is relatively small, it might be too difficult to analyse without suitable software. Further in this chapter there is a short description of the softwares used in the study, along with the description of purpose for using the software in this study.. 2.2.1 Excel Microsoft Excel is a pre-programmed spreadsheet application that can be used as a visualisation tool, organising & manipulating the data. Excel sheets enable us to perform a wide variety of operations, like calculating and manipulating data. Charts function in Excel helps us to transform the raw data into 2-D & 3-D visualisation figures. Another primary function of Excel is to act as a database tool, enabling us to use Excel as backend support for various applications (Jeleel Adekunle, 2010). 7.

(18) Mobility data collected from Apple is visualised based on the mode of transport (car, public transportation, walking), data from Google is visualised based on destinations of visit (workplaces, transit, residential, parks, retail and recreation, grocery and pharmacy) and data collected from Environmental Barometer is visualised based on bicycle movement. Mobility data is also pictured in Excel by considering the baseline as January 7, 2020. Mobility data is visualised using Excel. Traffic accident data collected from STRADA is analysed and visualised using Excel based on different factors like: a) Severity: Minor, Moderate, Severe, Fatal b) Mode of Transport: Bus, Cycle, Walking, Truck, Tram, Motorcycle, Moped, Car, Other c) Type of accidents: Single accidents, Intersections, Passing, Overtaking, Other. d) Age intervals: 0-6, 7-14, 15-17, 18-19, 20-24, 25-34, 35-44, 45-54, 55-64, 65-74, 75+ e) Gender. 2.2.2 QGIS Quantum Geographical Information System (QGIS) is an open-source desktop application used to analyse geospatial data. In QGIS, it is possible to analyse vector data (point, line) and raster data (Matrix of square/Pixel), which also can be imported from external sources. Besides this, QGIS also consists of several inbuilt packages, enabling users to write the code and store the SQL database data (Saul, 2017). Yearly data regarding traffic volume, speeding, and speed limits in Stockholm have been implemented in QGIS to create heatmaps. In order to visualise the data, a base map of Stockholm is used. Data valid for January 2018 to December 2020 have been used, and in order to get information regarding changes, only the overlapping road segments have been used.. 2.2.3 PTV Visum Safety PTV Group was founded more than 40 years ago and is a software for transportation planning. In PTV Visum, it is possible to create macroscopic and mesoscopic traffic models, which can be used to analyse traffic flows. For PTV Visum, a PTV Visum Safety tool can help determine where and why traffic accidents occur and by using PTV Visum Safety, it is possible to detect the black spots in the transport network. Accident data (based on police reports) can be implemented in PTV Visum Safety, and it is possible to create heat maps. In PTV Visum Safety, it is also possible to filter data with regards to chosen attributes (PTV Group, n.d.). Data regarding mobility and traffic accidents in Stockholm has been implemented in PTV Visum Safety to visualise the changes in mobility and traffic accidents during the last years. Heatmaps for mobility and traffic safety are used to see if there are any clear correlations between mobility and accidents. Regarding mobility, traffic flow changes and the shift in transportation mode during the pandemic have been analysed. Traffic accidents have been analysed based on the transportation mode involved in the accidents and the age and gender of 8.

(19) the persons involved in the accidents. It has also been investigated if there are any significant differences in the type of accidents, such as accidents involving only one vehicle or head-tohead collisions, during the pandemic and the previous years. To use accident data from STRADA in PTV Visum Safety, the data has been converted. Converting the accident data can be done by contacting PTV Group in Germany, which a contact form can be done on their webpage.. 9.

