Understanding Spatial–Temporal Patterns in Trespassing on Railway Property

Abstract

Trespassing on railway tracks is a growing problem in rail transport, with multiple causal factors, including increasing urbanisation, high-frequency rail traffic, higher volumes of traffic, etc. The predominant factor is human behaviour (lack of knowledge about trespassing, poor decision-making by road users and others). This research aims to analyse the available data to determine the frequency, patterns, and factors contributing to trespassing on railway tracks and to identify potential locations with the highest recorded trespassing. This is achieved by conducting a case study using data from various sources on trespassing from 2001 to 2023 on the Italian railway network. The methodology of this study consists of data collection on trespassing, data cleaning, and three-step analysis (description of variables used, and application of R programming language for descriptive statistics, correlation, and association analysis). The outcome of this study is the description of the data collecting process of trespassing on the Italian railway network, the identification of temporal factors, e.g., month, day, and hour of trespassing, and spatial factors, e.g., location and railway line where trespassing occurs most frequently, and a list of current and planned prevention measures on the Italian railway network. In the future, trespassing locations can be analysed according to the topology of risk.

1. Introduction

A trespasser is any person who is on railway property where this is prohibited, with the exception of users of a level crossing, according to the definition of the European Union Agency for Railways (ERA) [1]. This definition varies slightly from country to country, but generally, trespassing refers to any place on a track where it is not permitted to cross (e.g., stations, open tracks, bridges, etc.) [2,3,4,5]. Crossing a level crossing when barriers are lowered can also be defined as trespassing, but this again depends on the specific legislation of the country in question.

Trespassing on railway tracks is a significant problem with increasing urbanisation. Road users are often unaware of the risk of trespassing and crossing the tracks for various reasons, such as taking a shortcut to work or school, the distance to the legal crossing being too great, or because they believe that nothing bad will happen if they are careful.

The particular dangers and problems associated with trespassing are evident from the accident reports. Eurostat data from 2022 shows that 64.1% of fatalities are due to trespassing, and 28.6% occurred at level crossings [6]. Trespassing is not only a dangerous act in the UK; for example, trespassing caused delays of up to 75 h in the 2023/2024 period, representing a 15% increase compared to previous years [7]. In Belgium, Infrabel reported 512 intrusions on railway tracks in 2024, with an FWI (Fatality and Weighted Injuries) of 3.2 [5]. The trespassers are not deterred even by strict laws, as in France, where the consequences of trespassing include fines of EUR 3750.00 and a six-month prison sentence. Nevertheless, 34 deaths were recorded in 2023 [8]. In Croatia, the consequences of trespassing are evident in eight train collisions with trespassers in 2023, resulting in the death or serious injury of trespassers [9]. A high number of casualties caused by trespassing on tracks, at 1348 (51.3% of fatalities), were also recorded by the FRA (Federal Railroad Administration, Washington, DC, USA) in 2023 [10]. In Australia, 5747 near-miss incidents occurred between 2016 and 2022. These incidents had the potential to cause injury, death, or disruption to rail services [11]. In Italy, the subject of this study encompasses different types of trespassing (inconveniences, incidents, anomalous situations, and other BDP events) [12].

This study aims to describe the process of detecting, collecting, and analysing trespassing data on the Italian railway network, identify temporal patterns such as the year, month, and/or hour, and determine the locations with the highest frequency of recorded trespassing. It also reflects on whether the trespassing follows a trend or occurs randomly. To further complement the analysis, prevention measures that are being implemented or planned to be implemented are added. The former, however, was not included in various data sources of recorded trespassing from 2001 to 2023. The combined dataset does not allow for more in-depth analysis. Therefore, this study does not seek to explain why trespassing occurs but rather provides a statistical analysis to identify recurring patterns and possible critical locations.

The structure of this study consists of six sections. Section 2, the Background, presents a literature review of similar research. Following the literature review, Section 3 explains the methodological approach to analysing trespassing. The methodology is broken down into more detailed steps, allowing for a more precise analysis. Section 4, the Results, presents the main findings of the analysis. The results are divided into several subsections for ease of interpretation. Section 5, the Discussion, and Section 6, the Conclusion, summarise the findings with a critical analysis, highlighting the strengths and limitations of the research, as well as opportunities for future work.

