2 November 2010—Just after midnight on 8 August, a massive landslide sent nearly 2 million cubic meters of rock and mud hurtling through several towns in Zhouqu County, China. The natural disaster killed more than 1500 people in the county, which sits in a valley between two mountains in the Gansu Province. Nearly three months later, more than 250 people are still unaccounted for.
Though their breakthrough is too late for the Zhouqu disaster, researchers at Loughborough University, in Leicestershire, England, reported late last month that they have developed an electronic system that can warn of an impending landslide by picking up the telltale sounds of shifting earth.
Engineering geologists have known for decades that in the buildup to a landslide there is barely perceptible movement (on the order of about a millimeter a day) along an underground plane known as the shear surface. The shear surface controls the volume of the landslide and the speed at which a chunk of earth can move down a slope. Scientists also know that it’s possible to predict a landslide by checking for sounds made by particles rubbing against one another at the shear surface. The problem, says Neil Dixon, a professor of geotechnical engineering at Loughborough, is that the sound is quite faint and attenuates rapidly. Waveguides— which are made of a hollow pipe inserted in a hole dug deep enough to reach the shear plane—make detecting the sound a little easier. But soft materials such as clay make sounds so faint that they strain the sensing ability of even the most discriminating electronic ears.
Earlier attempts at landslide warning systems had problems that the Loughborough group has solved. Like the new device, earlier versions converted acoustical energy to digital signals. Some operated on low-frequency sound, which travels farther before it attenuates. The main drawback was that this range of frequencies is in a band that’s very noisy, and the systems returned too many false positives. But high-frequency devices struggled to capture the shear surface sounds because of attenuation and required off-site computing to handle the system’s data-processing duties.
The Loughborough team sidestepped these problems with a high-frequency device that takes readings at a rate of 20 to 30 kilohertz. They limited the effect of attenuation by boosting the sound with as low tech a solution as it gets: gravel. They filled the space between the outside of the waveguide and the borehole in which it sits with gravel, which grinds as the ground around it shifts, sending noise up the waveguide.
But Dixon is quick to note that movement isn’t the key indicator that a landslide is imminent—acceleration is. "If you’re detecting movement of millimeters per day consistently, you need to investigate," he says. "Acceleration is cause for alarm." The new device measures acceleration by counting the number of times the voltage exceeds a preset threshold. Setting a floor for the signal and simply counting each time the threshold is exceeded significantly cuts the amount of data processing required and dramatically reduces extraneous noise, all without degrading the device’s sensitivity. Dixon says its "hearing" is so acute that it can detect movement as tiny as 0.001 mm per minute. Though the system is continuously translating sound waves to voltages, a tally of the number of times the threshold voltage is exceeded is taken for every 15-minute period. Stable ground could deliver no hits at all, while shifting soil could generate hundreds, thousands, or hundreds of thousands of significant data points in the same time frame.
Dixon says that because geology varies and the history of a particular slope over many years is unique, only the acceleration of movement can offer clues to an impending landslide. Nor is there a standard time period between the sounds from the shear surface that foreshadow a massive fallout and the point when the landslide actually begins. A multiton chunk of earth can creep for months without breaking away and cascading down a hillside. Or there could be just an hour of warning.
Besides acting as an early warning system, the device can be used to measure the effectiveness of efforts to stabilize soil, such as draining or regrading, Dixon adds.
Among the tasks remaining for the group, which includes researchers from Loughborough University and the British Geological Survey, is refining the device’s sensors and bringing down the cost, which Dixon estimates is currently a few thousand U.S. dollars. He says this is critical if the devices are going to be installed in significant numbers in poor countries where landslides claim thousands of lives each year. He says the team hopes to have a version ready for commercial production by the second quarter of 2012.
This article was modified on 18 November 2010.