(20) 3. Current and Previous Studies on Mobility and Traffic Safety In this chapter, theory regarding mobility and traffic safety is presented. The literature study has been carried out to gain more profound knowledge regarding mobility and traffic safety. Search words used are mobility, mobility covid-19, mobility pandemic, traffic safety, traffic safety covid-19, traffic safety lockdown, and traffic safety pandemic.. 3.1 Mobility Mobility is defined as the movement from X to Y by people or goods. Mobility based measurements can help calculate transportation system performance like planning decisions. Mobility can be differentiated based upon users, modes, and demographics. From a user’s perspective, mobility can be referred to as a person travelling a number of kilometers by cycle, motorbike, foot, and any other mode of transport. From a transport mode perspective, mobility refers to automobile vehicles’ usage, for example, cars and public transit. Besides automobile vehicles, walking and cycling can also be considered. Mobility in terms of land use refers to access to public facilities, for example, parking lots and motorway entries. By knowing the mobility rate, it can be estimated whether the existing infrastructure can accommodate the public needs or not. Then by the conclusions drawn from mobility, authorities can take decisions like increasing the number of buses or trips performed by buses from X to Y. Mobility can be measured by conducting travel surveys to figure out the person-kilometers and tonkilometers relationship. The travel surveys can further analyse Level of Service ratings and Multi-modal system performance (Todd, 2003). According to Fiona (2019), in the Urban Mobility Index, Stockholm has stood second with a 57,1% index score in the survey conducted in 895 major cities globally to reduce congestion, enhance air quality, and reduce pollution at city centre areas by encouraging other modes of transport. Besides this, the Urban Vehicle Access Regulations (UVAR) in Stockholm brought up many recommendations to adapt to the increasing mobility rate (Fiona, 2019): -. -. Any low emission zones inside the city are quoted as environmental zones, and in these areas, heavy vehicles emitting harmful gases were banned from entering, encouraging the public to use other modes of transport within the city. They are imposing congestion charges on vehicles entering the city, shallow emission zones. Restrictions imposed on vehicles’ weight and length entering the city. By having the restrictions on vehicles’ weight and length when entering the road at specific periods. By improving the existing infrastructure to help in accommodating the increase in the share of cycling and walking lanes. Adding new metro stations and expanding the existing metro network in the city.. 10.

(21) By imposing the restrictions mentioned by Fiona (2019) in traffic in Stockholm in 2019, the shares of transportation modes in Stockholm in 2020 is cycling (45%), Car (30%), Public Transport (15%), Walking (10%), i.e. 45% of the trips generated are performed by bicycle, 30% of the trips generated are performed by car, etc., as shown in Figure 3.1 (Stockholms stad, 2020b) .. Figure 3.1, Mobility triangle in 2020 (Stockholms stad, 2020b). 3.2 Traffic Safety According to Tiwari and Mohan (2016), traffic is heterogeneous and not only are there many types of road users, but the road users are also moving at different speeds. Amundsen and Hydén (1977) defined traffic accidents as “a situation where two road users approach each other in time and space to such an extent that a collision is imminent if their movements remain unchanged”. However, according to Ryo (2018), traffic accidents are somewhat vague and can be interpreted differently depending on the person and mainly accidents involving vehicles are considered as traffic accidents. Ryo (2018) believes that the concept of traffic accidents also should include other accidents than accidents involving vehicles. All accidents occurring in the transport system are relevant, even spraining the ankle at a curb since it regards the inhabitants’ safety during transport. Ryo (2018) also mentions that it is not applicable only to consider the number of fatalities since everyone will die eventually and that it is more accurate also to calculate and consider years of life lost (YLL).. 3.2.1 Improving Traffic Safety According to Tiwari and Mohan (2018), it is important not to blame an individual for an accident if the interest is to avoid similar accidents but instead investigate the system’s flaws. There will always be some tired and unfocussed drivers while driving, and it is of interest to minimise the degree of the damages if an accident occurs. According to Vadeby and Forsman (2018), speed is of great importance in traffic since there is a close correlation between the risk of crashing, the speed travelled when crashing, and the degree of the injuries. It has been estimated that the number of fatal accidents can be reduced by 21 % by lowering the average speed by 5 %. The corresponding value for severe accidents is 16 %. According to Tiwari and Mohan (2016), driving on congested roads is safer since the speed is lower and the risk of an accident being severe decreases compared to driving at free-flow speed. Therefore, municipalities strive to have some slightly congested roads during some periods in the day when planning traffic. However, when the density is higher, the risk of an accident occurring increases. According to Tiwari and Mohan (2016), it is also possible to improve safety by 11.