2. Background

Each act of trespassing on railway tracks could potentially result in serious consequences of trespassers being seriously injured or fatally wounded. Many researchers have analysed the severity of accidents caused by trespassing. In South Africa, the problem is that out of 379 trespassing accidents, up to 20% of accidents were not reported [13], while in New Zealand, over a period of 10 years (1994 to 2003), the number of trespassing accidents has remained constant (10–20 incidents) [14]. A Turkish study revealed that even if trespassers survive the initial impact with the train, in 11.4% of cases, they succumb to injuries [15], while in India, 96.6% of trespassers have succumbed to injuries at the location of the accident [16]. In Finland, a yearly decrease of 4.4% in trespassing has been observed [17], and 32% of collisions in the city of Chicago (32%) are due to trespassing [18]. Trespassers are often left with lasting injuries, like brain injuries (Queensland, Australia, (14.2%) [19] or in India (69.3%) [20]), amputation of legs (66.7%) [21], or damage to lungs, livers, or kidneys [16]. In the USA, it is shown that children often need amputation operations and require an average of 5.7 operations after an accident [22]. Collisions potentially cause stress disorders in train drivers, which could last for up to a year after an accident, and often do not reduce over time [23].

The severe consequences of trespassing may be caused by several factors, such as the specific location of trespassing, temporal factors, human behaviour and profile, motivation for trespassing, and others, which are explored in the following sections.

Locations of frequent trespassing are concentrated near large cities where there is a high traffic density [24]. The common locations of railway trespassing include stations or stops, shortcuts for daily use, tourist routes, level crossings, and meeting accommodation places [25]. Other reasons for trespassing include a lack of legal crossings (such as level crossings or pedestrian crossings) and poor urban planning [25].

The time frequency of trespassing is more challenging to determine. For example, in a survey conducted in Finland, trespassers indicated that most trespassing occurred during the day (17.9% in the afternoon hours) [26], and 67% of respondents trespassed at least once a week and were aware of the existence of legal crossings [24]. In the USA, the highest number of fatal accidents occurred in March and August. Notably, 57% of accidents happened on Fridays, Saturdays, and Sundays, 60% of accidents occurred at night, and 75% of accidents took place in city areas [27].

Trespassers are predominately males [26], and include the local population, dog owners, students, and pupils [25]. It is not uncommon that the trespasser is under the influence of alcohol [14,27,28]. The trespassers were found in both upright and lying positions [29], were using a shortcut [30], were often unaware of the dangers (18.2%), and stated a lot of misconceptions about trespassing [26]. Poor infrastructure at railway stations, long distances to the legal crossing, and fear of underpasses can also contribute to increased trespassing [31].

Some respondents trespass to work (64.3%) or school (42%), while others who do not trespass see no reason to do so (41%) [25]. Young people often have a limited ability to assess risks of trespassing [32].

Other factors that could contribute to trespassing include the increasing length of railway lines [33], higher frequency of trains, and speed of trains [34]. It is apparent that numerous factors impact railway trespassing. Factors may vary between each country or trespassing location, which is why it is necessary to analyse each location separately. This research background forms the basis for the methodology employed in this study.

3. Methodology

The methodology approach consists of four subsections: (1) location, (2) data collection, (3) analysis and method, and (4) tools (software). This structure helps to clarify the approach taken in this study.

3.1. Location

This case study was conducted using an example of trespassing on railway tracks on the Italian railway network. The current railway network in Italy consists of 24,636 kilometres of railway, out of which 16,879 kilometres are in operation, with 2000 railway stations [35]. This study did not involve primary data collection on-site. Instead, it relied on secondary sources.

3.2. Data Collection

Various data sources of recorded trespassing from 2001 to 2023 are used [12,36,37,38,39,40,41]. The data were chosen due to the availability of data regarding trespassing on railway tracks, which were systematically collected on the Italian railway network, and due to the multidisciplinary nature of the research in question. Because data were obtained from multiple sources, it had to be organised and merged into a single database to enable efficient analysis.

3.3. Analysis and Method

The method for analysing trespassing on the Italian railway network involves several distinct steps, as illustrated in Figure 1. This process includes data filtering, cleansing, categorisation, data analysis, and the exploration of prevention measures that are currently implemented or planned to be implemented on the Italian railway network.

Cleaning the data is a necessary process to translate the data from Italian to the English language and create a categorisation for more detailed analysis. The parameters considered for the analysis include spatial details (location and section of railway lines), temporal aspects, and descriptions of trespassing (type of trespassing, accident types such as fatalities or injuries).

The parameters of month and day were extracted from the temporal records of the trespassing (date). The time of trespassing was recorded in the format hh/mm. In this form, the peak or rush hours could be easily identified. Additional analysis included correlation and association analysis. The final step of the method involves recording both the current and planned measures for trespassing prevention on the Italian railway network.

3.4. Tools (Software)

The analysis was conducted using R programming (version 2024.04.2) to better understand trespassing and its characteristics in Italy. This tool supported the identification of temporal patterns, spatial relationships, and potential correlations and associations between variables in this study.

4. Results

Trespassing on railway tracks was analysed from multiple perspectives. First, the process of data collection for trespassing on the Italian railway network was described. This was followed by an analysis of the descriptive statistics and correlation analysis of the trespassing database, as well as a list of prevention measures currently implemented or planned for implementation on the Italian railway network. The data for analysis were collected from multiple sources [12,36,37,38,39,40,41].

