Train horns: Pedestrian safety and residential wellness

The iMOVE project, Train horns: Broader social effects and pedestrian simulations is a mixed-methods research that offers comprehensive insights into how train horns impact pedestrians’ behaviours at crossings.

The study also examines nearby residents’ sleep quality and mental health. It helps inform transport industry organisations and policymakers about the effectiveness and impacts of train horns on road users and the broader community.

The research was conducted by the Australasian Centre for Rail Innovation (ACRI), the National Transport Research Organisation (NTRO) and the Queensland University of Technology (QUT).

The final report is available for download below.

Objectives

Railway lines and crossings are typically marked with whistle boards directing train drivers to sound their horn to enhance safety at railway crossings.

But this may negatively impact sleep for residents living nearby. Not much was definitively known about this issue, so this Queensland-based project harnessed three studies combining objective data and self-perceptions to offer baseline information.

Those studies sought insights on:

1. How train horns impact nearby residents’ sleep quality and functioning;
2. Their self-reports about horns’ impact and
3. The horns’ effect on pedestrian behaviour around level crossings.

Methodology

The project comprised three studies:

1. Recording and monitoring participants’ reactions in virtual reality scenarios around a level crossing with trains and horns
2. An online survey of participants living up to 1km from a train station or railway tracks for their self-perceived effects of train horns; and
3. An objective measurement of participants’ sleep using an electronic wristband.

The first study comprised 37 participants aged between 18 and 68. They wore a virtual reality headset that simulated walking 100m to a railway level crossing. Linked to the headset was an Xbox adaptor to control their virtual character. Subjects listened to playlists via iPhone 6 earphones to create a realistic audio distraction. This allowed researchers to monitor and measure head movement and eye positioning to gauge their responses.

Researchers tested in virtual environments spanning active or passive crossings with or without a train (travelling 80km/h) and monitored reactions to a soft or loud horn. Participants were also asked to talk about their thoughts and actions and describe what they heard and saw as their cyber character navigated the scenarios.

The analysis used quantitative and qualitative methods.

For the objective measures, SCANeR™ studio software and the recorded scenario video were analysed using a Generalised Linear Mixed Models (GLMMs) approach. This model helped gain insights into the influence of independent variables (such as passive/active level crossing control, no/soft/loud train horn, music on or off) and how factors interacted with the dependent variables. Researchers used the chi-square test to determine correlations between environmental factors and the sequence of objects participants perceived.

The second study was an online survey of broader social effects of train horns and involved 334 participants who lived near one of seven Brisbane railway stations.

The survey aimed to measure the following:

• Frequency they had noticed and been annoyed by trains (light and vibration exposure) and train horns in the past 12 months.
• Rate the severity of any experience of depression, anxiety and stress in the previous week (researchers used the Depression, Anxiety and Stress Scale-21)
• Self-reports of sleep quality over the past month, spanning habits, duration, efficiency, disturbance and time to get back to sleep afterwards, medication use, daytime dysfunction and waking-hours enthusiasm (researchers used the Pittsburgh Sleep Quality Index), and
• Perceived sensitivity to noise via a five-point Likert scale (using the 21-item Weinstein Noise Sensitivity Scale).

Based on the responses, researchers divided people into three zones. Those who lived within 450 metres of a railway station (zone 1), 450—1000 metres away from a station and railway tracks (zone 2), and more than 1000 metres (zone 3). The analysis involved one-way ANOVA (or its non-parametric equivalent), pairwise comparison tests and a binary logistic regression model.

That model assisted researchers in assessing how six independent variables (age, gender, train-horn impact group, night-time or daytime train horn frequency and noise sensitivity) impacted the likelihood of participants reporting poor sleep quality.