(22) separating the different transport modes, and in an ideal network, each transport mode would have a separate lane.. 3.2.2 Traffic Safety in Stockholm, Sweden Every year, hundreds of people died in traffic accidents in Sweden (Transportstyrelsen, 2021). Nevertheless, according to Kristiansen et al. (2018), Sweden is considered to have one of the safest transport systems globally; only 3 out of 100 000 persons die in traffic accidents annually, which is considered a low value compared to other countries. Since 1997, Sweden has been working proactively regarding traffic accidents by introducing the Vision Zero policy program, where the vision is that no fatal nor severe accidents should occur in the transport system. Since Vision Zero was implemented, the number of fatalities and severely injured in traffic has decreased. Between 2000-2010, the number of car users who died in traffic decreased by 60 %. However, just a few studies show a direct cause-and-effect between the policy and the number of decreased traffic fatalities. Although, the positive development seems to be highly connected to the Vision Zero policy program. A measure of the program is to have a no-blame culture, meaning the individuals involved in the accidents should not be blamed, and instead, the roads and vehicles should be evaluated. Some measures taken because of Vision Zero are to lower the speed limit on the roads and implement more roundabouts at intersections. Using roundabouts instead, the risk of an accident to happen increases compared to using traffic lights, but the accident’s risk to be severe or fatal decreases significantly due to reduced speed. Hence, the minor and moderate accidents might increase when implementing roundabouts, but there is a tradeoff: fatal or severe accidents might decrease, which is the purpose of Vision Zero. According to the report “Safer City Streets: Global Benchmarking for Urban Road Safety” released by the International Transport Forum in Paris, among the major European cities, Stockholm is having the safest roads (Tanya, 2018). Besides this, a report published in 2011 presents Stockholm road fatalities are 50% less than the fatalities in San Francisco and based on the population of Stockholm and rate ratio analysis conducted, it clearly states that walking and cycling as the mode of transport is the safest mode in the city (McAndrews, 2011).. 3.2.3 Risk Compensation According to Norén (2019), Sweden has worked progressively to increase traffic safety for many decades. September 3 1967, is a historical date in Sweden, it was the day the traffic shifted from left-hand traffic to right-hand traffic, and the day is also called the “H-Day”. The main reason for the shift was to increase traffic safety. Sweden had to keep up with international standards. The majority of the other countries in Europe had right-hand traffic, which made it difficult and a higher risk for accidents when driving in other countries and when foreigners drove in Sweden. Another reason, according to Stockholmskällan (n.d.), the steering wheel was placed to the left in most of the cars, and by implementing right-hand traffic, the traffic safety was increased due to better vision for the driver. Before implementing the right-hand traffic, catastrophic consequences in traffic due to the right-hand traffic were imagined by the inhabitants. However, according to Norén (2019), the inhabitants were well prepared and 12.

(23) informed of the change, and at H-Day, more than 100000 citizens voluntarily guarded approximately 19000 pedestrian crossings all over the country in order to avoid accidents. According to Brüde (2013), additional measures to increase traffic safety were implemented in 1967, for example, speed limits on rural roads and improving the conditions of the roads. According to Perakslis (2016), the accident rate dropped by 40 % after the implementation of right-hand traffic in Sweden and a reason could be that the drivers were more vigilant in traffic in the beginning. However, when people do not experience the negative effects of the risks, people tend to underestimate the risks and be overconfident, and the traffic accident rates in Sweden later increased to previous levels. According to Houston and Richardson (2007), when the drivers feel safer, they tend to take higher risks and drive more recklessly, which is a phenomenon called risk compensation. Due to risk compensation, when traffic safety measures are implemented, the decrease of accidents and fatalities might be lower than predicted.. 3.2.4 Accident Prediction Models Accident prediction models are just to predict accidents in traffic, and according to Hauer (2010), the two main purposes are: 1. In order to set policy or program targets, it is of importance to make a prediction regarding the accident statistics and what would be in the future if no measures are taken. The safety targets are set with regards to the prediction of no implementation of policy or target. 2. In order to estimate the effect of the safety measures, a prediction of what would have been if the measure was not implemented is needed. The difference between the purposes is the timing. The first purpose is before a measure is implemented and in order to set realistic targets, and the second purpose is after the implementation of a measure in order to evaluate the effect (Hauer 2010). In order to predict traffic accidents, accident prediction models (APM) can be used. According to Eenink, et al. (2005), there are many different APMs, and the formulas and values of the parameters might differ for different regions and countries. However, according to Eenink, et al. (2005), the most important is to include traffic volume and length of the road segment. Further in this chapter, four different APMs are presented. The findings have been summarized in Table 3.1.. 13.

