Hello everyone, hope you’re all staying healthy. It’s been a while since we updated you on our live Raspberry Shake data software project rsudp, and we have added a lot of great features since its initial release this past fall. rsudp is much more than the live visualization script it started out as; I can proudly say that it is now a fairly diverse and extensible suite of educational and scientific tools for live seismic data visualization, event detection, notification, and learning, which are capable of running harmoniously and continuously together.
rsudp testing/demonstration mode — rs-test — a simple yet incredibly powerful tool, which emulates the standard live client mode but uses data recorded in the past. rs-test runs a number of internal tests but can also be used to demonstrate what an earthquake looks like as it is recorded and detected in real time (i.e. in a classroom, or to test detection settings with your own miniseed data)
a new deconvolution setting ("GRAV"), which takes calculates the fraction of acceleration due to gravity recorded by your Shake (fraction of lowercase g to physics nerds)
alarm times in debug messages, saved plots, telegrams, and tweets are now reported with 0.01 second precision (previously they were reported with 1 second precision)
Social media additions
a Twitter posting module that can automatically send tweets immediately when an event is detected and when images of the event are saved
(examples here: https://twitter.com/paleosurface_eq)
a standardized module template to quickly get started with rsudp development ideas without the need to write new modules from scratch
a custom thread class (rsudp.raspberryshake.ConsumerThread) for consumers to inherit which contains all internal flags that the Producer needs to function
updates can come directly from the Python Package Index, PyPI, by executing the following command in your rsudp conda environment:
pip install -U rsudp
Final notes
It hit me how unique of a tool rsudp was a few months ago when developers from both obspy and matplotlib mentioned offhand what an interesting project they thought this was. rsudp stands on the shoulders of both of those incredible pieces of software and pushes their boundaries to create something we can be pretty proud of — thanks to Raspberry Shake’s data casting, perhaps the most up-to-the-second event detection and visualization software currently available in the passive seismic community. All free and open source. Cheers!
Hi Graham, glad to hear you got it working. Yes, as you’ve now figured out, rsudp will run on both Pi 3 and Pi 4, although from my testing the Pi 4 seems to have a lot fewer crashes due to RAM issues.
I havent had any crashes on my Pi4 but it crashes all the time on Windows 10 - It doesnt seem to run for more than about 5 minutes . I posted some logs from my Winows PC in an earlier post
I saw that and still don’t know what the issue might be. The symptoms are consistent with a RAM problem, but it is definitely machine-specific because several (Windows 7, 8, 10) machines I’ve tested it on run it fine for days or weeks without error.
and it hangs there.
I tried uninstalling Anaconda (which was there from some previous excursion with Rsudp) and reinstalling Rsudp, but the same message comes up about ffplay.
I inserted my station name R00B9 in rsudp_settings.json, but that’s all.
@iannesbitt, thank you for this. I was happily running headless using xvfb on an old version and eager to update when I saw this announcement today.
I upgraded my Pi4/Buster installation via git pull and running the install script again (per instructions).
One small glitch, the install ended with an error:
fatal error: ffi.h: No such file or directory
A web search suggested I try the following if I’m on Debian or Ubuntu:
sudo apt-get install build-essential libssl-dev libffi-dev python-dev
Without bothering to check what was already installed, I did the apt-get and reran the install script and all is fine.
As for updating settings, it seems that exec is no longer supported and the following line is ignored:
“exec”: “eqAlert”,
I’ve removed it and I’m assuming that my eq_screenshots = true under plot will still be honored.
Am I correct in that assumption?
Finally, I just want to thank you for this great software. I’ve been using it to alert me on quakes by monitoring the screen shots directory and emailing when new files are added. I post the earthquakes here: https://www.paloaltoweather.com/earthquakes.html
Usually, the heliograph is more interesting to me (the untrained eye), so I just post that.
Note: I also look for quakes every morning in the heliographs and post many that rsudp does not alert on. I’m going to play more with the settings, starting with your quiet vault settings and backing off from there.
Yes, as long as plot is running, setting "eq_screenshots": true should save an image.
Cheers, thank you for the kind words. I’m glad you’re getting good use out of it. As you can imagine, there are a lot of nonlinear variables in earthquake detection, so a single-station linear detection algorithm is imperfect in terms of separating the signal from the noise, but it is a lot of fun to play around with.
