Can we detect BE-4 testing at Corn Ranch by public seismometer data?

[LouScheffer]

It's well known that rocket launches are strong enough to be picked up on seismographs.  Test firing should be similarly visible, and perhaps couple to the ground even better.  Furthermore, there is a network of seismometers in he USA, including one near Van Horn, Texas, and at least some of the data from these is publicly available on the web.

Looking at this data, there are number of events that to me (I'm not a seismologist) seem plausible as engine tests.  They are all listed as magnitude 3.5 events 100-133 km from the seismometer, vaguely in the area of Corn ranch. ( Corn Ranch is about 80ish km away from the seismometer location, but as rocket tests are not earthquakes it's not clear to me that the magnitude and distance estimation tuned for earthquakes will work well for engine tests.)  They last just a few minutes, with no before or after shocks.

For example, shown below is the signal for December 17th.  It has two similar events about 4 hours apart, which might be two tests.

If anyone here knows the actual time and dates for BE-4 test firings, this hypothesis would be very easy to test.  Anyone know?  Maybe the data can be extracted from photos of the test firings?  If this works it would be very easy to keep track of test firings - just check the seismo data once per day or so.  For even more automation, there are APIs to get the seismometer data.

[DanClemmensen]

It's well known that rocket launches are strong enough to be picked up on seismographs.  Test firing should be similarly visible, and perhaps couple to the ground even better.  Furthermore, there is a network of seismometers in he USA, including one near Van Horn, Texas, and at least some of the data from these is publicly available on the web.

Looking at this data, there are number of events that to me (I'm not a seismologist) seem plausible as engine tests.  They are all listed as magnitude 3.5 events 100-133 km from the seismometer, vaguely in the area of Corn ranch. ( Corn Ranch is about 80ish km away from the seismometer location, but as rocket tests are not earthquakes it's not clear to me that the magnitude and distance estimation tuned for earthquakes will work well for engine tests.)  They last just a few minutes, with no before or after shocks.

For example, shown below is the signal for December 17th.  It has two similar events about 4 hours apart, which might be two tests.

If anyone here knows the actual time and dates for BE-4 test firings, this hypothesis would be very easy to test.  Anyone know?  Maybe the data can be extracted from photos of the test firings?  If this works it would be very easy to keep track of test firings - just check the seismo data once per day or so.  For even more automation, there are APIs to get the seismometer data.

Maybe a GoFundMe for a small seismometer net around Corn Ranch? Should be cheaper than overflights. I'm not competent to build such a net, but it is theoretically easy, using old Cell phones that have accelerometers.

[catdlr]

Where is the engine testing facility located on this map?

To verify also pinpoint where the SpaceX engine testing facility is to see if the TexNet system is sensitive enough to pick up the testing you desire to record.

https://www.beg.utexas.edu/texnet-cisr/texnet/operations-status

[catdlr]

TexNet Seismic Station Requirements

These are the physical equipment requirements to station a seismometer and connect it to the TexNet (not to be confused with Tex-Mex) system:

TexNet Seismic Station Requirements
Version 06.15.21

The minimal requirements for a portable station to be used for earthquake location within TexNet are:

Sensor:
3-component orthogonal axis seismometer
Nominally flat velocity sensor response 10sec to 100Hz
Steel case for posthole/direct burial installation
Sensor cable should have a 1000PSI connector

Datalogger:
24-bit digitizer
Sampling rate at least up to 200 sps
Integrated seedlink server for continuous streams of both waveform data and State-of-Health
Data transfer via Internet allowing access from BEG network (TexNet Hub)
Timing using GPS
Communication preferably using a web based user interface
Local storage

Enclosure, power, and communication
Ensure autonomous operation through solar power
Ensure autonomous operation on solar power for a minimum of two (2) years.
A pole mounted water-tight enclosure with mountable solar panel on top. Enclosures must remain functional during deployments lasting at least two (2) years or longer.
Preferably two 110Ah batteries, a power regulator (with a user defined fail-safe shutoff voltage) and a 160W solar panel or comparable set up
Data should be sent by telemetry to TexNet in real time

Other requirements
Sensor installation depth should be from 3-6 ft
Orientation preferably should be so that horizontal orthogonal components are to North (magnetic) and East. If not, azimuth clockwise from North (magnetic) should be defined.
Provide metadata information in the form of: (a) Dataless SEED[1], (b) Response files[2], and (c) FDSN Station XML[3]
Station uptime should be more than 95%
Station should be fenced to avoid noise

[catdlr]

Might just be easier (cheaper) to monitor monthly or in 2-week periods the collection of earthquake events recorded on the TexNet EQ catalog via this site

texnet/earthquake-catalog

[whitelancer64]

Where is the engine testing facility located on this map?