4.1. Data Collection on Trespassing in Italy

The data collection on trespassing on the Italian railway network can be divided into six steps (Figure 2). There is no detection system in place to detect trespassing on the Italian railway network. Therefore, data about trespassing locations and trespassers were collected from rail staff and other authorities or can be collected from any member of the public present when the trespassing occurred. Trespassing can occur on a section of a railway track or at a specific location, such as a railway station, railway stop, or milestone. The second step involves a preliminary assessment, during which, if necessary, the train driver is instructed to reduce speed or stop the train, and the staff of the railway infrastructure manager investigates the trespassing incident. The severity of the trespassing act determines the next step in the detection process. If there are no consequences or the consequences are minor, the railway infrastructure manager usually just records and stores the data that trespassing has occurred and sends it to the NIB (National Investigation Bodies). Another important factor is the ANSFISA (National Agency for Railway and Highway Safety), which is responsible for ensuring the safe operation of railway transport. Trespassing, which results in significant consequences, is investigated by the NIB and the police, with external consultants sometimes present at the location, including experts and academic members. Ultimately, all trespassing incidents are stored, processed, and transmitted by the NIB to the ERA [36,37,38,39].

All railway infrastructure managers in the European Union (EU) are required to establish and manage their operations in accordance with the Safety Management System (SMS). The SMS ensures that railway infrastructure managers comply with all safety requirements. The SMS follows the Plan–Do–Check–Act (PDCA) cycle, which identifies rail hazards, manages risk, and ensures collaboration and coordination among different entities within the railway system to manage and mitigate shared risks that arise at points where their operations interact or overlap [1,42].

The ERA required EU Member States to systematically record railway accidents and incidents in the ERAIL database until 2020, as per EU Directive 2016/798. The site was discontinued, and all future data on railway accidents and incidents are now sent by the NIB to the ERA and stored in the combined database. Data such as date, location, and consequences (e.g., injuries or fatalities) are collected for analysis [1,43].

4.2. Descriptive Statistics of Trespassing on Italian Railway Network, Italy

Data on trespassing from 2001 to 2023 on the Italian railway network includes the collection of recorded trespassing and consists of the first three steps of trespassing detection, processing, and storing. Data are provided by the Sapienza University of Rome, Faculty of Civil and Industrial Engineering, and multiple other sources [36,37,38,39]. Variables for analysis include location, railway line, temporal data, description of trespassing, various accident types (e.g., fatalities, serious injuries, minor injuries, etc.), and the nature or type of trespassing. The data on the description (nature) of trespassers’ behaviour did not include the reasons for the trespassing occurrence, and there were also cases with missing data. The most common types of behaviour were walking across or along the railway tracks, walking outside the authorised points, attempting to cross the railway tracks, and trespassing close to or at railway stations. In the first step of the analysis, the trends of trespassing are established. Some percentages of trespassing are rounded to two decimal places to distinguish the differences, given the minimal variations. The highest number of trespassing was recorded in 2023 (11.71%). The patterns show an exponential growth in trespassing from 2001 (0.6%) to 2023 (11.7%). Thereafter, the number of recorded trespasses increased annually.

In Figure 3, trespassing is analysed according to the month and day, from 2001 to 2023. The highest number of trespasses were recorded in May (9.34%). This is closely followed by March, April, June, and July. Further analysis shows that the percentage of trespassing events did not decrease significantly until November, a trend that continued through January. A more noticeable increase can be observed from February, which marks the transition from winter to spring. The day with the highest percentage of trespassing is Saturday (15.17%). Compared to other days of the week, there is no significant difference. For example, compared to other days, the difference in trespassing is up to one percent. The only day that stands out is Sunday, with a noticeable difference in recorded trespassing.

The highest number of trespassing occurred between 16:00 and 16:59 (9.4%). This was closely followed by the period from 17:00 to 17:59. These hours are considered the rush hours of afternoon traffic. The number of trespassers increase as rush hour approaches and decrease after it ends. From 14:00 to 19:00 h, the difference in the number of trespassing is between 0.5% to 2%. Interestingly, the percentage of trespassers do not decrease after the morning rush hour between 7:00 and 9:00, but increase.

To determine whether the temporal trends obtained by summarising trespassing events by month, day, and hour from 2001 to 2023 are consistent across years, an analysis was conducted for each year. Figure 4 presents a summary of the frequency of appearances by month, day, and hour (highest recorded trespassing). The month with the highest frequency of trespassing events is October, followed by June. The frequency varies from zero to five occurrences from 2001 to 2023. The day with the highest frequency of occurrence is Saturday, followed by Thursday (range of zero to eight occurrences). However, the difference through each year is only apparent in the first few years, and the only day with a significant drop in trespassing is Sunday, with no frequency of the highest day of recorded trespassing. When hours are analysed by each year, the time range with the highest frequency of occurrence is 16:00 to 16:59 (up to 10 occurrences). These results appear to confirm that trespassing predominantly occurs during the afternoon peak hour.

