Methodology for assessing the road traffic risk in urban areas

In the assembly of concerns for reducing the negative effects of urban road traffic, special consideration is paid - both at the local and global level - to issues related to road traffic safety. Spotting urban zones vulnerable to traffic crashes, understanding the causes of crashes, identifying some directions of action to reduce the traffic risk and their application from the urban planning stage can improve road safety. In this framework, the paper presents a methodology for estimating the intrinsic risk linked to road traffic. The main objective is identifying the relationships between the crash frequency and the intrinsic factors contributing to their occurrence. The proposed methodology is particularized for Bucharest based on the disaggregated statistical data analysis on severe accidents recorded in 2017-2022. The resulting model allows the classification of risk levels and, further, identification, planning and scheduling measures to improve urban road safety.


Introduction
In contemporary society, the risks caused by anthropic activities have been aggravated [1].The concepts of sustainable development tend to globalize concerns for diminishing multiple risks of ecological and sociological nature (greenhouse effect, pandemic, fossil fuels resources, social inequalities, etc.).Road traffic risk is part of the risks produced by human society.The programs for road safety are undoubtedly on the political agenda of the authorities at different levels [2,3].
Traffic rules and control mechanisms have always been established in road systems, regardless of their type of road infrastructure.Parallel, the technical conception of road vehicles has considerably evolved to reduce collision severity and determine behavioral changes.Complementary, starting from the "vision of zero victims (dead and seriously) in road traffic", adopted by the Swedish Parliament in October 1997 [4], new concepts regarding road safety responsibilities have been added.The main responsibility for low road safety performance is not assigned to drivers but to the system planners.The main goal of the new vision is to eliminate the likelihood of events causing victims in road crashes.Key actions based on scientific and tested experience must be adopted at the beginning of the decision-making process [5].
Consequently, the one who designs and implements the system is the first responsible for road safety.From this perspective, at the urban level, an objective of the Sustainable Urban Mobility Plan (SUMP) consists in enhancing road safety from the planning phase [6].The solutions to ensure mobility demand, dependent on the urban area's size, shape and pattern, are potentially responsible for the volume and structure of the related traffic.The mobility solutions (correlated to the goals of sustainable development) must harmonize the accessibility needs (for people and freight) with the life quality (including the imperious necessity for road safety) and with the requirements of the protection of the natural and anthropic environment [7].
From a phenomenological point of view, the urban area where the road traffic risk occurs is a complex spatial system.The defining input classes for associated road traffic risk can be grouped into four sets: features (mobile entities -vehicles, pedestrians, etc. and functional entities of the traffic infrastructure), stakeholders (local authorities, managers of traffic infrastructures, etc.), spatial pattern and temporal pattern [5,7].In Bucharest -the study area of this research, all these classes have faced significant changes in the last two decades.Radical transformations in socio-economic life, major changes in the pattern of the commercial sector (caused by the development of large shopping centers), new dense residential and offices zones, structural and spatial changes of the interest points (for work, education, leisure time, etc.), as well as changes in the working practices and travel behaviors determined by Covid-19 lockdown have led to traffic heterogeneity increase, i.e., road traffic risk exposure.Identifying vulnerable urban areas to road crashes, understanding the causes, identifying action directions to reduce the traffic risk, and implementing them from the urban planning stage can improve road safety performance [8][9][10].
In this framework, this research develops a methodology for a detailed analysis of the recorded crash and spotting urban areas with intrinsic traffic risk.The purpose is to rank the zones with high traffic risk and to ensure the necessary information support for assessments in cost-effectiveness analysis (CEA) [5,11] and, further, for planning the proposed actions to improve road safety performance.

