Combined Inversion of Active and Passive Seismic Data Based on Optimized Data Selection

Imagine trying to figure out what's inside a giant, dark cake without cutting it open. You can't see the layers, the fruit, or the frosting, but you can tap on the outside and listen to the sound it makes. In the world of geophysics, scientists do something similar to understand the Earth's crust. They send sound waves (seismic waves) through the ground and time how long it takes for them to bounce back or travel through. Since sound travels at different speeds through different materials—like how it moves faster through a hard rock than through a soft clay—the travel times act like a map, revealing the hidden structure of the planet.

Usually, scientists get this data in two ways. The first is "active" seismic data, which is like a controlled drumbeat: they set off a known explosion or vibration at a specific time and place to see how the waves travel. The second is "passive" seismic data, which is like listening to the Earth's own whispers: they wait for natural earthquakes or tiny tremors to happen. The problem is that these natural tremors are unpredictable; they happen in messy clusters, often leaving big gaps in the map. If you try to combine the clean, organized drumbeats with the chaotic whispers, you might end up with a blurry picture where the crowded areas look too loud and the quiet areas get ignored. This paper tackles that exact puzzle: how to mix these two very different types of data to get the clearest possible 3D map of what's underground, without letting the messy parts ruin the picture.

---

The Great Underground Map-Mix-Up

Imagine you are trying to draw a map of a giant, invisible cave system. You have two tools to help you. The first tool is a team of explorers with flashlights (the active data). They stand in a perfect grid, shining their lights in every direction. Because they are organized, their light covers the center of the cave beautifully, but they can't reach the far, dark corners. The second tool is a swarm of fireflies (the passive data). These fireflies light up randomly whenever they feel a tiny vibration. They are great at lighting up the dark corners the explorers missed, but they are chaotic. They tend to swarm in huge, dense clouds in the middle of the cave, leaving the edges in the dark, and sometimes they blink so fast in one spot that it looks like there's a giant fire there when it's just a bunch of bugs.

The researchers at ETH Zurich, led by Kathrin Behnen, wanted to combine the explorers' perfect grid with the fireflies' wild coverage to make the best map possible. But they knew that if they just threw all the fireflies into the mix, the map would get distorted. The dense swarms in the middle would make the computer think that area was super important, while the edges would get ignored. It's like trying to take a group photo where 90% of the people are standing in the center; the edges of the photo would look empty and weird.

The Solution: The "Smart Filter"
To fix this, the team invented a clever "smart filter" using a mathematical trick called QR factorization. Think of this as a super-smart bouncer at a club. The bouncer looks at the chaotic crowd of fireflies (the passive data) and asks, "Who here is actually adding something new to the picture?"

Instead of letting every single firefly in, the bouncer picks just the right ones. It selects fireflies that are standing in the dark corners (to fill the gaps) and ignores the ones that are just standing on top of each other in the middle (which would be redundant). The goal was to find the "Goldilocks" number of fireflies: enough to light up the whole cave, but not so many that they crowd the center and block the view.

The Experiment: The BedrettoLab
They tested this idea in a real-life underground laboratory called the BedrettoLab in the Swiss Alps. This lab is a tunnel dug deep into a giant granite rock formation. The scientists had a perfect setup:

• The Explorers: They used a "sparker" (a controlled sound source) and hydrophones (underwater microphones) to send 41,881 sound waves through the rock.
• The Fireflies: They listened to 30,521 tiny, natural earthquakes (induced by water injection experiments) that happened inside the rock.

After cleaning up the data, they had a massive catalog of events. But they knew they couldn't use all of them. So, they applied their "smart filter."

The Results: A Clearer, Bigger Picture
The results were like magic.

1. Three Times the View: By using their optimized selection, the team managed to illuminate a volume of rock three times larger than what they could see with just the controlled explorers.
2. No More Crowding: When they tried using all the passive data (the whole swarm), the map got messy. The center of the rock looked too "bright" and distorted, creating fake little bumps and wiggles that weren't really there. It was like the fireflies were so dense they blinded the camera.
3. The Sweet Spot: The "smart filter" selected about 20,000 passive data points (out of over 250,000 available). This specific number was the perfect balance. It filled in the dark corners without overcrowding the center. The resulting map was smooth, accurate, and showed the true shape of the rock.

What They Found Underground
With this new, super-clear map, the scientists could finally see the secrets of the rock.

• The Fracture Network: They found that the rock was full of cracks and fractures. Where the rock was cracked, the sound waves traveled slower (like running through a field of mud). Where the rock was solid, the waves zipped through fast.
• The Earthquake Connection: The tiny earthquakes (the fireflies) mostly happened right inside these slow, cracked zones. This confirmed that the rock breaks where it is already weak and full of cracks.
• The "Weird" Direction: The scientists also noticed something strange. The sound waves didn't just travel at different speeds; they seemed to travel faster in one direction and slower in another, like a road that is smooth going north but bumpy going east. This is called anisotropy. The data suggested that the cracks were all lined up in parallel, like a stack of pancakes, causing the sound to behave differently depending on which way it traveled.

What They Ruled Out
The paper is very clear about what doesn't work.

• Random Selection Doesn't Work: If you just pick fireflies at random (like a lottery), you end up with too many in the center and not enough in the corners. The map ends up with the same distortions as using all the data.
• Using All Data is Bad: Throwing every single data point into the mix doesn't give you a better map; it actually ruins it by overemphasizing the crowded areas and creating fake, small-scale artifacts.
• Passive Data Alone Isn't Enough: You can't just use the natural earthquakes to make a map. Without the controlled "explorers" to anchor the picture, the map gets blurry because the computer gets confused about where the earthquakes actually happened versus where the rock is weird.

The Bottom Line
This paper suggests that to get the best map of the underground, you don't need more data; you need smarter data. By using a mathematical filter to pick the most useful, non-redundant pieces of information, scientists can see deeper and clearer into the Earth. They proved that about 10% of the available passive data was enough to create a model that was three times better than the active-only version, without the messy distortions. It's a reminder that sometimes, in science, less is actually more—if you pick the right "less."