The locations of trespassing are examined according to the type of railway line. The highest category of the railway line is the high-speed category, which is designed explicitly for high-speed trains (at least 250 km/h); the main railway lines serve as the backbone for national and international traffic, the secondary railway lines support regional and local traffic, and the railway junction represents a location where multiple different lines intersect [36]. On the main railway line, the highest percentage of trespassing was recorded (49.5%), followed by the secondary railway line. The most significant issue is that in 16.3% of recorded trespassing, there are missing data. This gap in the data collection poses a challenge to fully understanding trespassing trends, if they are analysed by railway line type. A better approach is to analyse the trespassing that occurred on the section of the railway line (between railway stations or stops) and the trespassing that occurred at a specific location (e.g., station, stop, milestone, etc.). The majority of trespassing occurred in a section of the railway line (65.5%).

The DTPs (Direzione Territoriale Produzione, in English Territorial Operational Directorates) are areas of division of the railway network and are under the jurisdiction of the Italian railway infrastructure manager. There are a total of 15 division areas [44]. The DTP Milano has the highest percentage of recorded trespassing (14.7%), followed by the DTP Firenze. The lowest percentage of trespassing was recorded in the DTP Cagliari. Milan, Bologna, Florence, and Rome are all cities in Northern and Central Italy with denser populations and railway networks, which could explain the higher percentage of trespassing.

Trespassing is categorised into several types, including an anomalous situation (another category which does not fall into any other category, with no significant consequences), inconvenience (from 2016, which replaced the anomalous situation), BDP abnormality (abnormality in rail operation), near-miss inconvenience, minor incident (light injuries, delays or similar), significant incident (high material damage, serious injuries, fatalities, and significant delays), and vandal act (acts typically include graffiti, destruction of property, etc.) [38].

Figure 5 only shows trespassing types higher than 10% in the overall share of trespassing types. Notably, some categories were present before 2016 but not after, and vice versa. This can again be explained by new reporting requirements from the ERA, meaning some trespassing types were categorised differently from 2016. There is also an exponential increase in the number of trespassing (anomalous and other inconvenience events), which could be attributed to the former.

The severity of trespassing is categorised into minor injuries, serious injuries, and fatalities (Figure 6). Minor injuries are accidents or incidents that cause disruptions in rail transport and result in light injuries, while serious injuries severely affect a person’s health, often leading to the need for hospitalization or lasting consequences. Fatalities are accidents that result in the death of one or more individuals, as well as significant material damage.

Notably absent is the period from 2001 to 2007, when no minor injuries were reported. Since 2011, there has been a significant spike in the number of minor injuries. The highest number of minor injuries occurred in 2015 (25). From 2015, the number of minor injuries has dropped to 12 in 2023. The year with the highest number of serious injuries was 2005, with thirty-two, followed by 2011 and 2014, both with thirty-one serious injuries, with one less serious injury occurring in each of these years. These are peaks in the number of serious injuries. There is a presence of a constant drop and a rise in serious injuries. If fatalities are analysed, there is an immediately apparent fluctuation in fatalities from 2001 to 2003. The highest number of fatalities occurred in 2023 (65) and 2018 (64). Following 2020, there was a noticeable rise, with 30 additional fatalities. This chart highlights the variability in annual fatality numbers, with both upward and downward trends.

In many cases, trespassing caused high train delays (Figure 7). The more severe consequences often result in higher delays to the normal operation of rail traffic. This is the case for 2018 (a high number of fatalities and delays) but not for 2023 (a high number of fatalities but fewer delays). In 2001, there were the fewest trespassing events, but the highest delays in rail traffic (7740 min, or 129 h). More peaks are noticed in 2002, 2003, 2011, and 2018.

4.3. Relationship Analysis

Based on the previous analysis, correlation and association analyses are used to test possible relationships between different variables. This will include testing the relationship between categorical and numerical variables.

4.3.1. Correlation Analysis

The correlation analysis was conducted using three test examples. The results of the analysis are shown in

Table 1. Results of correlation analysis.

