On 10 August 2020, the Grizzly Creek Fire started to sweep through the steep, wooded slopes of Glenwood Canyon in Colorado. Sparks from a vehicle traveling along Interstate 70 are believed to be the cause of the raging forest fire.
The fire blazed into October, shutting down the major interstate through the Rocky Mountains for 13 days. The flames left scarred, barren hillsides in their wake and a deadly calling card for the months ahead: landslides.
Months later, with no trees or vegetation to stabilize the slopes, rocks started falling down them and onto the interstate. “The tree roots hold rocks in place a bit—they are a little like a web,” said Paul Santi, a professor of geology and geological engineering at the Colorado School of Mines in Golden. “You remove those and the stability they are providing, and you are going to see the movement of particles, from sand-grain size to boulder size, after a wildfire.”
Burned hillslopes are especially vulnerable when rain begins to fall. In Glenwood Canyon and many other burned areas, the main concern is giant debris flows, which are a mix of soil, rock, and other debris saturated with water. After the 2018 wildfires near Montecito, Calif., a series of debris flows killed 23 people, injured 167 more, and damaged 408 homes. In Glenwood Canyon, mudslides could close a major transportation corridor for days, or more. And the potential for such risks might stretch for years, depending on how long it takes for vegetation to regrow and restore slope stability. “Typically, the increased risk for debris flows is usually something on the order of 1–3 years after a fire,” Santi said.
A Climate Connection
Colorado has seen temperatures rise by about 1°C since the early 20th century, and the hotter temperatures have been associated with droughts and extreme wildfire seasons. Because much of the state is mountainous, this change in climate could also lead to more landslides.
“As we have more fires—and longer, more intense fires—and then we couple this with more intense weather events, we are going to see a higher frequency of debris flows,” Santi said.
Scientists are also starting to discover other ways in which climate change is leading to changes in certain types of landslide activity, including debris flows, earthflows (rapidly descending earth), and rockfalls. In certain mountainous areas of Europe and Alaska, researchers have recorded a higher incidence of rockfalls as average temperatures rise from one year to the next. In other locations, an increase in extreme rainfall events and/or stronger hurricanes—both linked to climate change—can trigger debris flows.
“The connection between climate change and landslides is something that’s been talked about for a long, long period of time,” said David Petley, an Earth scientist at the University of Sheffield in the United Kingdom. “There is a kind of inevitability that there will be a signal there.”
Consequently, in the past 2 decades, scientists have begun to probe how different aspects of climate change are affecting the stability of slopes in North America, Asia, Europe, and Oceania. The results could help people know whether the ground beneath their feet, or the slopes above their houses, might pose a threat in the decades to come.
From Rock Climbs to Rockfalls
In 2005, a series of huge rockfalls in the Mont Blanc massif in the French Alps swept away the Bonatti pillar, a classic climbing route on the Drus, at 2,754 meters above sea level. For scientists Philip Deline and Ludovic Ravanel at the Université Savoie Mont Blanc/French National Center for Scientific Research in Chambéry, the rockfalls were troubling but unsurprising. For years, they’d been tracking rockfalls and the rising risks for the climbing community in the region.
Using reports from mountaineers and a large database of old and new photographs, Ravanel has created an inventory of rockfalls in two areas of the Mont Blanc massif from the end of the 1800s to current times. Nearly 80% of the rockfalls occurred between 1990 and 2015, as air temperature increased. Nearly a third of all the rockfalls have occurred in the past 5 years, Deline said. “We have shown there is a clear connection between the warming temperature and rockfall frequency,” Deline said.
The scientists believe that warming temperature is heating permafrost, ground that has remained permanently frozen for at least 2 years. In mountainous areas, permafrost zones typically exist above a certain altitude. In permafrost zones, rocks are glued together by ice fillings in their cracks and crevices. As the air temperature increases each year, scientists believe that the warming, and even thawing, permafrost is weakening rock faces and leading to rockfalls. The theory is difficult to prove: “The permafrost is not something visible like a glacier,” Deline said.
When the scientists analyzed their inventory of rockfalls, they found that all of the rockfalls stemmed from areas in permafrost zones and that rockfalls were more likely in regions of temperate permafrost (close to 0°C). They also found that the frequency of rockfalls increased during the hottest summers, with significant rockfalls seen in the extremely hot summers of 2003 and 2015.
In a relatively recent rockfall from their inventory, approximately 50,000 cubic meters of rock crashed down a mountainside. Some of the large sections of rock contained ice. Deline and Ravanel used carbon dating to estimate the age of the ice and found it was several thousand years old. This shows that the permafrost hadn’t thawed in that time, Ravanel said.
