Seismic data

Latest plots

Raspberry Shake network, station RCB43_SHZ_AM_00. View live data. Browse the data archive.
Plots rendered at 2017/05/29 15:03:13 UTC.
Page loaded at 2017/05/29 15:03:30 UTC.


This section still a work in progress as of May 29, 2017.

Plots are updated every minute at 15 seconds past. The station is located in a garage about 100 feet from a residential street with some pothole-like features in it. Vehicles hitting these features at speed account for a large portion of the spikes seen in the data. Other potential sources of noise include cars entering/exiting driveway, foot traffic, MetroNorth rail line (1000ft), Interstate 95 (1200ft), and waves on the shore of Long Island Sound (2000+ft). During the warmer months, lawnmower noise may contribute as well. The anthropogenic noise sources generally fall in the 10-15Hz range of the frequency spectrum as observed at this station. Water waves on shoreline (referred to in seismology as microseismic peaks or microseisms) are ever so slight in the seismic record. As such they tend to be best observed at times of relative acoustic quiet, such as the overnight hours local time, when the wind is low. Dominant period microseisms observed by this station tend to be in the ~0.5-0.25Hz (2-4 sec) range during periods of calm in LI Sound (here's a wave report from a nearby buoy). Earthquakes, should they be detected, will likely cover a much larger portion of the observable spectrum than the rest of the noise sources. More info below.

Spectrogram of the past five minutes (unfiltered):

A plot of energy density over time. Spectrogram color units are dimensionless.
Spectrogram plot


0.7 - 2 Hz bandpass:

Spectrogram plot


Helicorder plot since UTC 0000hrs:

Helicorder plot


0.7 - 2 Hz bandpass helicorder plot (can detect P waves from large, faraway earthquakes):

Helicorder plot of 0.7-2 Hz band


More info

These plots are made with a combination of ObsPy and Matplotlib formatting. The spectrogram above scales based on the density of energy at the given frequency, based on a moving window with an overlap of 0.9/1. The helicorder plot resets every day at 0000hrs UTC. Traces plotted on the helicorder are scaled down by 996 (3e4 in hexadecimal terms) for readability, although this is subject to change as I get more familiar with the signatures of environmental noise that the geophone picks up.

As it turns out, there are plenty of sources of environmental noise associated with placing a geophone near "downtown" Greenwich, CT. My hypothesis is that these sources of noise are somewhat amplified by unconsolidated holocene fluvial material that underlies the property. I do not know the depth to acoustically "fast" unweathered bedrock, but based on characteristics of nearby outcrops (and observed belowground conditions of a recent road excavation) I assume it's probably at least 3-5 meters below where the station sits.

Below you can see examples of signatures of the station's most easily discernable events. As discussed above, by far the most plentiful and dramatic noise sources are passing cars hitting subsidence areas and joints in the road. Generally speaking, at this station, passing vehicles produce vertical vibration in the 10-15 Hz range. Spectra of larger vehicles such as dump trucks are broader and traces are more diffuse.

There are two distinct noise sources in the street closest to the station, one a filled-in water line replacement trench, and two, a sewer manhole cover depression. Often, car noise recorded by the station is distinct from other sources of noise because it features two separate peaks, one slightly larger than the other, presumably from vehicle tires traversing each of these features individually. The larger peak is often associated with traversal of the trench (the further south road feature) and thus the travel direction of vehicles can sometimes be discerned by which peak occurs first. (I know what you're thinking and the answer is yes, I have sat on my porch watching the seismic signals of passing traffic to try to figure these things out.)

  • Car traveling north
  • Amtrak Train on MetroNorth line
  • An acoustically quiet period with a particularly good set of water waves
  • Person (5ft away) standing up quickly from tying shoelace, then walking away and closing garage door

  • If you'd like to learn more about ambient and anthropogenic ground motion, perhaps the best summary I've found is one from the German national particle accelerator research center DESY.


    Large Global Earthquakes


  • Low frequency waves from the M 6.1 earthquake in the western Aleutian Islands on March 27, 2017 felt here in Greenwich
  • Low frequency waves from the M 6.6 earthquake in the Kamchatka Peninsula area on March 29, 2017 felt here in Greenwich
  • Low frequency waves from the M 6.2 earthquake in the Atacama Desert area on April 15, 2017 felt here in Greenwich

  • P (compressional) and S (shear) waves from significant (>M6.0) earthquakes can traverse the diameter of the globe. Although S-waves cannot pass through the Earth's outer core (the outer core, being liquid, effectively does not transmit shear), S-waves can traverse the globe by reflecting one or more times off of the crust-aesthenosphere boundary, or by diffracting around the outer core-aesthenosphere boundary. (After reflection, P becomes PP and S becomes SS.) In my opinion—and in the opinion of Gempa, the organization that processes all of the RaspberryShake data worldwide—P-waves from these earthquakes are best observed in the otherwise-relatively-quiet 0.7-2.0 Hz range. When S-waves are present, they tend to be best observed in even lower frequencies, generally from 0.05 to 0.7 Hz.


    Local Earthquakes


  • M 1.3 1km N of Morris Plains, NJ, S-waves from which may have been very lightly felt in Greenwich around 17:45:11 UTC on March 25, 2017 (1:45:11 pm local time, or about 21.5 seconds after the event originated).

  • No significant local earthquakes have been detected at this station yet. As mentioned above, a very slight quake in Northern New Jersey may have registered on the device, but it's very hard to tell. A relatively local earthquake should occupy a much larger portion of the spectrum than most anthropogenic noise. Earthquake waves will also be much more uniform in their spectral power distribution. Where for example vehicles have a very dense spectral peak at about 12 Hz, an initial earthquake S-wave will more or less evenly occupy most of the spectrum from 0-20 Hz. Larger earthquakes sometimes tail off to primarily lower frequencies as time increases. On a spectrogram, this will have a bit of a "play button"-type appearance, if the play button were cut in half horizontally from the rightmost point to a point halfway up the vertical axis. The following images are a screenshot of the USGS software called Swarm of a 3.1 earthquake off the coast of the California at the Pacific-North American plate boundary as recorded by a RaspberryShake—with callsign RA6B2—near the town of Mendocino, CA.



    And the same waveform processed with ObsPy instead:


    There are differences between the softwares and the color scales, but overall the spectrograms show a fairly even onset of a broad band of frequencies from 1 to more than 25 Hz as the S-wave reaches the station. Light "ringing" continues observably for almost 30 seconds. I chose this light quake as an example since on the East Coast, this relatively small event most closely emulates the type of signagure this station is most likely to detect after an earthquake. Next time I'm able to see a larger or closer event at my favorite station to watch (Mauna Loa/MLOD), I will post that for comparison.