Development of a Regional Landslide Forecast Model for Slowly Moving Landslides in the Western Canada Sedimentary Basin

The Great Earth Squeeze: Why the Ground Moves and How We Can Predict It

Imagine the ground beneath our feet isn't as solid as a rock, but more like a giant, slow-motion sponge. In some parts of the world, specifically under the Western Canada Sedimentary Basin, this "sponge" is made of layers of ancient clay and soft rock that have been squished by glaciers for thousands of years. Over time, these layers have started to slide downhill, very slowly, like a glacier made of mud. These aren't the scary, instant landslides you see in movies that crash down in seconds; these are "Deep-Seated Landslides" (DSLs). They move so slowly you might not notice them for years, but they are relentless.

Why should we care? Because this slow-motion squeeze is happening right where our modern world lives. Pipelines, roads, and power lines are built right across these sliding slopes. While the engineers who build these things know the ground might move a little, there's a limit. If the ground moves too much, or moves too fast, the pipes can crack and the roads can buckle, costing millions of dollars to fix. The big question scientists have been asking is: Can we predict when these slow-moving giants will speed up? If we know the ground is about to get "wetter" and heavier, we can fix the pipes before they break, saving money and keeping people safe. This paper is all about building a crystal ball for the ground, using rain and snow data to guess when the earth will start sliding faster.

The Earth's Weather Forecast: A New Way to Watch the Slides

The team behind this study, led by researchers from BGC Engineering and the University of Alberta, decided to stop guessing and start measuring the relationship between the weather and the sliding ground. They realized that just like a sponge gets heavier and slippery when it soaks up water, these deep landslides speed up when the soil gets wet. But instead of just looking at rain falling today, they needed to understand the "memory" of the soil—how wet it was last year, the year before, and how the snow melted in the spring.

To do this, they gathered a massive amount of data. They looked at over 30 years of movement records from more than 500 sensors (called slope inclinometers) buried deep inside these landslides across Western Canada. These sensors told them exactly how fast the ground was moving every year. Then, they paired this movement data with a super-computer model of the Earth's weather called ERA-5. This model acts like a giant, global diary of soil moisture, snow melt, and rain going back to 1950. By comparing the "sponge's" movement with its "water diary," the team tried to find the secret recipe that makes the slides speed up.

The Recipe for a Moving Mountain

The researchers found that three main ingredients in the weather recipe were the most important for predicting when the landslides would get active:

1. How wet the soil was at the start of the season: Specifically, how much water was deep in the ground (about 1 to 3 meters down) by February.
2. The trend of the previous year: Was the soil getting wetter or drier over the last 12 months?
3. The "Total Water" hit during the season: This is the big one. It's the total amount of water (from rain and melting snow) that hits the ground over a rolling 90-day window during the landslide season (March to October).

Using these three ingredients, the team built a new tool called the Regional Landslide Activity Index (RLAI). Think of this index like a "traffic light" for the ground, but instead of just Red, Yellow, and Green, it has four levels: Low, Typical, High, and Very High Activity.

Here is how it works: The computer takes the weather data and calculates a score called the Regional Landslide Activity Score (RLAS).

• If the score is low (under 0.95), it's a Low Activity Year. Most landslides will be sleeping or moving very slowly, like a sleepy turtle.
• If the score is between 0.95 and 1.40, it's a Typical Activity Year. Some slides might speed up, but most are doing their usual slow dance.
• If the score is between 1.40 and 1.80, it's a High Activity Year. This is when the ground gets restless. The model suggests that about 71% of the landslides in the region will start moving faster than usual.
• If the score is above 1.80, it's a Very High Activity Year. This is the "Red Alert" zone. The model predicts that almost all the landslides will be moving much faster than normal, like a crowd of people suddenly running.

Testing the Crystal Ball

The team didn't just make up these rules; they tested them against history. They looked back at specific years when landslides caused real problems, like the big acceleration events in 2017, 2020, and 2021. They ran their model backward (a process called "hindcasting") to see if it would have predicted those trouble spots.

The results were promising. For example, in 1997 and 1998, the model correctly predicted a "Very High Activity" year before the ground actually started moving fast. In 2020, the model warned that a huge portion of landslides would speed up, which is exactly what happened. The authors note that while the model isn't perfect—it sometimes predicted a "High" year that turned out to be "Typical"—it rarely missed a major event. It's much better at catching the big risks than missing them.

Looking Ahead: The 2026 Forecast

The paper concludes by showing off the first real-world test of their new system: a forecast for the 2026 landslide season. Since they can't know exactly how much rain will fall in the future, they used a statistical trick. They ran thousands of computer simulations (called Monte Carlo simulations) based on 75 years of weather history to guess the range of possible outcomes.

The result is a set of maps that show the probability of the ground moving fast. In March 2026, they released a map showing the "95th percentile" prediction—basically, the worst-case scenario that is still likely to happen. As the season goes on and real rain data comes in, the uncertainty shrinks, and the forecast gets sharper. By the end of the season, they will know exactly how active the year was.

What's Next?

The authors are excited, but they know this is just the beginning. Right now, the model relies on data from about 641 specific spots. But they are planning to make it even better. They are adding data from new sensors that measure movement every few hours, and they are waiting for a new satellite (NISAR) to launch in 2025. This satellite will be able to "see" the ground moving from space, even in places where there are no sensors on the ground.

The paper makes it clear that this isn't a magic wand that will stop landslides. The ground will keep moving. But by giving pipeline and road owners a better way to predict when the ground will get restless, they can fix things before they break. It's about turning a surprise disaster into a planned maintenance job, saving millions of dollars and keeping the infrastructure safe. The authors suggest that with more data and better models, they might eventually be able to predict exactly where a specific slide will speed up, not just that the whole region is getting wet. For now, though, this new "weather forecast for the ground" is a huge step forward in understanding the slow, sneaky giants beneath our feet.