For the third study, 36 participants from the online survey volunteered to wear electronic wrist devices (GENEActive to collect actigraphy data) to measure their sleep, daily activities and temperature over about a week. They were aged between 31 and 75, with about half living close to a train station or level crossing (the impact group) and the rest between 1 and 3 kilometres away (baseline group). Participants also had an NSRT_mk4 device setup on their balcony or yard to record sound levels. The study also used a NGARA Sound Acquisition system to measure real-time sound pressure and continuous sound levels. These devices were installed in the railway station office to collect train horn sound data.

Researchers used GGIR software integrated with R to analyse the actigraphy data, and Audacity software for the NGARA system’s audio data. This allowed them to match participants’ environmental noise levels and sleep patterns with actual train horn acoustic information.

Researchers also delved into the participants’ online survey responses for subjective data to produce a descriptive analysis. The study developed a Generalised Linear Model to examine which factors influenced participants’ total sleep time in the impact group.

Report findings

The project found the sound of train horns – whether loud or soft – had little impact on a pedestrian’s behaviour and decision-making. The project also showed that train horns did not influence subjects’ emotional states.

The project revealed a disconnect between residents’ self-perceptions and objective data. Residents in the impact zone rated their sleep quality lower than those in the baseline zone. More impact-zone residents reported more than five interruptions to their sleep compared with the other group.

Listed below are some of the main findings:

Pedestrian safety

• Most pedestrians (82%) stopped at the level crossings after they perceived the approaching train and before the train horn was sounded.
• The use of train horns (on either loudness level) did not show any significant influence on pedestrians’ walking behaviour and decision-making.
• Level crossing control significantly influenced pedestrian crossing behaviour, and pedestrians behaved more cautiously at active level crossings than at passive level crossings.

Sleep and mental health

• Residents in the impact zone reported significantly poorer subjective sleep quality and longer sleep latency than participants living in the baseline zone.
• Nighttime train horn frequency and noise sensitivity were significant factors for predicting sleep quality in terms of good or poor.
• Increased train horn frequency at night and higher noise sensitivity increased the likelihood of participants being involved in the poor sleep quality group.
• Participants’ emotional states, i.e., depression, anxiety and stress levels, were not influenced by train horns.
• The maximum sound pressure level and sound impulsiveness rate of NGR trains were significantly higher than EMU/IMU/SMU trains.
• Train horn sound impulsiveness rate was negatively associated with total sleep time.

Recommendations

Researchers have suggested the need to sound train horns at active crossings could be reviewed given the horns’ marginal effect on motorists and pedestrians.

The researchers wrote in the final report: “The think-aloud approach showed that train horns were barely rated as a notable feature by pedestrians at both active and passive level crossings, which helps explain why the train horns played a minor role in influencing pedestrians.”

However, the researchers say this does not mean sounding train horns at level crossings has no safety value.

Future directions for the work

Future research could examine train horns’ effectiveness on other types of road users, such as motorcyclists and cyclists. As well, train horns’ acoustic features and whistle board locations should be studied to maximise their effectiveness while reducing noise impacts.

Also regarding further research, it was noted that:

It would be ideal if the equipment could be setup at a location closer to the whistle boards or level crossings where train horns are normally sounded, assuming that a waterproof area and power supply are available. For the residence noise monitoring, it would be better in the future if another set of NSRT_mk4 could be installed in participants’ bedrooms to record the indoor noise level.

Expected project impacts

The findings from this study, revealing the minimal effect of train horns on pedestrian behaviour, yet negatively affecting the sleep patterns and mental health of nearby residents, is expected to lead to a review of the requirements for sounding train horns at active level crossings. Modifications to train horn acoustics may also be an outcome of this study, to retain the safety benefits of their warnings to road users but with less noise impact.

Paul Bennett: Senior Technology Leader, Asset Performance, NTRO

Download the report

Download your copy of the final report, Train Horn – Broader Social Effects and Pedestrian Simulations, by clicking the button below.

Included in the 65-page report is more detail on the background, methodology, findings and more, an includes a copy of the sleep diary and survey form.

DOWNLOAD THE REPORT