On 3 November 2025, the lead bogie of a passenger train derailed at Shap in Cumbria after hitting a landslip. The train was travelling at 133.5km/h (83mph) and continued to run derailed for 560 metres, according to the Rail Accident Investigation Branch (RAIB). Of the nine staff and 86 passengers on board the train, four were treated for minor injuries.
In its initial investigation, RAIB says the cause of the landslip was due to a washout and that a drainage channel was unable to accommodate the volume of water during a period of extreme rainfall.
On 19 December 2025, RAIB published its Urgent Safety Advice 01/2025: Use of remote earthwork monitoring equipment. It was identified that the cutting slope at Shap was fitted with equipment to detect ground movement. This was recording data, although it had not been formally entered into service and was not sending alerts to the Network Rail control centre. However, similar equipment was operational on other parts of the railway infrastructure.
Around four hours before the derailment, the sensors nearest to the landslip began to show minor movement of the earthwork, but below the threshold needed to trigger an alert. This continued for the next two hours. At around 04:30, when it is believed the landslip occurred, the two sensors in the path of the debris were tipped over, but too quickly to detect and transmit their movement and to generate an alert. RAIB also concluded that the sensors’ wireless signal was also unable to pass through the layer of material which covered them.
RAIB is to be praised for quickly making the industry aware of its initial findings and Safety Notice. The RAIB full investigation will take time, but it will be very thorough and determine the sequence of events which led to the derailment and include consideration of the design, maintenance, and management of earthworks and drainage in the area, effectiveness of on-site monitoring equipment, operational response to adverse weather warnings, the performance of the train during the derailment, and any underlying factors which might have contributed to the derailment.
Climate change
Earthwork failures such as at Shap are one of the major risks to rail, with the increasingly volatile climate increasing the risk. Several factors make railway slopes more likely to collapse. These include weather conditions, the type of material, and the steepness of the slope, along with other influences such as vegetation and human activity. Failures can happen suddenly or develop gradually over time.
As would have appeared to have happened at Shap, extensive wet weather can soak the ground, with water building up inside slopes and making them weaker, leading to slow-moving landslides or earthwork failures. Dry conditions can make the structures shrink and crack, then, when it rains, water seeps into the cracks increasing the risk of failure. Dry weather can also reduce the vegetation that binds the soil, further increasing the risk.
Cold weather and freezing cycles can break down the soil as water in the cracks freezes and expands. Strong winds can strip vegetation or uproot trees, leaving loose soil that further raises the risk of failure. Both of these increase the risk of slope failures. The Department for Transport, Met Office, and the British Geological Survey, have created a series of transport hazard summaries to explain the natural hazards that are not the result of malicious acts, and how they may change in the future.
They say that the UK climate is projected to become more variable, with more extremes of hot, dry, and wet weather. Hotter, drier summers will dry out soils and climate change is expected to worsen the lengths and severity of droughts. Increased heavy rainfall combined with periods of dry weather are expected to lead to an increase in earthwork failures, which will require additional inspections and maintenance to ensure safety.
Slope failures similar to that at Shap have occurred many times over the years. On 28 January 2025, a landslide caused disruption to passengers travelling to Gatwick Airport by train, blocking one of the lines on the route.
In August 2020, extreme rainfall overwhelmed a poorly installed and maintained drainage system near Carmont, Aberdeenshire, washing debris onto the railway line. A train derailed, tragically causing the deaths of three people and seriously injuring three others. The line was closed for almost three months. An investigation found that the derailment would have been prevented with correct installation and regular maintenance of the drainage system, prompting improvements to the management of earthworks and drainage on the railway network by Network Rail.
A 350,000-tonne landslide near Harbury in January 2015, blocked a rail line between Southampton and the West Midlands. Freight and passenger trains between Banbury and Leamington Spa were suspended for six weeks.
And its not just rail that is affected. In October 2023, heavy rain triggered seven landslides at the Rest and Be Thankful pass on the A83 in Scotland. This landslide-prone stretch of road was closed for several days, and 10 people had to be airlifted after becoming stranded. Transport Scotland reported spending £3.6 million clearing landslides on the A83 that year.
More recently, on 22 December 2025, the canal wall collapsed on a stretch of the Llangollen Canal in Whitchurch, Shropshire, with a 50-square-metre hole opening up. Two boats fell into the hole, with another left teetering on the edge above. Several people were evacuated from boats in the vicinity and fortunately no one was hurt. The Canal & River Trust, which is responsible for the affected stretch of canal, said that the embankment had been inspected in recent weeks and that no issues had been identified.
Radio systems
Going back to rail, fortunately at Shap there was no other train involved in any collision and the GSM-R radio system was used to alert and stop any other trains and raise the alarm. It can be easy to look at the past with rose tinted glasses and assume things are now worse, but this is not necessarily the case. Some years ago, near to Shap, there were two landslip failures on the adjacent Settle Carlisle line which didn’t have the benefit of GSM-R or the robust procedures that are now in place.
On 31 January 1995 near Aisgill, Cumbria, at about 18:55, a train was derailed by a landslide and was subsequently ran into by a similar train travelling in the opposite direction. In the 24 hours before the derailment, 2.5 inches (64mm) of rain had fallen in the area.
A Carlisle to Leeds train could only proceed as far as Ribblehead railway station, Settle, as the route was blocked by flooding, so it had to return to Carlisle. The train hit a landslide north of Ais Gill Summit. The driver did not observe the landslide before the crash, could not attempt to stop, and the first carriage derailed across both tracks.
The forerunner of the GSM R system was the National Radio Network (NRN). This had been designed for the British Rail Regional structure, but the Railtrack 1994 reorganisation had failed to take into account the radio design. The Settle Carlisle route was manged by the Railtrack London North East Zone, the radio base station control equipment was managed by the Railtrack North West Zone, and the radio emergency control point was at the Railtrack West Coast Zone in Crewe!
