Big Data Analytics and Landslide Prediction

Published Oct-09-18

Breakthrough:
Data-driven models and tools to predict landslides and mudslides up to two weeks in advance.

Company:
University of Melbourne, Australia

The Story:

Big Data Analytics and Landslide Prediction Landslides are large movements of rock, earth and debris down a slope driven by gravity. They can occur suddenly or move slowly over a longer period and can be triggered by a number of factors. These include rainfall, seismic activities and human causes such as mining operations.

Landslides can cause widespread destruction and fatalities such as the landslide in Kedarnath in India in June 2013 that killed more than 5,700 people.

Big Data and Landslide Prediction

Predicting when a landslide event is more likely to happen is notoriously difficult. However, the task is being made easier with an advanced modeling tool developed at the University of Melbourne. It uses big data analytics and applied mathematics to predict the boundary of where a landslide is likely to occur up to two weeks before it occurs.

Warning signs are always there in the lead up to an event but the difficulty is predicting where they will occur. “These warnings can be subtle,” said Professor Antoinette Tordesillas from the University of Melbourne's School of Mathematics and Statistics. “Identifying them requires fundamental knowledge of failure at the microstructure level – the movement of individual grains of earth.”

Scientists cannot see the movement of these individual grains but they can identify the properties that characterize failure in the small-scale. Among them is how their highly disordered movement becomes synchronized over time. In the beginning, the movement is disordered but it becomes ordered closer to the point of failure.

The model developed by the university analyzes and decodes the data on the movement and turns it into a network from which it can extract relevant data patterns on motion and how they are changing with space and time. Detecting the ordered motions early enough is where the system will come into its own.

The potential of this analytics technology extends beyond the prediction of landslides. “We can now predict when a rubbish landfill might break in a developing country, when a building will crack or the foundation will move, when a dam could break or a mudslide occur,” commented Professor Robin Batterham from the Department of Chemical and Biomolecular Engineering at the University of Melbourne. “This software could really make a difference.”

Predicting Earthquake-Triggered Landslides

Meanwhile, researchers at Indiana University have developed a model to help experts estimate the likelihood of landslides that will be caused by earthquakes anywhere in the world. The estimates can be available within minutes

The model describes a mathematical relationship between where landslides happen and five variables. These variables include how much the ground shook during an earthquake, the wetness of the ground and the type of land cover present. Multiple versions of the model were tested on previous earthquake-triggered landslides and researchers selected the model with predictions that best matched where landslides occurred.

By entering data on ground shaking for a specific earthquake (available from the U.S. Geological Survey ShakeMap tool) scientists will be able to use the model to compile a map highlighting the probability of landslides in areas near the quake.

Saving Lives

Being able to predict forthcoming landslides well in advance will give agencies and authorities in locales that are likely to be hit time to prepare and evacuate citizens if needed, potentially saving a lot of lives.

Share on      
Next Story »

What Can we Solve for You?
landslide1.jpg
mudslide2.jpg