To verify also pinpoint where the SpaceX engine testing facility is to see if the TexNet system is sensitive enough to pick up the testing you desire to record.

https://www.beg.utexas.edu/texnet-cisr/texnet/operations-status

Blue Origin's test site is north of Van Horn, SpaceX's test site is southwest of Waco.

[Steve G]

Why is it so important to find out if Blue is testing its rocket engines?

[catdlr]

Where is the engine testing facility located on this map?

To verify also pinpoint where the SpaceX engine testing facility is to see if the TexNet system is sensitive enough to pick up the testing you desire to record.

https://www.beg.utexas.edu/texnet-cisr/texnet/operations-status

Blue Origin's test site is north of Van Horn, SpaceX's test site is southwest of Waco.

Thanks whitelancer64 for pinpointing the two locations.  According to the area near the SpaceX facility, using the sensors monitored by TexNet would be useless as they don't sufficient number or any sensors nearby.

[LouScheffer]

Where is the engine testing facility located on this map?

To verify also pinpoint where the SpaceX engine testing facility is to see if the TexNet system is sensitive enough to pick up the testing you desire to record.

https://www.beg.utexas.edu/texnet-cisr/texnet/operations-status

Blue Origin's test site is north of Van Horn, SpaceX's test site is southwest of Waco.

Based on these plots, likely my "detect testing" hypothesis is wrong.  The Delaware basin field is about the indicated 100-130 km northwest of the Van Horn station near the Mexican border, and SpaceX tests appear not to show up at all. 

To check this, we know the time and duration of SpaceX tests, and can look at the seismic data at that time.   I looked at the NSF testing summaries, but they don't have the date and time stamped on them.   Anyone know where I could find this data?

[whitelancer64]

There are some stations to the northeast of the Blue Origin test site, and two in particular: Station PB28, Culberson South, and Station PB35, UTL Block 46. These two are a part of a cluster of 6 stations near to each other.

These two stations are a bit closer to the Blue Origin test site than the Van Horn station.



And it might be worth emailing whoever is in charge of all these seismic stations. They might already know some information about Blue Origin testing, since they'd have to determine if an event is an earthquake or a rocket launch / test.

[LouScheffer]

And it might be worth emailing whoever is in charge of all these seismic stations. They might already know some information about Blue Origin testing, since they'd have to determine if an event is an earthquake or a rocket launch / test.

I emailed the TexNet admin, who forwarded my request to the technical guy, Alexandros Savvaidis.  He was happy for the interest and said they indeed classify events, and only publish the ones they classify as earthquakes.  But he also said:

We are able to identify other events, with quarry blasts being the most common one.
In order to identify rocket launces our sensors have to be close to the launching pad. Otherwise in most cases the ground motion is not easily traceable due to noise level (ambient noise at the sensor site).

For both launching sites you mentioned, we do not have sensors close to the launching pads so it would be difficult to locate the event from the sensor recordings. However, if you have interest in checking out let me know of a few rocket launching time (preferrable night launches where the ambient noise is lower) and I can check the sensor data of our closest site.
Also, if there is an opportunity to install one of our sensors close to a launching pad that would be great.

Thanks, and Best Regards,
Alexandros

This leads to two follow-up questions.  When was an exact time and date for a BE-4 test firing?  And how close is the nearest friendly, public place to Corn Ranch?  I suspect we could also add a seismometer to the NSF site at McGregor, but I don't know how much information that might add (though if it's sensitive enough, could it detect the mythical methane thruster firings?)

[LouScheffer]

From the YouTube video "This is BE-4", the engineering camera views indicate the test was done on 22 October, 2022, at 02:45.  The only question would be if the camera is set to local time (which I think is 5 hours ahead of Texas in October, hence 07:45 UTC) or UTC itself, hence 02:45 UTC.

I looked at stations VHRN and MNTX (two of the closest stations that are not in the oil fields) and did not see anything obvious at either of these times.  But the web interface looks only at the vertical component, and you'd suspect from the horizontal test stand that it would generate mostly horizontal components.

Also, as opposed to earthquakes, we know the location.  So cross-correlating two stations, with appropriate lags, might be a good approach at digging the signal from the noise.

I'll send a note to Alexandros and see what he says.

[whitelancer64]

Per https://www.fdsn.org/networks/detail/TX/

Station PB28   Culberson South   31.66864   -104.50077

Station VHRN   Van Horn   30.786736   -104.985193

BE-4 test site: 31.43003 -104.71977

Distance between PB28 and BE-4 test site: 33.684 km (20.930 miles)

Distance between VHRN and BE-4 test site: 75.86 km (47.137 miles)

I feel like 21 miles should be close enough to get some kind of reading.