(24) Table 3.1, Summary of the APM’s. Accident Prediction Models. Literature. Short description. Generalized Linear Models (GLIM). Archer, J., (2005). Indicators for traffic safety assessment and prediction and their application in micro-simulation modelling: A study of urban and suburban intersections. Royal Institute of Technology, Sweden. Estimation of the number of accidents at a specific intersection between primary and secondary roads. In the basic model, flows, type of intersection, speed, and location are considered.. Hierarchical Bayes-Spatial Mode. Vandenbulcke, G., Thomas, I., & Panis, L.I., (2014). Predicting cycling accident risk in Brussels: A spatial case-control approach, Accident Analysis & Prevention, 62, Pp 341357. A Bayesian modelling approach. Bernoulli model with a logistic link is appropriate for predicting the probability of having a bicycle accident at location i.. Power Model. Nilsson, G., (2004). Traffic Safety Dimensions and the Power Model to Describe the Effect of Speed on Safety, Bulletin 221, University of Lund, Lund Sweden. Prediction of fatal and severe accidents based on speed changes. Not suitable for predictions of the number of fatalities.. Umbrella Crash Adedokun, A., (2016). Application of A PTV visum safety accident Prediction Road Infrastructure Safety prediction model, which is in XML Algorithm Assessment Methods at Intersections, format. Norrköping: Linköping University.. Generalized Linear Modelling Approach According to Archer (2005), for linear models, the dependent variable (i.e. the number of traffic accidents in this study) often needs to be normally distributed. Since traffic accidents are not usually normally distributed but rather Poisson or negative binomial distributed, Generalized Linear Modelling (GLIM) approaches have been found to be more useful when predicting traffic accidents. To describe the relationship between the accident frequency in an intersection and the traffic flow from the primary and secondary road, Hauer and Lovell (1998) used the basic GLIM, see Equation 3.1, where it’s used to find the frequency of predicted number of accident occurrences. In the model, the average annual daily traffic (AADT) for the primary road segment (Qp) and the secondary road segment (Qs) is considered, and a, b, and c are constant values depending on the type of intersection, speed, and location, 𝛌 is represented as mean of actual accident occurrences. Equation 3.1. 14.

(25) Kulmala (1995) and Maher and Summersgill (1996) later proposed an extended version of the basic GLIM where other influence factors are included, see Equation 3.2. In the extended version, x represents any additional variable related to d and d is a model parameter. Equation 3.2 Hierarchical Bayes-Spatial Model According to Vandenbulcke, Thomas, and Panis (2014), due to the increasing trend of using bicycles as a transportation mode, an APM for cyclists is of great importance in order to make the transport system safer for cyclists. Cycling is better for people’s health with the aspect of getting exercise. However, the risks of getting into an accident, and being severely injured, must not be forgotten. In order to predict the risks of accidents for cyclists at a location, i in the network, Equation 3.3-4 is used. Yi is binary and is the dependent variable, i.e. if an accident occurs at a location i or not, and pi is the probability for an accident involving cyclists at location i occurring. a represents the intercept from the model, b is a vector of parameters, and the risk factors are basically represented in vector xi which consist of Explanatory variables related to an increased probability of bicycle accident. The risk factors if intervened, then can be changed in order to reduce the accident probability. When there is a lack of information for the values of the parameters a and b, the values are in general specified as a=0 and b=1,1-6. However, according to Vandenbulcke, Thomas, and Panis (2014), the hidden statistics for cyclists in accidents are relatively high since many people do not bother to report the accidents to the police. So as the dependent variable is binary, a conditional Bernoulli with a logistical link (2 staged formula as equation 3.3, 3.4) is used inorder to find the probability of bicycle accident at a particular location. Where the equation 3.3 is an accident risk model followed by logit distribution (Vandenbulcke; Thomas; and Panis , 2014). Equation 3.3 Equation 3.4 Power model According to Nilsson (2004), the differences in speed are considered in the power model, and the number of fatal and/or severe accidents is predicted based on historical accident data. The power model is presented in Equation 3.5-6, where y1 is the prediction, y0 is the number of fatal/severe accidents before the change, v1 is the speed before, and v0 is the speed after. The power model is a commonly used model due to its simplicity and is often used for calming traffic measures. The power model has been found useful to predict the number of fatal and severe accidents. However, the power mode is not suitable for predicting the number of fatalities since the flow is not known and how many people are involved in the accidents.. 15.

(26) Equation 3.5 Equation 3.6 Umbrella Crash Prediction model According to Adedokun (2016), the umbrella crash prediction model is provided with the PTV Visum Safety plugin and it is possible to predict the number of crashes for a given period of time and length of road using Equation 3.7. Where AP is the predicted number of accidents for the given length of road at a given time period; R is the accident rate per million vehicle kilometer; Cr is the ratio of crashes of a given crash type, i.e the proportion of fatal crashes or severe crashes or minor crashes to total crashes occurred on a given length of road; AADT is the Average Annual Daily Traffic of the road section; L is the length of the road in km. This model is used to predict the number of accidents across the given length of road irrespective of the transportation mode used.. Equation 3.7 Model Validation According to Ramzai (2020), for any study conducted regarding the prediction of an event that is going to happen in the future, model validation plays a crucial role. Where this step in the study decides whether the model used in the study is accurate to the study or has a lot of error gaps between the actual and predicted numbers. If this is not conducted there are chances of a model with poor predictions. According to (Zach, 2021) and (Jomnonkwao, et al., 2020), Mean Absolute Percentage Error (MAPE) can be used to weigh the prediction accuracy of the model. MAPE is further presented in chapter 5.3. 3.2.5 Loss of Statistics What is important to remember when analysing crash data is the loss of statistics. Not all accidents are reported. According to Tiwari and Mohan (2016), there is a significant loss of crash statistics and incredibly minor or severe accidents. In India, the unreported fatal accidents are estimated to be 5-10 %. According to Ryo (2018), the risk of unreported fatal traffic accidents in Sweden is almost non-existence. However, there is a risk of unreported severe traffic accidents, and Ryo (2018) found that the risk of the accident not being reported increases with a lower degree of damage. According to Ryo (2018), one reason statistics is lost is because the police and health care have different methods for reporting accidents. The police reports mainly more severely traffic accidents involving motor vehicles, and the chance of the accidents being reported increases for severe accidents and if a crime is suspected. The accidents reported by the police are reported relatively uniformly across Sweden, which is not the case for the traffic accidents reported by the health care since each of the hospitals follows their standards. The hospitals only report injuries of people who visit the hospital and report a higher percentage of emergency cases and a lower percentage of a minor, not emergent, 16.

