Indian university uses IoT to predict landslides
August 9, 2018
Indian university Amrita Vishwa Vidyapeetham (AVV) is using IoT technology to save lives in mountainous regions of the country by installing systems that give advance warnings of landslides so people can be safely evacuated before disaster strikes.
After commissioning India’s first such system in Kerala’s Western Ghats, it is now readying the second installation in Sikkim to guard against rainfall-induced landslides in the Sikkim-Darjeeling belt. The project is jointly funded by the Ministry of Earth Sciences, the government of India and AVV.
Maneesha Ramesh, director of AVV’s Amrita WNA centre for wireless networks and applications, who spearheads landslide research at the university, said: “Landslides are the third most deadly natural disasters on earth, killing over 300 people every year globally. The number of fatal landslides in India is higher compared to other countries.”
A report by Indian Roads Congress estimates that 15% of India’s landmass is prone to landslide hazard, including areas such as the Western Ghats and Konkan Hills, Eastern Ghats, North East Himalayas, and North West Himalayas.
“In North East Himalayas, the Sikkim-Darjeeling belt is at the most risk of landslides, which is why we chose this area to install our landslide detection system,” Ramesh said.
As per David Petley’s global database on landslides, the world’s top two landslide hotspots exist in India – the southern edge of the Himalayan arc and the coast along south-west India where the Western Ghats are situated. Not only is tectonic activity higher in the southern Himalayan arc, monsoon rains and manmade changes to the slopes have made these hills much more prone to landslides.
“This system has been actively monitoring the area for landslides and has issued several successful warnings to date,” said Venkat Rangan, AVV vice chancellor, about the system deployed in the Western Ghats in 2009 in Kerala’s Munnar district. “Impressed by this success story, the government of India approached Amrita to develop a similar system for the Sikkim-Darjeeling region which is very active geologically and is vulnerable to rainfall-induced landslides”
The IoT system for landslides being installed in Sikkim was custom developed for Himalayan geology. It consists of more than 200 sensors that can measure geophysical and hydrological parameters such as rainfall, pore pressure and seismic activity. It will monitor a densely populated area spanning 150 acres around the Chandmari village in Sikkim’s Gangtok district. This area has seen landslides in the past, the first one being reported in 1997.
The system collects real-time, continuous data from the sensors, performs basic analysis at the field management centre on the site in Sikkim and relays it to the data management centre at AVV in Kerala’s Kollam district. The university researchers are using these data to characterise and learn the geological and hydrological nature and response of the hill with respect to the dynamic and real-time meteorological variations to develop the landslide early warning model for that area.
To improve the system’s reliability and enhance the early warning duration, a three-level early warning model has been developed. The first level, based on the rainfall threshold, has successfully completed the testing phase and is ready to go live and issue alerts for potential landslides at the state level. In the second level, the system would generate a factor of safety value for various points on the hill in real time that will provide a more specific warning for the Chandmari region based on the rainfall, moisture and pore pressure sensor data from the field. In the third level, the system would use data derived from the movement and vibrational sensors to issue landslide detection warning.
“This multi-level warning system will help disaster management authorities to take steps to mitigate and manage potential landslide threats in a proactive and effective manner,” said Ramesh. “In this regard, Amrita has performed several community engagement programmes to disseminate knowledge regarding the impact of landslides, the working of the proposed warning system and its capability to warn about imminent landslides.”
Rangan added: “For inculcating this vision in the faculty and student community, Amrita has integrated several unique initiatives for the natural fusion of multiple disciplines leading to the delivery of an affordable solution for saving human lives. Indian higher education has to be geared in this direction to solve the pressing needs of the world.”
Landslides can be triggered by natural causes such as vibrations from earthquakes and the build-up of water pressure between soil layers due to prolonged rainfall or seepage. In recent decades, human-made causes have become significant in triggering landslides, including removal of vegetation from the slopes, interference with natural drainage, leaking water or sewer pipes, modification of slopes by the construction of roads, railways and buildings, overloading slopes, and vibrations from traffic.
“Even though many factors contribute to making a slope vulnerable to landslides, major triggering factors are rainfall and earthquakes,” said Ramesh. “Areas of steep slopes, those having tectonic activity and hilly terrains with heavy rainfall are at great risk of landslides.”
Landslides can be estimated at different accuracy levels through rainfall threshold based models, in-situ sensor-based monitoring technology, interferometric synthetic aperture radar based technology, ground-based radar technology, electrical resistivity tomography and satellite images.
“Rainfall threshold based models are most commonly used in the world and have issued several early warnings, but with more false-positives than accurate predictions,” said Ramesh. “In-situ monitoring methods deployed on the ground like what Amrita is setting up in Sikkim, can issue early warnings much more accurately than the rainfall threshold model. The landslide monitoring system installed by Amrita in Munnar has issued several successful early warnings till now.”
Unfortunately, the landslide-warning system being installed by Amrita researchers in Sikkim was severely vandalised by miscreants in April this year. This laid waste to two years of field work by the scientists, forcing them to start all over again.
According to Ramesh, several steps can be taken to reduce fatalities due to landslides.
“In the case of landslides, forewarned is forearmed,” she said. “More accurate landslide databases need to be maintained and regional, as well as site-specific rainfall threshold models, developed. Low-cost in-situ monitoring technologies have to be deployed in landslide-prone terrains. People have to be educated regarding landslides and the risks involved. Social media and mobile phone apps can be developed to collect data about the chances or precursors for landslides from the people and other governmental organisations.”
Amrita WNA has also initiated a landslide-research project with the British Geological Society and the UK Met Office to develop regional thresholds for landslide warning in the Sikkim-Darjeeling belt and the Western Ghats.
In 2008, Amrita set up India's first landslide laboratory, which serves as a test bed for development and validation of systems deployed in landslide-prone areas. It can create small landslides under controlled conditions and act as the feedback system for field deployment. The university pioneered the wireless sensor-based system for detection of landslides that is more accurate than the rainfall threshold model.