Traffic risk analyses
Usually, at the macroeconomic level, four intercorrelated influencing factors of traffic risk are emphasized [5,10]:  Drivers of road vehicles (behavior with perceived traffic risk)  Vehicles (technical characteristics, maintenance, total load, destination)  Infrastructure (location, road geometry, type and structure of the road system, infrastructure management, traffic management)  Socio-economic activities (dimension, pattern, temporal and spatial distribution).The characteristic variables of the socio-economic activities determine the interactions between the first three classes of factors.The traffic, as well as the travel/transport demand (being a derived demand) depend on the trade-off between the mobility offer and demand [5].The temporal dynamics of these interactions and the particularities of the four classes of factors generate spatial and temporal specificity for the size and nature of the traffic risk [8,9].In other terms, the spatial mobility supply and demand, the traffic risk exposure, and the frequency, location and severity of road crashes are influenced by the socio-economic space and, undeniably, by the implemented road safety measures [11].
Concerns for increasing road safety belong mainly to the state.Local authorities are insufficiently involved.Urban development plans usually do not consider road safety at meso-and macroscopic levels.The lack of a "culture of road safety" is reflected in the deficiency of traffic risk assessment in the large projects financed for urban areas [6].The most common methodologies for characterizing road safety are based on crash statistics.They intend to identify deficiencies and dysfunctions of the road infrastructure that cause driver errors.The data available in Romania compile the reports made by the Road Traffic Police following serious crashes.The collected data refers to the location and time of the crash occurrence, the road category, type of the road infrastructure feature (road section or junction, their type), local characteristics for traffic management, type of road surface, environmental settings (lighting, weather conditions), type and severity of the crash (number of injured and dead, number of involved vehicles), primary causes of the crash.
The traffic risk evaluations considering all the factors involved in crash occurrence are not frequent.This state is caused mainly by the sectoral character of the data provided by different sources and the difficulty of correlating the information (such as time series datasets on traffic -volume, debit and structure of the road traffic, urban space settings -exposure, transformations of the urban zones and functions).For instance, based on data available for Bucharest, Figure 1 shows a decrease in crashes during and after the COVID-19 lockdown (2020, 2021).Obviously, the reduction is associated with diminished road and pedestrian traffic flows.However, due to the lack of traffic time series data (in terms of traffic volume and structure), it is difficult to perform complex analysis to determine the correct dependencies between the size of the road and pedestrian traffic and the crash frequency.Therefore, a comprehensive interdisciplinary approach to the key concepts of the binomial" road safety -traffic risk" is necessary to lead to systematic modeling of phenomena.By defining and identifying the concept of "risk situation" in a spatial model appropriate for functional analysis, criteria can be established for representing the different particular aspects of road traffic in connection with the urban environment [9].Only in this manner can the road traffic risk assessment be included in an "ex-ante" hierarchy of project financing in urban mobility system planning.

Assessment of the traffic risk
Different types of models for estimating the number of crashes are presented in the literature [12][13][14].Most models are proposed to estimate the crash frequency for different types of intersections [14,15] and homogeneous segments of road arteries [16].There are also models developed to identify the relationships between the crash occurrence and road network pattern [17], respectively, the urban functions [8,9].
In this research, the basic equation used to estimate the number of crash victims, k, is [18]: where E represents the risk to exposure (vehicles-km),  the risk related to crash frequency (crashes/vehicles-km), and  the risk related to crash severity (victims/crash).
A study of the dependences of exogenous and endogenous variables (speed, driver behavioral variables, technical-functional characteristics of vehicles, attributes of the road, lighting and weather circumstances, etc.) is required for each of the three factors (E,  and ), The methodology developed and synthetically presented in the next section aims to create data sets that allow calibrating the dependencies between the number of crash victims and the composite risk exposure.

The spatial model of the study area
This research aims to develop a model that allows ranking the urban road infrastructure features with an increased risk of crashes and estimating the number of victims.The model results represent inputs in the cost-efficacy analyses necessary to prioritize financing measures to improve road traffic performance [5].The model is developed for the particularities of the urban space and the road infrastructure network from Bucharest.
The topological, geometric and structural design of urban road networks determine the traffic safety performance [7,9].In order to define and calibrate the traffic risk estimation functions, the development of the macroscopic digital model of the urban network in Bucharest is necessary.
The road network models are different in terms of applied disaggregation level, depending on the specific analysis objectives.In this study, the model of the road network includes (Figure 2  Several studies [8,9,19] indicate that different socio-demographic variables, such as the population, the active population, the degree of poverty and the economic activity, influence urban road safety.In order to be able to consider these features in traffic risk estimation, the digital model includes (Figure 2):  The geographical datasets of urban zones defined by their functions (the available data for Bucharest include 9675 zones delimitated for 12 urban functions).
 The geographical datasets of traffic analysis zones (TAZ) defined in Bucharest -Ilfov SUMP [20].The urban area is divided into 366 zones.Based on these data, for each TAZ, the densities of population, employees, school population, and road network are computed (as attributes that could influence the crash frequency) [7,9].The data available in this study for traffic risk estimation contain the records related to severe crashes (involving injuries, dead and/or significant financial damages) in Bucharest from 2017 -2022 (summarized in Figure 1).In order to identify the properties of the most vulnerable network features, the crash data is processed in Geographic Information System (GIS) environment based on procedures presented in [9].The data sets are used to disaggregate crashes in homogeneous classes.The crashes, grouped into three categories (those involving only motor vehicles, bicycles and motor vehicles, and pedestrians and motor vehicles), are assigned to feature classes of the urban road network (Figure 2).However, if the resulting size of certain samples for some crash causes is too small, then the level of data disaggregation will be reviewed [9,16].