“So this is one more element in order to show that we are in [a] period of the degradation of the permafrost,” Deline said. “We are becoming even more convinced that thawing permafrost is the trigger of this kind of event.”
From Rockfalls to Tsunamis
The connection between warming temperatures and rockfalls has been observed at high altitudes in other regions as well. When scientists used Landsat satellite imagery to analyze rockfalls in Glacier Bay National Park and Preserve in Alaska, they found that a cluster of large rock avalanches had occurred between 2012 and 2016, a period of record-breaking warm winter and spring temperatures.
Researchers hypothesized that thawing permafrost may be playing a role in these events and expressed concerns that a landslide into Glacier Bay could trigger a tsunami. “It would be a low-probability, high-consequence type of event,” said geologist Gabriel Wolken of the Geological and Geophysical Surveys Division of the Alaska Department of Natural Resources.
More recently, however, scientists have become concerned about a “higher-probability event with very large consequences,” Wolken said. As the climate has warmed in Alaska, Barry Glacier, on the edge of Prince William Sound, has retreated. Previously supported slopes are now weakened and have the potential to fail.
A small landslide from this slope into Barry Arm Fjord has the potential to generate a tsunami that could affect fishers and tourists in the immediate area. A full-slope failure would be like dropping about 500 Empire State Buildings into the ocean at one time, Wolken said, on the basis of the scientists’ current best estimates of the possible maximum volume.
A tsunami from such a major landslide would be devastating for the local community. The town of Whittier, 50 kilometers southwest of the slope, could be struck by a wave 10 meters high within 20 minutes. During the summer, as many as 500 people—fishers, recreational boaters, and campers—may be in the area and at risk, according to the Alaska Department of Fish and Game. Consequently, Alaskan officials have strongly recommended that people avoid the identified danger zones until scientists better understand the hazards involved.
Concerned about this risk, a group of scientists, including Wolken, issued a statement in May 2020 calling attention to the large, creeping rock slope above the fjord and the potential for a hazardous tsunami. For Wolken, the connection to climate change is clear.
“Fundamentally, this landslide probably wouldn’t happen, and the tsunami hazard associated with it wouldn’t exist, if the glacier [were] positioned where it used to be,” Wolken said. “We have all of these changes in cryospheric variables…such as glacial retreat and permafrost thaw and degradation, that are independently clearly linked to a change in climate.”
Rainy Day Slides
Although the connection between rising temperatures and landslides in permafrost zones is becoming clearer, most of the world’s landslides are triggered by rainfall. Discovering whether changes in rainfall patterns could lead to more, or fewer, landslides is proving tricky.
“The nature of the interaction between rainfall and ground movement is harder to pin down, as there are lots of variables to be considered simultaneously,” Petley said.
To determine how changing rainfall patterns might affect landslide activity in the Calabria region of southern Italy, for example, scientists analyzed a catalog of rainfall events related to landslide occurrences in the region between 1921 and 2010. The researchers then used the rainfall events from the past 30 years and a model of anticipated changes in rainfall patterns based on different greenhouse gas emissions scenarios to anticipate how landslide activity might change across Calabria in the future.
Using this approach and assuming that emissions keep rising through the 21st century and temperatures rise by 4.5°C in the next 80 years, the researchers’ methodology predicted a 45.7% average increase of rainfall-induced landslides in the region. “We expected an increasing trend, and we managed to quantify it,” said Stefano Luigi Gariano at the Research Institute for Geo-hydrological Protection in Perugia, Italy.
The scientists’ research—among the first to quantify the impacts of changing rainfall patterns on a regional scale—also predicted that there would be a greater number of rainfall events capable of triggering landslides. And those high-intensity rainfall events would be concentrated into certain months of the year. “There will be more rainfall in less time, which will produce more intense events,” Gariano said.
Warming temperatures can also have an impact on the intensity of hurricanes. “We’re seeing from our data that the strongest storms are in fact getting stronger,” said climatologist James B. Elsner at Florida State University.
Although more intense storms don’t necessarily cause more rainfall, scientists have determined that long-term climate trends contributed to the record-breaking rainfall of Hurricane Maria, which slammed Puerto Rico in 2017. This extreme rainfall led to 40,000 landslides on the island.
A Model Approach
To look at how changing rainfall patterns might be affecting landslide activity in High Mountain Asia, a team of scientists recently used two models: one that projects changes in precipitation coupled with another that determines whether landslide activity is likely due to the intensity of the rainfall and the surface conditions.
“It’s a simple decision tree model that considers if it is raining hard,” said research scientist Dalia Kirschbaum of the NASA Goddard Space Flight Center in Greenbelt, Md. “If the answer is yes, it goes on to see if the area is susceptible to a landslide.”
