A landslide in Rio de Janeiro, Brazil, tore toward Niemeyer Avenue in February and April, damaging a roadside bicycle lane that runs between Rio’s Leblon and São Conrado neighborhoods.

Richard Santos/Prefeitura do Rio

Rainy, with a chance of rockfalls: New landslide forecasting system debuts in Rio

With its steep slopes and rainy weather, Rio de Janeiro, Brazil, is no stranger to landslides. In February and April, landslides tore down a slope by the coast, leading to the collapse of a local cycling lane that runs along a major road. NASA Goddard Space Flight Center research scientist Dalia Kirschbaum hopes to lessen the toll of future events by forecasting them—and sounding the alarm through social media. She and her team in Greenbelt, Maryland, launched their forecast model for Rio, called the Landslide Hazard Assessment for Situational Awareness (LHASA), in October.

LHASA, which cost the city about $11,000 to install, creates what it calls “nowcasts.” Those are near–real-time forecasts of where landslides are most likely to happen in the next few minutes—based on past rainfall data and other variables, including hill slope angles, local rock and soil types, and whether an area has been deforested, as well as how close geologic structures such as faults are. Felipe Mandarino, a city information coordinator whose Rio-based data organization is helping customize LHASA for the city’s specific needs, hopes that with the model’s launch, landslides will become less and less of a surprise.

ScienceInsider spoke with Kirschbaum and Mandarino; this interview has been edited for brevity and clarity.

Q: How are LHASA’s “nowcasts” performing during heavy rain events in Rio?

Felipe Mandarino: We considered the model under a test phase during the last rainy season—2018 and 2019—and it performed quite well. We had landslides, and I was looking at the model in real time and saw that the model was mapping a hazard—and, in fact, landslides happened.

Dalia Kirschbaum: It’s really exciting, because this is a demonstration of what the capabilities can be of dynamically identifying landslide hazards in real time. This is a first at NASA.

Q: Is LHASA ready to help the city decide when to evacuate people?

F.M.: It’s ready to be used that way. The city [already] has evacuation plans and sirens in the parts of the city that have more risk to landslides. What this model gives us is a view of the whole city. We will from this rainy season onwards provide that kind of early-warning alert for the whole city.

Q: How does LHASA work with the siren system?

F.M.: One complements the other. The new LHASA model will be used for the parts of the city that do not have the sirens, which is certainly the biggest part of the city in terms of area—98% or something like that. The siren systems, they’re really for small areas, for the favelas. The slums of the cities. There is a specific protocol with a series of rain thresholds and decisions to be made regarding this system.

Q: What are some limitations of the model?

D.K.: Landslides are very difficult to estimate because there are so many different factors involved. We’re [trying] to figure out what information in terms of forecasted rainfall and data on past rainfall and soil moisture might help to better inform the forecasts. The other thing that we talked about was a better understanding of the human dimension to identify higher priority areas at the city scale. We’re working to add more information on population, on roads, on other infrastructure that may help to better characterize the exposure and maybe the risk associated with the landslide hazards.

Q: In a best-case scenario, how much advance warning will LHASA give about an impending landslide?

F.M.: It’s really hard to predict that, but I would say something like tens of minutes, probably. Not hours.

Q: What will it take to expand LHASA to landslide-prone places elsewhere in the world?

D.K.: There are no plans to have as detailed, or as long and involved a relationship as with Rio at this point. But there are already other implementations of the landslide model. One of my colleagues is working with IDEAM [the Institute of Hydrology, Meteorology and Environmental Studies], which is the meteorological service in Colombia, and they are working to implement a version of the model across the country using both rain gauges and satellite data, and the information they have on landslides. And then we’re also working with a group in Tajikistan and Pakistan as part of the same project, and we’re just starting something in Southeast Asia.