Data analysis
The data sets presented above are necessary for estimating the risk associated with road traffic and then applying CEA for ranking the measures to improve road safety performance (Figure 3).performance.The first phase involves data spatial processing in order to select potentially dependent variables for crash occurrence.It is assumed that a type of crash that repeats in the same location represents a predictable event [2,10,12].Crash frequency can indicate the existence of an influence of the road system in the cause of that type of accident.If crashes are systematic in certain types of locations, it is unlikely to be caused only by driver error or vehicle dysfunctionality.They depend on technical and urban environmental identifiable factors which can be analyzed and a priori reduced.Therefore, datasets of yearly disaggregate homogeneous crash classes are compared and intersected (Figure 4).
In this research stage, the set of road features with repeated crash frequency is obtained.Further research is needed to define and calibrate the traffic risk estimation functions associated.In order to complete this step, besides selecting features characteristic of urban road network and urban zones (presented above), supplementary processing of traffic flow data is necessary.Traffic characteristics are considered important factors in crash occurrence [2,21].Generally, average annual daily traffic values are used to characterize the flow in this type of estimation, even if, thus, significant effects of traffic non-uniformity are excluded [14,21].

Conclusion
The risk associated with urban road traffic detaches from other socio-technical risks through the diversity of the influencing factors and the consequence extent.Therefore, it requires particular examinations.The dispersed responsibilities of various authorities and institutions impose a systematic approach to political actions to increase urban road traffic safety performance.
Under the new technical and ethical visions, the binomial "road traffic risk -road traffic safety" represents a topic of interdisciplinary research oriented towards the responsibility of those who design and operate the system.The transfer of the crash responsibility from the user to the one who designed and implemented the system is the fundamental characteristic of the new paradigms that will gradually restructure the current situation in road traffic.
In this line, the research aims to provide a model for assessing the risk associated with urban road traffic and for identifying measures to increase urban road traffic safety performance from the planning stage (before crash occurrence and blackspot registration drawing attention to the need for safety actions).The paper presents the first stage of the study.The developed model of the urban space allows processing data regarding the urban road network, traffic patterns, urban patterns, and crash data necessary to analyze the factors influencing traffic risk at the disaggregated level.Further research is needed to calibrate the traffic risk estimation functions and obtain appropriate results to carry out the necessary CEA in prioritizing the proposed measures to improve road safety.
):  Urban Road Intersection Dataset with attributes related to section length, number of lanes, tram lines existence (simple, double), bus public transport lines existence, street side parking existence. Urban Road Intersection Dataset with attributes related to traffic lights existence, configuration (3-way intersections -T, Y or skewed Y intersections, 4-way intersectionsregular cross, skewed 4-legged intersections, more than 4-way intersections), type of infrastructure: only road traffic intersections and intersections including tram infrastructure, public transport stations existence, pedestrian crossing existence/type.

Figure 2 .
Figure 2. Selection criteria for disaggregate crash classes.Several studies[8,9,19] indicate that different socio-demographic variables, such as the population, the active population, the degree of poverty and the economic activity, influence urban road safety.In order to be able to consider these features in traffic risk estimation, the digital model includes (Figure2): The geographical datasets of urban zones defined by their functions (the available data for Bucharest include 9675 zones delimitated for 12 urban functions).

Figure 3 .
Figure 3. Flowchart for ranking the measures for enhancement of urban traffic road safety.performance.The first phase involves data spatial processing in order to select potentially dependent variables for crash occurrence.It is assumed that a type of crash that repeats in the same location represents a predictable event[2,10,12].Crash frequency can indicate the existence of an influence of the road system in the cause of that type of accident.If crashes are systematic in certain types of locations, it is unlikely to be caused only by driver error or vehicle dysfunctionality.They depend on technical and

Figure 4 .
Figure 4. Example of comparison of traffic crashes spatial distribution in Bucharest -2021 (a, b) and 2022 (c, d) dataset selections of cumulative frequency of vehicles involved in different crash categories.