(27) injuries, although the accident occurred in traffic. According to Ryo (2018), there is also a loss of statistics due to system changes and routines. In Sweden, in the fall of 2013, a system change was carried out in the system in which the police report accidents, which resulted in a loss of statistics and in 2018, it was only partly recovered. In 2015, routine changes were made in the hospitals’ system report accidents, which led to the loss of statistics.. 3.3 Traffic Accidents During the Pandemic Many people have stayed at home during the pandemic in many countries, trying to slow down the spread of the virus (Lu, Zhue, & Wang, 2020). According to Qureshi et al. (2020), the Missouri traffic volume decreased during the pandemic since more people stayed at home. However, the number of trucks travelling on the roads was unchanged, and sometimes even slightly higher than before. The number of trucks has not been reduced because more people did their shopping online instead of visiting the stores themselves. Qureshi et al. (2020) also found that the total number of traffic accidents decreased at the beginning of the pandemic. However, it was found that only minor and moderate accidents were decreased. The number of fatal or severe accidents were unchanged. Qureshi et al. (2020) speculate why the number of fatalities is unchanged during the pandemic, and one reason is since the traffic volume has decreased and there has been less congestion, it is possible to drive at higher speeds, which increases the risk of crashing and making the crash more severe. Qureshi et al. (2020) also speculate that another possible reason could be that the number of persons driving influenced by alcohol/drugs might have increased during the pandemic. Vingilis et al. (2020) mention several reasons why the number of fatalities and severely injured in traffic has not decreased during the pandemic. Increased speed is an apparent factor, and in Toronto, speeding has increased by 35 %, and stunt driving has increased by 200 %, according to the Toronto police. In Toronto, it has also been discovered that when the traffic volume at the roads decreases, people using other transport modes than cars are using the roads. Vingilis et al. (2020) also mention that young drivers take more risks, and since the mortality of the virus is considerably higher for older people, more older adults might avoid traffic; therefore, the percentage of young people in traffic has increased. According to Vingilis et al. (2020), young people might also use their vehicles to interact with each other since bars, restaurants, and events have been closed and younger people are therefore even more exposed to traffic. Vingilis et al. (2020) also suggest closed gyms as a potential reason for traffic accidents since it is not allowed to work out at gyms, people might choose alternative workout methods instead, e.g. walking, running, or cycling, which leads to them being exposed to traffic as unprotected road users.. 17.