Test	Variables	Pearson’s Product–Moment Correlation	Spearman’s Correlation Test
Test one	(1) Hour of trespassing (2) Minor injuries	✓ r = 0.004114333 ✓ p = 0.4656 > 0.05 (95%) ✓ CI = 0.0069 to 0.0152 ✓ df = 31,451	✓ S = 0.0073, ✓ p = 0.1955 > 0.05 (95%)
Test two	(1) Hour of trespassing (2) Serious injuries	✓ r = 0.0174 ✓ p = 0.001991 < 0.05 (95%) ✓ CI = 0.0064 to 0.0285 ✓ df = 31,451	✓ S = 0.0283 ✓ p = 0.0000004991 <0.05 (95%)
Test three	(1) Hour of trespassing (2) Fatalities	✓ r = −0.0257 ✓ p = 0.000005269 < 0.05 (95%) ✓ CI = −0.0367 to −0.014 ✓ df = 31,451	✓ S = −0.0026 ✓ p = 0.6457 > 0.05 (95%)

. Test one examined the relationship between the time of trespassing and minor injuries, using numerical variables. The minor injuries are summarised by the hour. The morning and the afternoon peak hours and evening hours from 17:00 to 22:00 represent periods with the highest number of trespassing. This could be the cause of the higher number of minor injuries; however, since these records are not recorded in a combined trespassing database, additional research is needed. The correlation between hours of trespassing and minor injuries has resulted in an extremely weak positive correlation, as indicated by Pearson’s product–moment correlation. The p also indicates that the correlation is not statistically significant, suggesting no strong connection with the occurrence of minor injuries and hours of trespassing events. These findings suggest that the specific time of day is not associated with the number of minor injuries. The Spearman correlation test was also conducted, which revealed almost no monotonic relationship, while suggesting the results are not statistically significant.

Test two examined the relationship between the number of hours spent trespassing and the occurrence of serious injuries. The high number of serious injuries occurred during the afternoon rush hours.

The correlation between hours of trespassing and serious injuries showed a weak positive correlation, as indicated by Pearson’s product–moment correlation. However, the p also indicates that the correlation is statistically significant, suggesting a connection between the occurrence of serious injuries and the number of trespassing acts. These findings suggest that the specific time of day is linked to a higher incidence of serious injuries. However, since the correlation is weak, it is reasonable to assume that multiple other factors also influence the occurrence of serious injuries at specific times. A Spearman correlation test was also conducted, which revealed an almost non-monotonic relationship (weak monotonic relationship), suggesting that the results are statistically significant.

Test three examined the relationship between hours of trespassing fatalities. The rise in fatalities is most pronounced during morning and afternoon rush hours, around the time of these periods. This somewhat overlaps with the distribution of trespassing by hours.

The correlation between hours of trespassing and fatalities showed a weak negative correlation using Pearson’s product–moment correlation. The p also shows that the correlation is statistically significant, suggesting a connection with the occurrence of fatalities and hours of trespassing, which suggests that the specific time of the day does affect the number of fatalities however, since the correlation is weak, again, as in previous cases, multiple other factors could affect the occurrence of fatalities at specific times. The Spearman correlation test revealed an extremely weak negative monotonic relationship (weak monotonic relationship), while suggesting the result is not statistically significant.

In all three tests, there was little to no correlation. Given the lack of a significant relationship, a different comparison was made between the frequency of trains per hour and the number of fatalities per hour. Two variables are compared and overlapped to determine whether fatalities follow similar patterns of a rise and fall as the variable of train frequency. Data on the number of trains were extracted from the timetable for Station Roma Ostiense [45]. The reason for this was the extensive size of the Italian railway network. It is important to note that the timetable data exclude passing and freight trains. Between 0:00 and 4:59, there was little to no overlap because no trains were passing during that period. After this period, including morning peak hours, small patterns are noticed. Between 10:00 and 19:59, the rise and fall in both variables—train frequency and fatalities—showed strikingly similar patterns. Small patterns of rise and fall are observed in the evening after 8:00 p.m. This suggests that fatalities may be partially explained by the frequency of trains.

4.3.2. Association Analysis

The relationship between DTP areas and the trespassing type was examined, including incidents, inconveniences, and other BDP events. The data were filtered to include only the classification of trespassing from 2016 to 2023 due to changes in reporting requirements. Incidents include minor injuries, serious injuries, and fatalities. An example of inconvenience is people entering and exiting the railway area without consequences, whereas other BDP events are related to some abnormality in the rail system and operations.

Pearson’s Chi-squared test is used to test the relationship between the variable DTP area and trespassing type (incidents, inconveniences, and other BDP events). With the assumption of independence, the value of X-squared = 441.63 suggests a significant difference between the observed values and the expected values (degrees of freedom, df = 28). The p < 0.00000000000000022 (significance level of 0.05) is very small, meaning the test is highly significant. In other words, the type of trespassing is not independent of the DTP area, demonstrating a statistically significant relationship between the two variables.

Figure 8 visualises the relationship between two variables.

Table 2. Explanation of results of association analysis showed in Figure 8.

