Smart boulders for real-time detection of hazardous movement on landslides

Landslides present a substantial hazard across coastal and mountainous regions in Europe and worldwide, and are becoming increasingly prevalent due to extreme rainfall linked to climate change. There is a need to develop new technologies for landslide monitoring and early warning systems, as traditional approaches alone are insufficient due to low temporal resolution and high costs. The SENSUM project (smart SENSing of landscapes Undergoing hazardous hydrogeologic Movement) has deployed manmade boulders, called SlideCubes, that monitor landslide movement in real-time across two coastal slow-moving landslide sites in southern England (Lyme Regis and Isle of Wight).

SlideCubes are embedded with low-power low-cost sensors that comprise an inertial measurement unit (IMU with accelerometers and gyroscopes) and magnetometers. The SlideCubes are part of a wireless sensor network (WSN) that communicates via Long Range Wide Area Network (LoRaWAN) and Internet of Things (IoT) technologies. Rain gauges and other third-party sensors can be easily integrated into the network to provide additional data sources. Our novel WSN allows for near real-time wireless monitoring of the landslides, only requiring field visits to replace sensor batteries every 9-12 months. The sensors are motion-triggered, significantly saving battery power, meaning the WSN requires little and less frequent maintenance than other sensor-based monitoring approaches. This allows long-term remote measurement of landslide kinematics (inferred from SlideCubes) and initiation of movement, which is key for early warning.

In the present work, initial findings from the SlideCubes installed at two UK-based sites are discussed. The movement events detected and recorded over 2 years are validated by periodic GNSS and drone imagery surveys. We present an overview of temporal and spatial motion across both landslide sites and evaluate sensor performance. Using gyroscope and accelerometer readings from field and laboratory data, we demonstrate how types of motion (e.g. rolling, sliding) can begin to be categorised, which is not possible with the accelerometer alone. This research will be developed in future with machine learning to detect hazardous movement including large magnitude catastrophic events. These findings will be integrated into a SENSUM early warning online portal, in development, for use by stakeholders.