The scientists used outputs from their Global Climate Change model to anticipate changing rainfall patterns between 2060 and 2100 in High Mountain Asia to produce a representation of how rainfall might vary seasonally in the future.
The researchers then fed the rainfall data into their Landslide Hazard Assessment for Situational Awareness model, which assesses the potential for landslides in a region on the basis of detailed information about slope steepness, bedrock, tectonic faults, and tree cover within a given area.
The combination of models predicted that the border region of China and Nepal could see a 30%–70% increase in possible landslide activity at the end of the century compared with that seen between 1961 and 2000.
Currently, the scientists’ model is focused only on the rainfall impacts. “There are many studies that have shown that rainfall is the most predominant trigger of landslides,” Kirschbaum said. “And in a changing climate, we anticipate that landslide impacts could be even more exacerbated in areas where you have more extreme rainfall and susceptible slopes.”
A Tangled Web
Despite the progress that has been made in assessing how climate change affects landslides, the picture remains a complex one. Climate change can lead to more landslides, but in some regions, it can lead to droughts or more vegetation, which can make landslides less likely.
“The mechanics of landslides are actually pretty straightforward, and we’ve understood them for a long time,” Petley said. But he added that deciphering when a landslide will happen, or how, is tricky. Often, landslides are simply the end point, or middle point, in a cascade of events, muddying the picture even further. Wildfires followed by rain can cause landslides, earthquakes can cause landslides, and even one landslide can lead to another.
“The actual way that a landslide manifests is really quite complex and unfortunately keeps us in business,” Petley said. “Prediction methods might be able to predict the first landslide, but can they predict immediately what happens after the first one?”
A Paucity of Data
In Alaska, many of the areas prone to landslides are remote and unpopulated, and their conditions are harsh. Consequently, scientists have faced challenges in seeing how changes in climate affect the potential for landslides. “It’s logistically challenging and costly to study the changes impacting slope stability in such a vast area,” Wolken said. “So we have a data gap.”
Data scarcity also exists in the long-term record of landslide activity. “If you talk to climate scientists, they say you can’t pick up a trend in a data set that is less than 30 years,” Petley said. “And the problem is that pretty much nowhere in the world do we have a 30-year record of landslide activity that has enough landslides in it so that we would genuinely know that change is occurring.”
Consequently, the lack of data is the key challenge facing scientists in trying to determine what impact changing weather patterns might be having on landslide activity. “Without more information, we can’t draw that line directly yet,” Wolken said.
Closing the Gap
Although current landslide inventories might lack the number of landslide incidents required to draw any meaningful conclusions about changing patterns, scientists are investigating both practical and inventive approaches to close the data gap.
Cell phone videos and high-resolution remote sensing provide much greater documentation of modern landslides than was possible in the past. Petley takes advantage of these in his Landslide Blog, which highlights significant landslide events from around the globe and provides commentary on their causes.
In 2013, geophysicists at Columbia University pioneered a new method to detect the unique signature of landslides in seismic waves, the vibrations caused by sudden movements of rock, ice, magma, or debris. Using this technique, scientists were able to find a series of seven previously undocumented massive landslides associated with Siachen Glacier in the Himalayas.
Such a technique could be used to search back through the historical seismic data and, with the help of artificial intelligence, pick out previously undetected landslide events, Petley said. “It does seem that although we don’t have comprehensive data sets yet, there is the prospect that through possible new techniques we may be able to project backward, which is really exciting,” he said.
Kirschbaum believes that the combination of high-resolution satellite imagery and machine learning algorithms will drastically enhance opportunities to increase the amount of landslide data available. “We’re really at the cusp of greatly expanding landslide inventories as a community,” Kirschbaum said. “With better knowledge of events today, we can build models that help take us into the past and understand the patterns between landslide activity and climate.”
In July 2020, Kirschbaum and her colleagues published a study demonstrating how they had used a model to reconstruct the patterns of landslide hazard in the Pacific Northwest using a machine learning model. The model was able to replicate well-documented landslide events and then apply what was learned to represent long-term patterns in potential landslide hazards. “While this model was over a fairly small study area, it shows the potential for doing this on a much larger scale,” Kirschbaum said.
Kirschbaum is confident that researchers are going to be able to build models that can look back into the past to understand patterns in landslide activity and that can also project into the future to anticipate how climate change may modulate landslide activity and impacts. “It’s a complex problem, but I think with new modeling techniques and new data sources, there is real potential to move our understanding of landslide processes and impacts forward,” Kirschbaum said. “So it is a positive picture moving forward.”
—Jane Palmer (@JanePalmerComms), Science Writer