Visual Element	Interpretation	Analysis
Dark red	Strong overrepresentation of trespassing type in the DTP area (observed >> expected)	Significantly more types of trespassing than expected.
Light red	Mild overrepresentation of types of trespassing in the DTP area	Mildly more types of trespassing than expected.
Dark blue	Strong underrepresentation of trespassing type in the DTP area (observed << expected)	Significantly less types of trespassing than expected.
Light blue	Mild underrepresentation of trespassing type in the DTP area	Mildly less types of trespassing than expected.
White	No meaningful deviation of trespassing type in the DTP area (observed ≈ expected)	The count of trespassing type is expected for the DTP area and no deviations are detected.
Dash border	The dash border represents results that are statistically significant and examples are given on strong overrepresentation and mild overrepresentation.
Solid Border	The solid border represents results that are not statistically significant and the examples are given on strong underrepresentation or mild underrepresentation.
Dash white Border	Trespassing types are expected in the DTP area but not observed (strong underrepresentation of expected values with statistical significance).
Solid white Border	Observed values align with expectations (no significant deviation).
Values −4 to 4 (Standardised Residuals)	Values −4 (strong overrepresentation) to 4 (strong underrepresentation) of trespassing types in DTP area.

provides a further explanation of Figure 8. The colours in the mosaic plot represent the patterns of trespassing types across DTP areas.

The examples of the strong overrepresentation of other BDP events in a notice in DTP Venezia (engl. Venice) are statistically significant. Furthermore, if DTP Milano is analysed, inconveniences are marked white with dash borders, meaning that one should expect more recorded events (expected events) than are recorded by official statistics (observed) [12,36,37,38,39,40,41].

4.4. Trespassing Prevention Measures on the Italian Railway Network

The currently implemented and planned to-be-implemented measures are listed in

Table 3. Measure for prevention of trespassing (current and planned for implementation).

Measure Type/Status	Measure Name	Objective	Source
Detection	Cameras at railway station	Detection of trespassing at stations	[46]
Planned	AI cameras for detection at the station	AI detection of unusual behaviour and movement of people	[47]
Detection	Employment of railway staff or security guards	Rail staff or security guards responsible for monitoring and conducting sweeps of the railway area and tracks	[48]
Detection	Psychical barriers (e.g., fences, walls, etc.)	Prevention of access to railway areas and tracks	[46]
Infrastructure	Sign that informs users of the proximity of legal crossing	Traffic signs on roads or next to railway tracks that inform users of the distance to legal crossing (level crossing or pedestrian crossing)	[47]
Current	Safety rules	Rules for your safety behaviour in the railway environment	[49]
Signalisation	Additional safety rules	Awareness campaign on railway and road safety	[50]
Education	Project “Train… to be cool”	Aimed at students across Italy	[51]
Education	Initiatives “In piazza con lo sport” (“In the Square with Sport”)	Street events on railway safety organized across Italy	[51]
Education	Prevention Campaigns—BE CAREFUL! Make a difference	Posters and informational brochures	[51]
	Summary	10

. The collection of current and planned prevention measures was included as an additional element to complement the analysis by providing additional context, rather than being directly linked to the combined dataset. There are no available data on the effectiveness of these measures. Measures are divided into four categories: (1) detection, (2) infrastructure, (3) signalisation, and (4) education. Detection systems are stations where there is a high frequency of passenger movement. In 2019, the installation of portals and cameras started with the objective of surveying, passenger counting, and anti-queue systems. The surveillance system is planned to extend to several main terminals in Italy (Rome Termini, Naples Centrale, Firenze SMN, Milan Centrale, Bari Centrale, Palermo Centrale) [46]. Artificial intelligence cameras equipped with motion sensors are planned to be installed to monitor unusual behaviour or movements of people, preventing unauthorised access to railway areas and tracks (trespassing) [47]. In the future, the Italian railway network manager is considering employing additional railway staff or security guards. Their objective would be to monitor unauthorised movement or the crossing of railway tracks while also doing sweeps alongside tracks and railway areas [48].

Physical barriers, such as fences and walls, are placed alongside the railway tracks to prevent unauthorised individuals from accessing the railway area. Additional signalling or traffic signs are placed alongside railway tracks and on the road to inform users of the existence and distance to a legal crossing (level crossing, pedestrian crossing, underpass, overpass, etc.). The purpose is to inform users that the legal crossing exists in proximity to their location, so they might choose not to trespass [47].

Education on trespassing dangers can influence and promote correct behaviour in railway environments. For example, the Italian railway network manager has written safety rules that passengers should follow to ensure safe movement and travel. There are three simple safety rules: “respect” level crossing, stay behind the yellow line at stations, and do not trespass on railway tracks [49]. In addition to safety rules, the Italian railway network manager conducts safety campaigns to promote correct behaviour. In 2023, the campaign focused on promoting correct behaviour at stations, near railway tracks, and level crossings. This included posters, safety videos, and safety messages on social networks for all participants in rail transport. This campaign was joined in collaboration with the Ministry of Infrastructure and Transport [50]. The railway police is a police branch responsible for policing the Italian state railways. The first listed initiative was part of the “Train… to be cool” project, which was directed at all students in Italy. Together with ANSFISA and different sports associations, the railway police has also conducted the initiative “In the square with sport”, where through organising sports events, participants learn about basic safety rules, e.g., yellow lines, safety culture, and dangers in the rail environment. Moreover, railway police also educate passengers with safety posters inspired by traffic signs and leaflets, along with information on correct behaviour, to avoid exposure to the risk of theft and fraud [46,51].

4.5. Summary of Results

The summary of results consolidates all the key findings into a single table,

Table 4. Summary of key findings.

Research Question	Findings	Examination
How are data on trespassing events collected?	○ Collected through 6 steps.	(1) Detection; (2) Preliminary investigation; (3) Consequences; (4) Investigative body; (5) Data analysis and storage; (6) Reporting to ERA.
Frequent location of trespassing?	○ Main railway line (49.5%). ○ Section of the railway line (65.5%)—excludes railway stations. ○ DTP Milano (14.7%).	Hypothesis cause: higher frequency of passenger and freight trains or higher population density Problem: missing data.
What type of trespassing event has the highest percentage of trespassing events?	○ Another inconvenience (in 2023, 3544 cases).	Hypothesis cause: change in reporting requirements and regulations by ERA.
Year, month, day, and hour with the highest percentage of trespassing events?	○ 2023 (11.7%) ○ May (9.34%) ○ Saturday (15.17%) (the highest frequency of occurrence in each year) ○ Afternoon peak hour 16:00–17:00 (9.4%).	Hypothesis cause: new reporting regulations by ERA. Other reasons: fewer trains, leisure time, and human behaviour (work). In some cases, trespassing follows patterns.
The year with the highest number of minor injuries, serious injuries and fatalities?	○ 2015: 25 minor injuries ○ 2005: 32 serious injuries ○ 2003: 65 fatalities.	The current dataset does not allow for more in-depth analysis.
The year with the highest total delays in rail traffic?	○ In 2001 (129 h).	The current dataset does not allow for more in-depth analysis. Hypothesis cause: more serious injuries (cumulative delays).
Correlation analysis: Tests one, two, and three	○ Minor injuries: no statistical significance. ○ Serious injuries: weak correlation, statistically significant. ○ Fatalities: weak negative correlation, statistically not significant.	The current dataset does not allow for more in-depth analysis.
Frequency of trains	○ Test: Station Roma Ostiense ○ Similar trends (rise and fall) from 10:00 to 20:00 h.	Overlap suggests that the frequency of trains follows similar patterns, which warrant further investigation. However, the current dataset does not allow for more in-depth analysis.
Association analysis	○ Example DTP Milano (many more expected inconveniences have been recorded and observed).	The current dataset does not allow for more in-depth analysis.
Prevention measures on the Italian railway network	○ Planned and implemented measures.	No data on the effectiveness of measures.

, for clarity and ease of interpretation. The current database does not allow for a more detailed analysis of findings. Instead, authors hypothesise possible explanations. This is the first step in a more in-depth analysis and possible direction for future research.

5. Discussion

The analysis of trespassing established temporal and spatial trends of trespassing in Italy was performed, and the examination of trespassing included correlation analyses to uncover underlying patterns. This provided an overview of preventive measures that have been planned or are currently being implemented to mitigate trespassing events in Italy. This study did not seek to explain reasons why trespassing occurs (human factors). The combined dataset does contain descriptions of trespassing. However, they are not continuously collected, and in some cases, there are missing data. Instead, the aim was to provide a statistical analysis of reported trespassing to identify recurring temporal and spatial patterns. Understanding these patterns is a necessary first step that enables more targeted future research into the underlying causes, including social, behavioural, or environmental factors.

The review of findings is crucial for drawing conclusions and determining possible steps for further research. An explanation of key trends and their implications can be used to develop and test theories that enhance railway safety. The last step in data collection for trespassing is the same as in other EU countries due to the ERA reporting requirements. Cameras are only implemented at stations, and no electronic detection systems exist. Some trespassing might go unreported. It is likely that, given the exponential growth of trespassing over the years, the detection process has also improved, as have changes in reporting requirements [36,37,38,39].