(28) 4. Statistics regarding mobility and traffic accidents This chapter presents the result and analysis regarding mobility and traffic accidents data. Based on collected data, it can be seen that the mobility has changed during the pandemic and when traffic volumes have decreased, there are possibilities to drive at higher speed, which have also affected the number of accidents. The total number of accidents has decreased but the number of severe accidents is almost as high as previous years and younger people are affected, which is further presented in this chapter.. 4.1 Mobility Changes During 2020 It is presented how mobility has changed during 2020 due to the pandemic in Stockholm municipality. Factors considered are: purpose of trips, transportation mode used, traffic volumes, and violations in traffic.. 4.1.1 Purpose of Trips During 2020, the purpose of trips has changed; see Figure 4.1, according to mobility data from Google (Google, 2020). The purposes included are workplace, transit, residential, parks, retails and recreation, and grocery and pharmacy. The baseline is January 18, 2020. As the data for the previous years is not available, a comparison of purpose with the before years is not possible. Therefore, the rate of change from March 2020 to February 2021 with the baseline is compared. Due to lack of data, it is assumed that the distance of each trip performed to the purpose is the same, even if that is most likely not the case in reality, for example, the length of each trip to visit parks is the same, and the distance of each trip travelled to visit the groceries shop is the same. In the figure “transit” meant mobility to places like public transport hubs like the subway, train stations, etc. As can be seen in the figure, the trips with the purpose of getting to workplaces, retails and recreation, and transit decreased in March 2020 and have not exceeded the baseline ever since. The trips to grocery stores and pharmacies increased at the beginning of the pandemic. In the data set, it is possible to see some peaks during the pandemic for visiting grocery stores and pharmacies, and the dates for the peaks correspond to holidays, for example, easter, midsummer, Christmas, and new years, where people usually gather and eat a lot of food. However, the peaks are not above the baseline, meaning that even in the peaks, the number of trips to the grocery store is lower compared to a normal day in January. It can be seen in the figure that the number of trips to parks increased during the pandemic from April to October, compared to the baseline. However, the baseline is January 18, 2020, which might be misleading. Since there are four seasons in Sweden, it is reasonable to believe that, even before the pandemic, more people are visiting parks during the summer months. In order to know if more people visited parks during the pandemic than before, data for previous summers should be collected and analysed. The figure also shows that in November 2020 to February 2021, the number of trips to parks is lower compared to the baseline. It would also be of interest, especially for the statistics regarding visiting parks, the weather and weekday of January 18, 2020 (the baseline) since that might have an effect on the number of trips that day.. 18.

(29) Figure 4.1, The purpose of trips in Stockholm Municipality. 4.1.2 Transportation Mode The transportation mode used when making a trip in Stockholm municipality has changed during 2020, see Figure 4.2, according to mobility data from Apple, where the transportation modes car, public transport, walking, and cycling are included. The baseline, which is set by Apply, represented the situation on January 01, 2020. As the data for the previous years are not available, comparison of transportation mode used with the before years is not possible. Therefore, the rate of change from March 2020 to December 2020 will be compared with the baseline, and the rate of January to February 2021 compared with the situation in January to February 2020. Due to lack of data, it is assumed that the distance of each trip performed by the mode of transportation is the same, for example: the distance travelled by every car is the same, and the distance travelled by every cyclist is the same, even if that is not very likely in reality. Mobility for all other modes, except cycling, is collected from Apple Mobility Trends and cycling data is collected from the Environmental barometer and has been adjusted to the proportion of the Apple Mobility data. As can be seen, there is a large decline in using cars, public transport, and walking as transportation mode in March 2020, which is when the pandemic was declared. The reasons for the decline might be that less people made trips, which was presented in Chapter 4.1.1, and also that more people started to use cycling as transportation mode, which is possible to see in the figure. However, cycling increased during the summer months. For all of the transportation modes, there is a slight spike of mobility in April 2020. However, the increase is greater for cars, and lower for public transport. For cars, the values are above the baseline from May 2020 to November 2020, meaning more persons used. In the figure, it is possible to see a decrease, for all of the transportation modes, from the end of June to the beginning of August, which might be because of vacations and summer break. At the end of August, the values for public transport are above the baseline. There are two obvious drops for walking, where the first, in May 2020, depends on missing data, and the second, at the end of December 2020, might be since it is winter and also holidays. For all of the transportation modes, there is a drop in the values in December 2020, which later increases 19.

(30) again, probably since people have to go back to work and schools. The values for all transportation modes exceed the baseline in the months before the pandemic was declared, which might be because of the baseline used. The values during the year give an explanation of the movements but the baseline might be misleading. However, when comparing the time before the pandemic (January and February 2020) to the corresponding values in 2021, it is possible to see a decrease for all transportation modes, which is most likely because of the pandemic and the encouragement to avoid unnecessary trips. The values are also lower than the baseline (January 1, 2020), which indicates that less trips are made even on weekdays during the pandemic than on a holiday before the pandemic.. Figure 4.2, Transportation mode used in Stockholm Municipality. 4.1.3 Traffic Volume Data regarding traffic volume in Stockholm Municipality has been collected from Trafikkontoret Stockholm. The data has been imported to QGIS in order to visualise the traffic volume and make comparisons. Figure 4.3 is visualised for which roads there are measurements of the traffic volume from 2018 to 2020. The lines’ thickness depends on the yearly average traffic volume 24 hours on a weekday; the thicker the line, the higher volume. The left map represents traffic volumes on road segments in Stockholm during 2018 and 2019 (purple lines), and in the right map, measurements valid for 2020 are also included (cyan lines). As can be seen in the left map, many of the road segments overlap, meaning for those road segments, there are measurements for several years, and it is possible to analyse the change of volume.. 20.

