I’ve found much better ways of examining earthquake data
received by my station. Thanks to Alan Sheehan, who was kind enough
to post his excellent report generation tool, I’ve now got some
actual data we can use. I’ve updated his code a bit and changed it to
better match my workflow; my repo is here. Like Alan’s, my code
is under the MIT License. Share and enjoy!
Here’s a sample of what we’ve been able to see:

This is the result of a magnitude 6.2 earthquake that struck off
the coast of El Salvador. Fortunately, no damage, injuries or
fatalities were reported.
Here’s a sample of what I’m pretty sure is a freight train going by:

This happened about 15 minutes after I saw the train go by another
crossing about 10km west of where the seismometer lives. The
equidistant lines in the spectrum sure seem like the ones reported in
this study, “Equidistant Spectral Lines in Train Vibrations” by
Florian Fuchs, Götz Bokelmann, and the AlpArray Working Group
(https://doi.org/10.1785/0220170092 …but see the previous link for
the actual paper).
Hello, world! A couple weeks ago, I took delivery of a Raspberry
Shake 1D. It’s pretty sweet. Right now it’s set up at my inlaws'
house, and I wanted to see how it’s doing by looking at whether it can
detect recent larger earthquakes.
Macquarie Island Region: Mag 5.9, June 9 2023, 21:21:42 UTC
I’m starting with the USGS map of recent earthquakes:

From there I can drill down to individual earthquakes – such as that
first one, 5.9 in the Macquarie Islands. Going to Waveforms
takes me to this page, which lets me find stations that recorded
data from it. Looks like there’s one in Corvallis, Oregon:

Clicking on that gets me the data:

Side note: this is narrowed down to VH channels. V means a sample
rate of ~ 0.1 HZ; H is a High Gain seismometer; and Z means vertical
orientation. From there I can see when the phases arrived. The original
earthquake, according to the USGS, happened at 2023-06-09 21:21:42
(UTC); the P phase arrived 15 & a half minutes after that, and
other phases past that. The P is barely noticeable, but the PP and S
waves definitely show up.
It’s interesting to compare this with the heliplot:

So – how does this compare to my seismometer? The Corvallis
station is 13,550 km away; mine is 13,990 km. The arrival
time should be a little later – by simple/stupid math, about 30
seconds later.
I’ll be honest: for this one, I’m not sure I see anything. The spike
around 21:52 seems like a candidate for the S wave.

Fiji Islands Region: Mag 5.8, June 10 2023, 09:12:50 UTC
Here’s the USGS page for this one, and the IRIS page.
Here’s the Corvallis data:

I’m curious to know that that 09:50 spike is about…but let’s keep
going. The S and SKS waves showed up pretty strongly at 09:35 or so.
Here’s what I saw:

Don’t know that that 09:28 data is, but there’s bupkiss at 09:35. Hm.
Anderson Springs, California - Mag 2.6, June 10 2023, 09:43:04 UTC
The peak at 09:50 caught my eye, so I tried looking for anything
around that time. I found a small one near Anderson,
California. Here’s the IRIS data:

That doesn’t seem a good fit either…but: Corvallis is 652 km away
from the epicentre, and it took 90 to 170 seconds for the waves to
arrive. I’m 1164km, about 1.8x further. Again, stupid math: 90-170
seconds becomes 160 to 306 seconds, or about 2.5 - 3 minutes. And
look at what I recorded 3 minutes after that quake:

Here it is really zoomed in:

This seems like a good candidate to me!
So what next?
All this is just a first pass through the data (and a very manual one
at that).
-
I’d like to do more digging. Finding some way to automate at least
the collection of links & data would be wonderful.
-
I’d also like to compare my data against this station in Queen
Elizabeth Park in Vancouver, run by Natural Resources Canada.
I do wish it had a little better data view.
February and March got away from me…but fair enough, because I
started my new job at a 🌠SPACE🌠COMPANY 🛰📡🤯. I’m pretty excited.
Also, though, I got COVID and then a sinus infection in March, which
sapped my energy. I got over it, but man, that was not pleasant.
So what did I do?
Hardware hacking
- Continued to work on firmware for weather vane.
Climate letters
- Three sets, rather than the 8 that should have been. But COVID.
Work
-
So much Terraform work. I’ve got the luxury of setting up the
codebase from scratch, and so far (🤞) I think I’m doing a decent
job of it.
-
Travelled to Edmonton for an offsite, and actually met my team in
person for the first time. I talked to almost everyone in the
company. Such an amazing bunch of people. And hey, our first
satellite is due to launch real soon now…
First thing to mention, which doesn’t really have a category: I walked
from my home in New Westminster to UBC in one day; it was about 32km,
which is the longest walk I’ve done in one day. I am mulling the
possibility of walking across the US when I’m 60, and this is the kind
of daily distance I’d want to maintain. I got some good blisters and
was sore the next day, but not crippled; I think I could have done
that again. It’s a good sign.
Webby
Hardware hacking
-
More work on the weather vane; got it mounted on a peanut butter jar
lid. If that sounds silly, then in my defense it turns out to be
very handy to have a standalone mount for a project.
-
Made an HTML page to display readings from the weather vane, using
javascript to rotate an arrow graphic to reflect the direction it
was measuring. Surprisingly handy.
-
Bought an Ikea Vindriktning, aiming to read its measurements
directly with an ESP32. Took a while to figure out how to get
it working – turns out that a common ground between the ESP32 and
the sensor board was necessary to get the UART working – but I
think it’s coming along.
-
Took apart a coffee maker that died on us to figure out what was
wrong, and it turns out to be a thermal fuse that blew – apparently
this is quite common. Will be picking up a replacement and seeing
if I can get it going again.
ML/AI/Earth Observation
Space
- After nearly 5 years of searching, I have finally got a job in the
space industry: beginning January 9th 2023, I’ll be working for
Wyvern Space. They are building satellites to do
high-resolution hyperspectral imaging; my position is senior devops
software developer, helping to build and operate their image
processing pipeline. I couldn’t be more thrilled. 😁
Trying to get back to doing these things on a regular basis.
Hardware hacking
-
More work on an electronic weather vane, following these
instructions. Lots of figuring out what size of bearings I
should order.
-
Some soldering to make a battery holder for some ESP32 camera
modules I’ve got.
-
Weather station:
-
Try to get the tipping bucket rain meter working; there’s a loose
connection somewhere, and periodically I see that Burnaby had 5
metres of rain in the last 24 hours. I never realized just how
much you have to pay attention to loose wires.
-
Sketch out a new rain meter based on inexpensive flow meters, then
order some. We’ll see how this works.
Webby
-
Going through a number of online courses/resources:
-
Coursera UMich Web Design for Everybody course: excellent,
though aimed at people quite new to development of any sort. One
thing: I’m lucky enough to have my employer pay for this, but the
lecturer, Colleen van Lent, writes:
My motivation for creating this course content was to spread the
mission of free education to everyone. Unfortunately, many of
the platform changes has put the material behind paywells. I
highly encourage students to take the courses individually
(rather than as a specialization) to access them for free. Even
then, some of the assignments may be hidden. I am hoping to
launch a new more open version in Fall 2018.
-
Shay Howe’s HTML & CSS course; also excellent
-
Javascript.info: awesome walkthrough of JavaScript
-
Trying to get the basics down, then look into React or some other
front-end framework.
-
Gotta say, I’m really fascinated by the tie-in between JavaScript
and DOM manipulation, which I had not really grokked before.
-
Project-in-progress is a refactoring (not a redesign, as I want the
look to remain about the same) of The Floating Head of Ayn
Rand, which has been more or less untouched for HOLY CRAP
twenty-one years. (State of the art at the time was
table-based layout, which I adopted enthusiastically 😬).
-
Changed the CSS for this site to have the post titles be a bit more
prominent:
.posts-list-item-title {
font-size: xx-large;
}
Data
- But also web: begin taking up work on the New West Trees page
again.
- Newest feature: adding links to the Wikipedia page for a
tree species!
- Coming soon: adding common names for species (eg: English Oak
instead of Quercus robur)…which turns out to be surprisingly
tricky.
- Tried pytaxize, which was a yakshave to get an NCBI API
token, then gave me problems re: rate limiting
- Tried pygbif; better results, but still not great for
trees. Example: Quercus palustris is resolved to just “Oak”,
but Wikipedia clearly resolves it to “Pin Oak”.
- But this gave me the idea of trying wikidata or
wikispecies; this is up next.