Any views expressed in this article are those of the author and not of Thomson Reuters Foundation.
Satellite data could not only identify hotspots for landslides, but could also play a part in predicting these potentially devastating events, particularly in remote mountainous regions, a study suggests.
Researchers at the University of California, United States, have developed a model using satellite data on rainfall, topographical features of slopes, and land cover — and by testing the model on a dataset of previous landslides say it predicts these historical events reliably and could be the basis of a real-time, global landslide prediction system.
"Landslides typically occur in mountainous regions where other sources of information, including radar and gauge measurements [used in standard global landslide models], are not available," Amir AghaKouchak, co-author and assistant professor at the Center for Hydrometeorology and Remote Sensing in Irvine, tells SciDev.Net.
"Further, in many developing countries ground-based observations are also limited due to a lack of investment.
"Our model has been developed with satellite data so that it can be used[globally] in remote and topographically complex regions. Most previous landslide studies have been at a local or regional scale," AghaKouchak adds.
“Efforts are underway to further develop the model into a real-time landslide prediction model.”
According to the researchers, the model "cannot be considered as a general landslide model" as it does not take earthquake-triggered landslides into account, and is not designed for small-scale landslides (local events not reported in the NASA global landslide inventory, which is the data used to calibrate the model).
But it can be "coupled with a local physical model to improve landslide monitoring prediction" by first using the satellite model to identify landslide hotspots and then applying a physical model for slope failure to the hotspots.
"Efforts are underway to further develop the model into a real-time landslide prediction model," AghaKouchak says. "But our research will largely depend on receiving grants and support."
Bjørn Nilsen, professor at the department of geology and mineral resources engineering, Norwegian University of Science and Technology, says: "This proposed approach for predicting landslide risk is definitely interesting.
"For monitoring larger, remote areas it may be even more valuable. But dense vegetation may for many areas represent a limitation and add uncertainty to the method. For further investigating the potential and reliability of the proposed methodology, site visits and investigations are recommended."
Oliver Krol, a scientist at the Fraunhofer Institute of Optronics, System Technologies and Image Exploitation in Germany, says: "This is a very ambitious project to develop on a global level".
But he adds that a focus on regional models would be preferable since local conditions could be covered by individual countries much better.
The research was published in Natural Hazards and Earth System Sciences.
Link to full paper