(31) Figure 4.3, Road segments (purple and blue lines) with data regarding traffic volume. The left map presents data for 2018 and 2019 (purple lines), and in the right map, data for 2020 is included (cyan lines).. Based on the traffic volume data provided by Trafikkontoret, the average number of vehicles at the road segments have increased at some road segments in 2020 compared to previous years and decreased at others. The average value is negative (-238) and also the total (-403129), meaning, overall, the traffic volume for the considered road segments has decreased. However, it must be taken into consideration that the values represent the yearly average and that the pandemic started in March and that some of the measurement might be from before the pandemic. The average number of measurements for each road segment in 2020 is 6.22 and most likely not all of the measurements are made before the pandemic. Many of the measurements might be from during the pandemic since the data is collected by Trafikkontoret, for which it might be of interest to know what has happened regarding traffic during the pandemic. However, the average number of measurements for each road segment in 2018 is 6.93 and in 2019 it is 6.92, meaning, in 2020 less measurements have been performed. The changes of traffic volume for some streets (Vasagatan, Sveavägen, Jarlsgatan & Sturegatan, Valhallavägen, Torsgatan, and Klarastrandsleden) in Stockholm based on mode of transport: cyclists, pedestrians and cars, are collected from the Environmental Barometer ((Miljobarometern, 2021) and TomTom (TomTom, 2021) for 2018 to 2020. The traffic volumes collected from the two sources are in terms of Annual Average Daily Traffic (AADT), which is represented in figure 4.4.. 21.

(32) Figure 4.4, Traffic volume on particular streets in Stockholm. 4.1.4 Violations in Traffic It has been investigated if there has been a difference in the violations in traffic during the pandemic compared to previous years. The violations considered in this study are speeding and drunk driving. Speeding The main reason for a revoked driver license in Sweden (>60 %) is substantial violations in traffic, for example speeding (most common), driving against red, and not stopping at intersections with a stop requirement. In 2020, the number of persons in Sweden that got the driver license revoked is a record high (Trafikverket, 2021). However, before drawing the conclusion that the overall speed has increased, based on revoked driver licenses, additional factors have to be considered. The number of revoked driver licenses are record high, what about speeding tickets? If it is possible to drive at higher speeds, people speeding might drive even faster, resulting in a revoked driver license instead of a ticket. Therefore, the number of tickets might have decreased. In the data sets regarding traffic measurements in Stockholm, provided by Trafikkontoret, average, median, and 85-percentile speed are included for the road segments, respectively. When comparing the measured speed on the road segments for 2019 and 2020, it shows that the average speed has decreased, increased, or has not changed between the years, see Figure 4.5. The average value for the change of average speed is positive (0.85), which indicates that the overall average speed at the considered road segments has increased. However, the data is yearly average, which means that year 2020 also includes two months before the pandemic. The speed data is also independent of the time in the day and when the traffic flow decreases, there might only be a significant difference in speed during peak hour.. 22.

(33) Figure 4.5, change of average speed 2019-2020. The measured average speed has also been compared to the speed limit for the road segments, respectively, in order to get knowledge if there is a difference between the type of road. Figure 4.6 presents the percentage of road segments where the average measured speed is higher than the speed limit of the road. As can be seen, in total, the speed has increased in 2020 compared to previous years but the percentage of road segments, for all speed limits, where the average measured speed is higher than the speed limit, have increased during 2020 compared to 2019, except for the road segments with the speed limit of 80 km/h. In 2020, at none of the road segments in the data set with the speed limit of 80 km/h, the average speed was higher than 80 km/h. However, in the data set, there are only 22 road segments with the speed limit of 80 km/h included, which might not be representable for the complete road network. When comparing the values for 2020 with the measurements for 2018, the trend is not as clear since the percentage of speeding is higher in 2018 for the road segments with the speed limits of 40, 70, and 80 km/h. For the road segments with the speed limit of 60 km/h, the values are the same. For none of the road segments, traffic calming measures have been considered and it is not known for which of the road segments measures might have been taken and installed in the previous years. It is also not known for which of the road segments the speed limit might have been reduced temporarily due to construction work, etc.. Figure 4.6, Percentages of the road segments where the average measured speed is higher than the speed limit. As mentioned in Chapter 3.2.1, according to Vadeby and Forsman (2018), the speed is of great importance since there is a close correlation between the risk of crashing, the speed travelled 23.