The main railway line is identified as a location with a higher percentage of trespassing due to the higher frequency of trains. Previous studies [33,34] have also referred to this. The case could also be that the main railway lines are better monitored. However, the current dataset does not allow for a stronger explanation. Furthermore, more trespassing occurs between railway stations on sections of railway lines rather than at stations, stops, etc. This is also the case in the Czech Republic [25]. The dataset described trespassing and the most frequently occurring behaviour of walking or crossing the railway lines, crossing outside designated points, near or at railway stations [12,36,37,38,39,40,41], which partially answers the former. The description, however, does not contain the human reasons for trespassing, as in previous studies [25,32], and warrants investigations at specific trespassing locations [26]. The analysis of trespassing by DTP areas reflects that more trespassing occurred in Northern and Central Italy. The strong notion is that trespassing events are higher in these areas due to a higher population density, as in a previous study [24]. The DTP areas are not the same as the geographical areas of Italy [44], and the population comparison relative to trespassing occurrence must be analysed accordingly.

The month of May is identified as the month with the highest percentage of trespassing. The only noticeable changes are in the winter months. The same conclusion can be drawn for Saturday, which is the day with the highest percentage of trespassing occurrences. The expected drop after morning peak hours (7:00–9:00) does not occur. Previous studies [24,26,27] have analysed the temporal aspect of trespassing, and this differs from country to country.

Due to the differences in reporting requirements for trespassing, the analysis of various types of trespassing from 2001 to 2023 was more challenging. Rather, more accurate representations of trespassing trends are likely to be produced by analysing minor injuries, serious injuries, and fatalities, as these accidents warrant investigation (with the exception of some minor injuries) [12,36,37,38,39,40,41]. The following question was asked: do these trespassing accidents correlate in any way with hours of trespassing? As results showed a weak or negligible relationship between the two variables, the conclusion can be drawn that other factors influence the occurrence of minor injuries, serious injuries, and fatalities more than the time of occurrence. A previous study [33] identified the length of railway lines [33], while another study [34] identified the higher frequency of trains and speed of trains as other causes for trespassing. The overlapping of variables, frequency of trains, and fatalities was an attempt to determine the connection between these variables, as the authors of [34] mention the frequency of trains as a possible cause. However, only data about trains that stop at the station were extracted from the timetable. Any passing trains or freight trains were not included. Therefore, the analysis for the night period warrants additional research. The correlation between the DTP area and types of trespassing revealed that some types of trespassing are overrepresented or underrepresented in certain areas. For example, in some areas, there should be significantly more trespassing incidents than expected, but that is not the case. The research on factors for trespassing, primarily human behaviour, must be conducted in each DTP area.

Human behaviour was not investigated in detail as a factor in trespassing, as the current combined dataset does not contain the necessary information. The only human behaviour was a description of the trespasser’s movement. The factors that contribute to why people trespass can be extracted from accident reports [6], analysed through video recordings, and identified using a questionnaire, among other methods [30].

The listed prevention measures are currently being implemented or are planned to be implemented on the Italian railway network to mitigate trespassing. These measures were added to complement the existing analysis. The implementation of these measures should be taken with caution, as there are no data on their effectiveness in reducing trespassing.

6. Conclusions

Trespassing on tracks is one of the leading causes of accidents in railway transport. The data available for analysis comprise all recorded trespassing acts from 2001 to 2023 on the Italian railway network. The objective of this study was to analyse temporal and spatial trends of trespassing on the Italian railway network. These involve the investigation of the year, month, day, and hours of trespassing, distribution according to location, distribution according to trespassing type and accident type, and the correlation analysis of trespassing with the hour and with DTP areas. The answers to the research questions revealed trends in trespassing that could be used to identify high-risk locations on the railway network. However, these trends do not reveal the entire nature of these events, or in other words, the following question remains: do they occur because of human behaviour, due to deficiencies in infrastructure, rail operations, or other factors? Nevertheless, due to the extensive size of the rail network, they can still be used to focus on areas that appear to be hotspots for trespassing. The problem arises with the possibility that many trespassers were not detected or recorded. Trespassing, in many cases, was without consequences. However, if train–pedestrian collisions occur in a high percentage of cases, the consequences are fatal. The focus should be on improving the detection of trespassing (e.g., cameras, advanced detection systems on tracks, access prevention, etc.).

This research successfully identified spatial and temporal trends of trespassing in Italy. The analysis revealed the complexity of trespassing and the numerous causal factors that influence it. Limitations are missing data and discrepancies in the database. Data on trespassing were not collected in detail prior to the new requirements posed by the ERA. This research identified temporal and spatial trends; however, it did not focus on factors like the human motivation for the occurrence of trespassing, partially because the dataset did not include a detailed description these factors.

Future research could focus on an in-depth analysis of factors contributing to trespassing occurrences and investigate human behaviour. The limitations mentioned in the database could be addressed by contacting safety authorities to create a comprehensive database and expand the current database by adding new variables. This would serve as the basis for developing prevention measures that target all or specific causal factors of trespassing.