[Robert_the_Doll]

There was the test of Qualification Engine 1 on 12-09-22 per the video seen in the Devin Garner video:

[LouScheffer]

There was the test of Qualification Engine 1 on 12-09-22 per the video seen in the Devin Garner video:

Nice find - the more firings the better.  But how do you know it's month 12?  For me the month is hidden behind the feather.  Also, at t=56 seconds, it switches to from 09/22 to 15/22.  So two different firings, I suspect.

Finally the day, date, and time format is different from the other video.  So again we have the problem of local time or UTC.

Thanks again for pointing this out.

[LouScheffer]

Turns out the seismic community is super good at making their data public.  There are web APIs to grab the raw data from almost any time period of any seismometer.  Here are the ones for PB28 and MNTX - they need two different commands since they are on two different networks, with different servers.  But at least they use a common format, which in both cases returns a .mseed file:
http://rtserve.beg.utexas.edu/fdsnws/dataselect/1/query?network=TX&station=PB28&starttime=2022-10-22T02:40:00&endtime=2022-10-22T02:50:00

https://service.iris.edu/fdsnws/dataselect/1/query?network=US&station=MNTX&starttime=2022-10-22T02:40:00&endtime=2022-10-22T02:50:00

Furthermore, there is a pubic library, ObsPy, that knows how to read this format.  When I asked it what was in the file, it returned for PB28:

6 Trace(s) in Stream:
TX.PB28.00.HH1 | 2022-10-22T02:39:58.510000Z - 2022-10-22T02:50:00.360000Z | 100.0 Hz, 60186 samples
TX.PB28.00.HH2 | 2022-10-22T02:39:58.990000Z - 2022-10-22T02:50:00.280000Z | 100.0 Hz, 60130 samples
TX.PB28.00.HHZ | 2022-10-22T02:39:59.650000Z - 2022-10-22T02:50:02.800000Z | 100.0 Hz, 60316 samples
TX.PB28.00.HN1 | 2022-10-22T02:39:58.655000Z - 2022-10-22T02:50:00.170000Z | 200.0 Hz, 120304 samples
TX.PB28.00.HN2 | 2022-10-22T02:39:58.120000Z - 2022-10-22T02:50:01.695000Z | 200.0 Hz, 120716 samples
TX.PB28.00.HNZ | 2022-10-22T02:39:59.905000Z - 2022-10-22T02:50:01.420000Z | 200.0 Hz, 120304 samples

It will likely be until the end of next week until I can try to interpret these.  But at least the data is there, is anyone wants to give it a go.  I'm thinking (just speculation) that the ratio of energy (per frequency bin) of horizontal/vertical might be a differentor.  It's the first thing I'll try.

[Robert_the_Doll]

There was the test of Qualification Engine 1 on 12-09-22 per the video seen in the Devin Garner video:

Nice find - the more firings the better.  But how do you know it's month 12?  For me the month is hidden behind the feather.  Also, at t=56 seconds, it switches to from 09/22 to 15/22.  So two different firings, I suspect.

Finally the day, date, and time format is different from the other video.  So again we have the problem of local time or UTC.

Thanks again for pointing this out.

Watch the video on YouTube proper. The initial firings are 12-09-22 and then it jumpcuts to one at 12-15-22. This confirms what Tory Bruno stated in a post that there were firings on a weekly basis.

Screencaps in case you cannot copypaste the URL:

[Emmettvonbrown]

Since Elon Musk was tracked by flight radar, it makes some sense to track Bezos ' Blue Origin by seismometers...

[LouScheffer]

I finally got a chance to look at the seismic data, and did not see anything obvious.

But from the recent Huntsville video, the camera clock is clearly set to local time.  (It was described as a late afternoon test, and the camera shows 17:03).  So maybe I was looking in the wrong place.

[LouScheffer]

Now we are making some progress.   Assuming the camera clock is in local time, I converted to UTC (+5 hours in October).  Then I processed the raw data by grabbing the seismometer records, doing a fourier transform on each 10 second segment, and then summing the power in the 1-20Hz range.  You can see that around the time of the firing, there are several points with about 100x the usual power in this band.

However, there are at least two problems.  One is that there are confounding events that reach this level - see the 'before' and 'after' plots for the 10 minutes before and after.  The second is that it looks like the event starts about a minute or so before the time shown on the camera.  How are times set on these cameras, anyway?  Are they typed in by the user or set in some automated way?