(34) when crashing, and the severity of the injuries. Vadeby and Forsman (2018) mentioned that it has been estimated that the number of fatalities can be reduced by 21 % if the average speed is lowered by 5 %. Due to the measured speeding, the percentage of road segments the average speed could be decreased by 5 % and the average speed is still higher than the speed limit, has been investigated, see Table 4. As can be seen, the percentage of road segments where the average speed could be decreased by 5 % and still be higher than the speed limit decreased in 2019 compared to 2018. The corresponding value for 2020 is higher than for 2019 but not higher than for 2018. However, the percentage of the number of road segments where 15 % of the vehicles drive at least 20 km/h faster than the speed limit has increased in 2020 compared to previous years. One reason for the increase of speed could be that the traffic volume has decreased, and it is possible to drive faster. The table presents values for the overlapping measured road segments in 2018 to 2020. Table 4.1, Percentage of overlapping road segments with measured speeding. Percentage of overlapping road segments where the speed limit is exceeded even if the average speed decreases by 5 %. Percentage of overlapping road segments where 15 % of the vehicles drive at least 20 km/h faster than the speed limit. 2018. 30 %. 3%. 2019. 17 %. 4%. 2020. 18 %. 10 %. In order to get knowledge of where the overlapping road segments with the highest speed violations are located and if there are any differences of the locations compared to before the pandemic, QGIS has been used, see Figure 4.7. In the figure, maps of Stockholm municipality are presented where the road segments where 15 % of the vehicles drive faster than the speed limit in 2018 and/or 2019 (blue) and in 2020 (red) are highlighted. As can be seen, some of the locations overlap, meaning the locations of the speeding’s are the same during the pandemic as before. However, it is possible to see some changes of locations as there are some road segments highlighted by red (2020) and not blue. It is also possible to see that for some of the road segments highlighted in blue (2018/2019), there are no red highlighting, meaning the exceeding of the speed was not as severe in 2020. Reasons for the change of locations of severe speeding’s could be the change of mobility during the pandemic.. 24.

(35) Figure 4.7, Road segments where 15 % of the vehicles drive at least 20 km/h faster than the speed limit (blue: 2018/2019, red: 2020). Traffic Accidents Involving Alcohol Another risk factor in traffic is driving or being exposed to traffic under the influence of alcohol and/or drugs. Based on accident data provided by STRADA, the percentage of traffic accidents where the injured person was confirmed to be influenced by alcohol at the time of the reported traffic accident has increased during the pandemic, see Figure 4.8. The figure presents the percentages of persons influenced by alcohol involved in traffic accidents in Stockholm and the transport mode used, before (January 2018 to February 2020) and during the pandemic (March 2020- December 2020). The transportation modes presented are cycling, walking, moped, car, and others. Based on the accident data, there have been no reported traffic accidents involving alcohol for other transportation modes. As can be seen, the total percentage has increased during the pandemic. The increase for cars is four times higher than before the pandemic. A reason for the increase could be that the alcohol consumption might have increased during the pandemic, especially among adults, due to lost jobs and possible depression, which could be the reason for the increase amongst cars. However, the share for walking and travelling by moped have decreased. For the category “other”, which includes electric scooters, the share has more than tripled during the pandemic. One reason why the percentages for walking has decreased and the percentage for the category “other” has increased could be that instead of walking, an electric scooter is rented. The statistics presented include all degrees of damage (minor, moderate, severe, and fatal) and the hidden statistics for accidents, especially minor and moderate.. 25.

(36) Figure 4.8, The share of persons involved in traffic accidents that were under the influence of alcohol at the time of the accident. 4.2 Traffic Accidents in Stockholm 2018-2021 Based on data provided by STRADA, the number of traffic accidents in Stockholm has decreased during the pandemic, see Table 4.2. The accidents have been categorised as fatal, severe, moderate, and minor in the table. The categories “no injuries” and “degree of injury is uncertain/unknown” are included in the table but have not been considered in the rest of the analysis due to lack of data. The number of accidents in 2021 might be misleading since only January and February are included. However, it is still of interest to keep the values for 2021 since the mobility data for 2021 will be analysed due to lack of earlier data. The number of traffic accidents might also increase over time due to delays in the registration of accidents. Although, in this study, the presented values will be used. Further in this chapter, degrees of traffic accidents, years of life lost and years of life with disability, type of traffic accident, transportation modes involved in accidents, and age and gender of the persons involved in the accidents are presented. Table 4.2, Traffic Accidents in Stockholm (Jan 2018 - Feb 2021). Year. Fatal. Severe. Moderate. Minor. 2018. 8. 75. 1018. 2147. 401. 57. 3706. 2019. 11. 67. 961. 2294. 599. 55. 3987. 2020. 4. 57. 558. 1744. 214. 26. 2603. 2021. 0. 7. 124. 233. 53. 1. 418. (Jan-Feb). 26. No injuries. Degree of injury is uncertain / unknown. Total.

No results found