How do scientists turn those little wiggles on paper into images of Earth’s interior?
One powerful technique is called Seismic Tomography, which works much like a CT scan of the Earth. Just as doctors take X-rays from different angles to build a 3D view of your body, geophysicists use seismic waves from multiple directions to map what lies deep underground. This is how we know that the Earth has distinct internal layers with different compositions, as well as smaller features like subducting slabs, magma chambers, and sediment basins.
To conduct a seismic tomography study, you must start by designing the network of seismometers, based on what you want to see in the subsurface. Based on the location you’re interested in, you may use land stations, OBS, or a combination of both. You may be looking at features in the shallow crust, just 1-50 km below your feet, and will want to use a more dense array of seismometers in a small area, trying to capture earthquakes nearby your area of interest. Alternatively, you may be interested in the deeper structure, perhaps looking at the deep crust, mantle, or even the core boundaries deep in the earth! For this you’ll need to capture earthquakes from all around the world (called teleseismic), requiring a broad network that is more spaced out.
Once you have your array set, you must consider how long you will let it sit out for. This will depend on the battery life of instruments, access logistics, and how much seismic activity you expect. If you’re in a seismically active area, you may get plenty of useful data quickly. In quieter regions, you might leave your instruments out for months or years to accumulate enough earthquake events. But what if you are looking at the shallow structure and there are no earthquakes occurring nearby? Are there just gaps in the world we can’t image in detail?
In these situations, scientists will turn to other seismic sources. There are actually many other sources of these seismic waves, and they can be used to conduct these seismic tomography experiments too! We divide them into 2 groups:
With the array set and source determined, you can start collecting data! But wait, which channel of the seismometer should you look at? Well, the data you want to look at depends on what part of the Earth you want to see, and what properties you want to measure!
Different seismic waves types will appear more strongly on different channels, based on how the wave moves. So, it’s important to know what channel to look at to find the seismic wave you’re interested in. Additionally, these different waves are sensitive to different subsurface materials, so knowing what you want to image is important for choosing the type of wave you’re interested in looking at.
Body Waves
These waves travel through the interior of the Earth, they’re the first to arrive after an earthquake.
Surface Waves
These waves travel along the Earth’s surface, arriving after body waves but often carrying the most energy and damage potential, especially in large, shallow earthquakes
Each channel of a seismometer acts like a filter, tuned to specific directions of motion. Knowing which channel highlights which wave type ensures you’re reading the most useful signals for your imaging goals.
From Signals to Subsurface Images
Once the data has been recorded and the seismic waves have traveled through the Earth to your array, it’s time to turn those signals into an image of the subsurface. This step is where geophysics meets computation. Scientists use specialized coding packages and inversion algorithms to process the data. But first, you need a model to start with.
1. Building a Starting Model
Before the computer can simulate how waves move through the Earth, it needs a starting model of the region you’re studying. This model is essentially a map filled with estimates of seismic velocity, or how fast different types of seismic waves are expected to travel at each point underground. These velocities vary with rock type, temperature, pressure, and fluid content. The model is usually based on known geology, regional studies, and educated guesses. It’s a rough draft of what the Earth might look like in your area of interest.
2. Simulating Wave Travel Times
Next, you calculate how long it would take seismic waves to travel from the source to each station in this starting model. You calculate the expected travel time for each wave, based on the path and the current velocity values in the model.
3. Comparing to Real Data: Residuals
Then you compare those expected travel times to the actual arrival times measured in your data. The difference between the two is called the residual. If the wave arrived earlier than you calculated it should, the starting model might have a velocity that’s too slow in that area. If the wave was late, the model might be too fast. These mismatches guide the next step.
4. Inversion: Adjusting the Model
When you run the inversion algorithm, you adjust the seismic velocities in the model, trying to minimize the residual (observed arrival time – calculated arrival time). This is an iterative process, you will calculate the travel times, make small tweaks to the model, then calculate them again and see if the residuals have improved. With each pass, the model should get a little closer to matching the real Earth.
This process doesn’t produce a single “perfect” solution, there is always uncertainty due to noise in the data, limited coverage, and simplifications in the physics. But with good data coverage, the residuals shrink, and the model becomes a much more accurate image of the subsurface.
5. Using Multiple Events and Paths
One seismic wave gives you information along a single path, but when you use data from many seismic sources traveling through the area from different angles, you build up a network of intersecting paths. By combining all these paths, the model can resolve structures in 3D. The denser your array and the more diverse your wave paths, the sharper and more detailed your final image will be.
The end result is a velocity model, a 2D or 3D image showing patterns of seismic wave velocity in the Earth. These models can reveal subducted slabs, magma reservoirs, fault zones, sediment basins, and more. These models provide physical insight into the otherwise invisible world beneath our feet.