The injured driver managed to radio Crewe control room using the emergency button. The controller at Crewe had the ability to make a ‘group call’ to all trains in the area, which might have alerted the second train to the obstruction in time to prevent the collision, but no training had been provided how to do this. The transcript of the call was as follows:
Driver: Blea Moor to Carlisle, derailed blocking both roads. Can you stop the job between Kirkby Stephen and Blea Moor.
Control: We’ll arrange all that, driver. Over and Out.
This may have given the false impression that the approaching southbound train would be warned to stop, and no precautions required by the Rule Book – i.e, walking 1.25 miles (2.01 km), placing three detonators, and displaying a red handlamp – were taken. The Crewe controller informed the London North East Controller in York of the incident and although they tried to contact the individual trains by radio, they had no ability to make a ‘group call’ to all trains in the area.
About six to seven minutes later, the southbound 17:45 Carlisle to Leeds train struck the derailed northbound train, with the collision fatality injuring the conductor of the derailed train. Twenty-six people were treated in hospital with five serious injuries. The radio system had logged all the calls, but on different systems with no centralised clock. This meant the data presented to the inquiry had numerous time errors.
Five years later, on 15 January 1999, a similar incident occurred around 10 miles (16km) from the site of the 1995 accident, at Crosby Garrett Tunnel, north of Kirkby Stephen. A landslide caused a Carlisle-bound Class 156 Sprinter to derail, and the train went through the tunnel.
Things had improved, with the Settle Carlisle line all managed by the North West Zone and its control in Manchester, the NRN had been improved with better call logging, and all the controllers had been trained in the ‘group call’ facility. However, while no fault with the radio system could be identified, no emergency call was instigated by the driver. This may have been due to the driver not holding the emergency button for the required 0.5 seconds, or that the train radio was faulty.
More importantly, though the driver promptly left the train to protect it with detonators, he had made it only 300 yards (270 metres) when he heard an approaching train and managed to place a single detonator and wave a red signal lamp. The driver of the approaching train almost managed to stop but collided with the derailed train at just under three miles per hour (4.8 km/h), crushing the cab and pushing the train back through the tunnel. There were no fatalities and the injuries were not serious.
Thirty years later, geotechnical resources, technology, and techniques, along with the radio emergency system using GSM-R, are much improved. With GSM-R a driver can easily initiate their own emergency group call. However, with incidents such as Shap and increasingly adverse weather, what more can be done and how can technology help?
Risk mitigation and technology
The transport hazard summaries say that various methods can be used to reduce the risks associated with landslides and earthwork failures. Examples include: (i) regular inspection, assessment, and maintenance, especially of drainage systems and at-risk structures; (ii) upgrading or replacing earthworks to meet modern design standards; (iii) engineering solutions to stabilise slopes and prevent further movement; (iv) active monitoring of high-risk sites using remote sensing methods and other technologies; (v) implementing operational measures like speed and weight restrictions near vulnerable earthworks; and (vi) developing climate change adaptation plans to identify the most vulnerable regions and improve preparedness and resilience.
There is no one ideal solution to detect earthwork and slope failures, which illustrates the challenge. However, the technology available includes:
Tilt meters on posts provide a cost effective, quick to install, and simple solution. However, they provide limited asset condition ahead of any failure and there is very little predictive capability and false positives are possible. Tilt meters can also miss translational failures, unless they are designed and installed across a slip plane.
Drone examinations can provide good physical inspections of slopes and drainage channels but require competent operators and can’t provide continuous monitoring. LiDAR (Light Detection and Ranging) technology can help drone-based surveying by capturing point clouds. These can provide detailed terrain information in areas with dense vegetation.
Satellite InSAR (Interferometric Synthetic Aperture Radar) monitoring can deliver advantages over traditional monitoring methods, such as cost savings, minimising time spent on the track and on slopes, faster data collection, repeatable measurements for trend analysis, and reducing human error and subjectivity. But like any inspection it can’t deliver continuous monitoring, and both drone and satellite monitoring will help, rather than replace, manual examinations.
In-ground instrumentation can provide good predictive capability with continuous condition monitoring and good event detection. However, this solution requires drilling and a bespoke design, making it expensive to deploy and standardise.
Fibre optic sensing offers potential large area coverage and has great theoretical capability. It’s not a new technology and Network Rail has experience of trialling various applications for many years. Standalone systems can be expensive to provide and repair, but could something be provided using the existing lineside fibre cables which already run along many railway routes?
Twenty years ago, fibre optic sensing could detect copper cable theft but suffered from too many false positives. However, over the years optical and processing technologies have improved greatly to deliver far beyond simple acoustic sensing. Systems have been connected to the railway fibre cables to provide continuous railway monitoring over tens of kilometres. Slow and progressive track formation movement has been trend-monitored with a resolution of tens of nanometres (<0.0001mm), providing evidenced insight to focus further interventions. Testing at The Global Centre or Rail Excellence has demonstrated the detection of events from the trackside cable route and showed reliable locational detection of low-volume spoil-drops.
The technologies available and under development to help manage railway earthworks is impressive. However, there is no silver bullet and it’s likely that the solution is for a number of technologies to be brought together to manage the issue. For example, a slope might be characterised using LiDAR collected from a drone combined with aerial imagery, monitored using satellite, with an early warning system based upon the use of smart sensors or fibre optic monitoring. The key will lie in the integration of multiple approaches to provide a comprehensive understanding of the asset and managed by competent experienced geotechnical engineers who are assisted, but not replaced, by AI.
Further information on the hazards discussed in this article can be found in this series of Transport hazard summaries:
Image credit: