How insights from Earth observation data can optimise rail operations

Groundbreaking satellite technology is empowering the rail industry to make better decisions here on Earth.

In brief:

• Data from Earth observation satellites is transforming rail operations by providing insightful analysis, identifying problems before they escalate and optimising operational efficiency.
• Advanced satellite sensors delivering detailed Earth observation data, combined with AI, enable rail network management to become more efficient, effective, and safer.

Australia's 33,000-kilometre rail network snakes its way through some of the world's most inhospitable landscapes. New Zealand’s 4,000-plus kilometres of rail line runs the length of the country, covering steep slopes and tough terrain, as well as 2,100 level crossings, more than 1,700 bridges and 150 tunnels.

Keeping a close eye on every section of these networks from the ground is an impossible task. But using Earth observation satellite data from space, we can monitor the track remotely, zooming in to spot small problems before they become big ones.

Space technology, or Space Tech, may seem like science fiction. But satellite technology can help us make better decisions here on Earth today.

Tens of thousands of satellites have been launched since Sputnik was sent into orbit in 1957. For decades, the information captured by these satellites remained in the hands of space agencies. Then, in 2008, NASA released its Landsat satellite data to the world. The European Union followed with its Sentinel series a few years later. As the cloud computing giants began connecting information gathered in space to computational power on planet Earth, insights that were once hidden to all but a few, became free and easy to access.

Unlocking Earth’s secrets from space

EY Space Tech Lab harnesses advanced satellite insights to transform rail transportation worldwide.

While satellite data has been around for decades, most use cases were focused on the scientific opportunities. But as the resolution and frequency of imagery has improved, costs have fallen, and commercial applications and new business ventures have emerged.

Satellites in orbit can provide consistent, repetitive observations of specific locations because they pass the same location on Earth at regular time intervals. The lack of atmospheric drag contributes to orbital stability, offering precision measurement in almost real time.

Just as the cameras in our smartphones have become more sophisticated with each year, satellite-mounted sensors are gathering deeper, granular and more varied data, generating insights across a range of applications and industries:

• Optical sensors allow us to see views of Earth from space revealing fine-grained views of our planet’s natural systems and human activity, with commercial providers achieving resolutions of 30 centimetres.
• Powerful hyperspectral sensors on satellites – which analyse a wide spectrum of light in great detail, allowing for specific wavelengths to be tuned and analysed. This means we can unlock capabilities such as detecting specific species of vegetation or assess the quality of water bodies accurately.
• Synthetic aperture radar (SAR) can penetrate clouds, smoke and tree canopy with up-to millimetre levels of precision, observing day and night to generate high-resolution ground imagery and SAR can even detect changes in the Earth's surface to see embankment stability issues early on, before they become disastrous.

The EY Space Tech Lab was established in 2020 and since then has amassed more than 150 use cases across a wide range of industries.

Earth observation tools can detect changes to soil minerals, vegetation types, emissions, water quality and even the slightest deformation to the Earth’s crust. We can measure the growth in a stem of rice to determine the right time to harvest. We can watch vegetation grow, water pool, cities expand.

But that information sits on a shelf unless we can interpret it. When satellite imagery is combined with processing power – backed by artificial intelligence (AI) and machine learning (ML) – we can analyse data and automate processes. We can feed treasure troves of information into complex models that the human mind can’t decipher to uncover new insights and evaluate future risk.

Rail resilience redefined

With precision data from space, we can prevent more problems here on Earth.

When heavy rains caused washouts along the rail line 473 kilometres north-west of Port Augusta, it halted freight services from Adelaide to Perth and Darwin. It took nearly a month to repair the damage along a vast stretch of track across 18 sites. In the meantime, West Australians faced the worst food supply crisis in living memory.

Whilst the weather may have caught people by surprise, the emergence of AI and Earth observation data can improve infrastructure resilience, help prepare for such events in future, help respond quickly to the effects and reduce downtime through the repair process. The water management model developed by the EY Space Tech Lab is one application that could be used in future events to identify significant areas of water pooling, alongside historical rainfall and topographical data, to help the rail sector minimise damage from water and other future risks.

This is just one of many use cases for the rail sector. Other key use cases are detecting vegetation encroachment risks to reduce disruption and improve safety along the rail network; improving how railway crossings are managed and detecting any impediments that may disrupt the network; improving track safety by monitoring the network for any significant changes and detecting embankment blockages along a rail network to improve safety.

An overview of the main use cases for rail operators are shown below.

On track to tomorrow

Earth observation breaks new ground in rail operations

One collaboration between the EY Space Tech Lab and a mining company used Earth observation data and machine learning to quantify which problems had the biggest impact on their network and take the guesswork out of track maintenance.

For another client, we meshed satellite imagery with machine learning to create a water detection model that allows us to assess potential risks. And for a third, we’ve identified where vegetation encroaches on the track and then, with the help of predictive analytics, determine where problems are most likely to occur in the future. We know what the vegetation will look like in six months’ time with a 94% degree of accuracy, which can support significant cost efficiencies in vegetation management.

These are just a handful of examples of how satellite imagery can help asset-heavy enterprises like rail operators to monitor infrastructure, plan for preventative maintenance, contingencies and redundancies, and reduce human interactions with hazardous equipment.

To shape the future, we must first establish a baseline. High resolution data is available now. NASA has captured imagery at 30-metre resolution, for more than 25 years. ESA’s Sentinel data, with imagery at 10-metre resolution, stretches back 10 years.