,
23 tweets,
6 min read
Read on Twitter
Today, #SeismogramSaturday is actually #SpectrogramSaturday. I'm going to talk about another way that seismologists might look at seismic data.
The key background information here is that earthquake waves, like all waves, have a quality called "frequency", which describes how many times the wave oscillates per second (a unit called Hertz, or Hz).
Sound waves are a good way to explain frequency--if the air oscillates many times per second it creates a high frequency wave, which is a high note. Low frequency waves are low notes. Humans can hear in the range of about 20-20,000 Hz. Middle C is 256 Hz.
Earthquakes are typically much lower frequency, with 50 Hz or so on the very high end, and frequencies that go down so low that they oscillate only once in minutes or hours. Nearly all seismic waves are below the range of human hearing.
Any given wiggle, however complex it looks, is actually just a bunch of frequencies playing at the same time. We can break the signal down into its frequencies using something called a Fournier transform.
It's not that different from a symphony--there are high frequencies (piccolos, for example) and low frequencies (the bassoon or tympani). They all play at the same time so the resulting waveform looks complex, but it represents something beautiful.
We learn a lot from earthquake frequencies. In some cases these can provide information about how the quake failed, and what caused that failure (this is at least very common in volcano seismology, which is my field of expertise).
Sometimes it's easiest to look at these frequencies using a "spectrogram", which shows both the seismogram (the wiggle plot) and the frequencies. An example is shown below. 
In this figure, you see the seismogram on the top and the spectrogram on the bottom. Both show the same data, but presented differently. The seismogram shows how strongly the ground was shaking at this station.
The vertical axis in the spectrogram is frequency--high notes near the top, low notes near the bottom. Color is volume. So when the quake hits at ~260 seconds into the file, we get strong (pink) energy at frequencies between 0-25 Hz.
This quake was a M2.5 event in California today--just a little guy in San Juan Bautista. No one reported feeling it. earthquake.usgs.gov/earthquakes/ev…
What's cool here is that in the spectrogram you can also see two quieter (yellow-blue) signals near the end of the file.
These aren't even visible in the seismogram because they're so much smaller than the M2.5, and they don't rise much about the noise. But their frequencies are different from the low frequency background noise, so we see them in the spectrogram.
There's also a hum at ~24 Hz in this spectrogram--you might not even pay attention to it. Something that perfect (its frequency doesn't vary) is probably electronic noise rather than an earth signal.
Larger earthquakes generally create lower frequency waves, and high frequencies attenuate (lose energy) pretty quickly. So large earthquakes recorded far from the source are low frequency, like this one:
You can see this quake (the M7.5 that hit Ecuador last month) more clearly in the seismogram, but it is there at the very bottom of the spectrogram (it would be clearer if the frequencies were presented on a log scale).
Some signals have really cool spectra. As an example, here's an event called a "tornillo" that was recorded yesterday in the Katmai region of the Alaska peninsula. Tornillos shake at one single frequency--they are like sine waves.
This tornillo hummed at ~2.5 Hz only. These are most often seen in volcanic regions and are probably related to hydrothermal or magmatic fluids.
Just for fun, here's a crazy one that I recorded in my garage a few years ago. In this case it's not a natural event--this is the way the ground shakes when a train starts up at the station ~0.5 miles from my house.
In that train signal you can see harmonic frequencies that started off low, then got higher, then slowly lowered again. Trains make really complex but kinda lovely spectograms. I know this because they shake my house like crazy and I'm just geeky enough to record them.
Spectrograms are a great tool for distinguishing different types of quakes, different types of signals, etc. It's difficult to identify volcanic tremor in a seismogram, but pretty clear in a spectrogram.
Many public agencies allow you to view their data in either seismogram or spectrogram formats. For example, here's where you can see real-time spectrograms for seismic stations in the Pacific NW. pnsn.org/spectrograms
Got questions about earthquake frequencies and spectrograms? Post 'em here. Happy #SeismogramSaturday and #SpectrogramSaturday, friends.
