Tikalon Header

Landslide Sensor

November 15, 2010

As recent hurricanes, earthquakes and volcanic eruptions have reminded us, we live on a dynamic planet. Although weather satellites help us to prepare, somewhat, for major hurricanes, earthquake detection hasn't seemed possible. Seismology helps somewhat in the prediction volcanic eruptions. Then there are landslides, which are abrupt flows of tremendous quantities of soil. Since most of the action occurs below the surface, monitoring the shift of position at the surface over time is not that helpful for prediction.

One of the
paradoxes of Zeno of Elea is about millet seeds. The paradox is that if a millet seed is dropped, no sound can be heard. However, if a bushel of seed is poured, sound can be heard. Zeno's paradoxes concerned the concept of infinity and the infinitesimal, and his thoughts about this millet seed experiment were not just that a single seed must make a sound, but that the smallest part of a millet seed would make a sound. Nowadays, we have microphones that can capture the sound of a millet seed dropping, and also signal processing techniques that make sound a valuable diagnostic tool for materials and machinery. We can hear the first cracks in a concrete bridge, and we can monitor the acoustic emission spectrum of a machine to discover when it needs maintenance. Acoustic techniques are being applied to give an early warning of landslides.

We've all had some experience with acoustical detection of fault conditions. We know there's something wrong when our automobile brakes start to squeak, or when the
mechanical drive on our computer starts going clack-clack instead if click-click. Over the last three and a half years, a team of scientists at Loughborough University, Leicestershire, England, and the British Geological Survey have been developing an acoustical sensor for detection of landslides.[1-4] As shown in the figure, the sensor is quite simple in construction, but it isn't used singly. It's deployed in an unstable hillside as a network of units connected to a central computer for analysis. Neil Dixon, principal investigator on the project and a professor of geotechnical engineering at Loughborough, says it's been known for fifty years that "...the movement of soil before a landslide creates increasing rates of noise."[4] Noise created by subsurface movement builds to a peak as the slope becomes unstable, and then there's a landslide. Movements of about a millimeter per day need to be monitored.[1]

ALARMS Landslide Sensor

ALARMS Landslide Sensor

The project was executed in two stages. There was a research stage, 'Assessment of Landslides Using an Acoustic Real-Time Monitoring System (Alarms) that was followed by a validation stage, Alarms: Low Cost Sensor Development and Exploitation. The principle of the sensor is explained in a press release by the Engineering and Physical Sciences Research Council, which is the primary agency for physical science and engineering research funding in the UK. The patent-pending sensor consists of a piezoelectric transducer, buried in the soil, that sits on top of a steel tube that acts as a waveguide that directs the generated sound to the piezoelectric transducer. To allow for efficacy under different soil conditions, the steel tube is centered in a borehole that's filled with gravel. It's the sound of the shifting gravel, grinding against itself and the tube, that's heard. Ultrasonic frequencies are monitored to eliminate background noises, such as nearby vehicle traffic. The signals are acquired at a rate of 20 to 30 kilohertz.[1] Says Dixon, "The development of low cost independent acoustic slope sensors has only become possible in very recent times due to the availability of microprocessors that are fast, small and cheap enough for this task."[3] The steel tube can be many meters long if the geological conditions warrant monitoring at deep soil levels.

Computer analysis is able to detect the slow, millimeter per day movements that indicate instability. The sensor is able to resolve movement of just a micrometer per minute, but the more important signal is if the time internal between sonic bursts is steadily decreasing. This indicates an accelerated movement, and it's indicative of a pending landslide. To detect such acceleration, the system counts the number of noise spikes that exceed a fixed threshold in fifteen minute intervals and compares the count in each interval with the count in the previous interval. If the ground is stable, there might be no noise pulses in the interval, whereas significant movement could generate thousands.[1]

The current cost per sensor is a few thousand dollars, but the team hopes to reduce cost to allow a widespread deployment of the system. The stakes are high. In August of this year a massive
mudslide in Zhugqu County, China, killed more than 1,500 people.[1]

References:

  1. Willie D. Jones, "Sensor System Yields Landslide Warnings," IEEE Spectrum, November, 2010.
  2. Loughborough University develops landslide technology with British Geological Survey, Loughborough University Press Release No. PR 09/144, October 21, 2010.
  3. The sound of the underground! New acoustic early warning system for landslide prediction, Loughborough University Press Release No. PR 10/164, October 22, 2010.
  4. The sound of the underground! New acoustic early warning system for landslide prediction, Engineering and Physical Sciences Research Council Press Release, October 21, 2010.