eric@scace.org said:

How much does Boulder move?  I'd guess not much so measuring motion will be
tough.  You could try to get a lower limit on the speed.

Looks like that part of the country is not interesting to the USGS:
https://earthquake.usgs.gov/monitoring/gps

Ballpark numbers.
With a good setup, GPS gives location to about 1 mm.
I live a few miles from the San Andreas fault system.  It is shifting about
as fast as your fingernails grow, roughly an inch per year.

Measuring rotation will be tough if your 2 stations are only 100 ft apart.  Do
you have a friend 1, 10, or 100 miles away?

PS: Make sure that your antenna mounts are sturdy.  You don't want them
drifting as the house ages or you bump into them.

There is a major USGS campus on the Boulder side of Denver.  You might wander
down there and ask around to see if you can find anybody familiar with either
GPS or earthquakes.  Or try their web pages.  There is probably a public
information contact.

California is rising about 2 mm per year.  Sea level is rising about 3 mm
per year.  Net is 1.
East coast is sinking about 3 mm per year.
(Major risk is surge and waves from hurricanes.)

--
These are my opinions.  I hate spam.

Eric,

Good idea to experiment with an antenna on each end of the house. You'll
get all sorts of data and your eyes will be eager to read house movement
into the plots. The problem is you won't know for sure if the results
are real or not; there are many factors, especially for a house made of
wood. See plots for my home/lab. [1]

So a suggestion is to place at least one of the antenna on a waterproof
turntable and continuously rotate it, off center, very slowly, maybe one
turn a week, or month. Then look at your data and see how well you can
detect that known movement.

By comparing actual GPS data with your calculated turntable antenna
location you can establish the position detection sensitivity of your
setup. Which is to say, if you can't detect a known antenna movement
of a few inches around a month there's little chance that you're going
to confidently detect an unknown seasonal or tectonic ground motion of
a few mm a year.

Alternatively, use a waterproof linear XY stage and each midnight shift
the antenna 1 to 12 inches north and 1 to 31 mm east based on the month
and day number. At the end of the year you will have created the coolest
GPS plot ever seen. It's kind of a slow motion version of "geowriting". [2]

/tvb

[1] http://leapsecond.com/pages/quake/

[2] http://leapsecond.com/pages/geowrite/

On 11/21/2019 11:36 PM, Hal Murray wrote:

Hi

Indeed, I do a better job of detecting the impact of humidity ( = rainy season ) on my deck
then I do detecting anything else on a fairly short baseline. I do get correct distances and
angles between the antennas (as verified with a tape measure).

One thing you can do (with some effort) is to collect a lot of data. There is nothing magic
about a 24 hour run. A two week run is also do-able. Lots of data will not take care of all
problems. It will get the portion related to SNR down to some very small numbers.

For continental drift sort of things, find a solid chunk of rock. If it’s connected to something
very deep, that’s best. Tie the antenna to it, possibly with a fairly short mast. Best to have a
good view of the sky so a short mast may not be practical. In the various papers on the subject
you see antennas on a 2’ mast mounted on tops of hills if it’s a “quick” setup.

=====

If you are going to all the trouble of setting up three or more L1 / L2 systems with good antennas
on them and monitoring that for months / years ….. I’d suggest that watching the PPS out of
each of the receivers might be interesting as well …. :)  The uBlox units mentioned earlier do have
a PPS out that is pretty good. They have a cousin, the F9T that is slightly more timing oriented.
The F9P board is a much better deal / easier to get. (though uBlox is very quick at shipping
the F9T’s direct from Europe).

Bob

On 11/22/19 5:04 AM, Bob kb8tq wrote:

That may only be 2 feet sticking up above the surface. There could well
be a 10 meter deep double pipe. I can't seem to find the website now,
but for SCIGN they had instructions on drilling a hole down to rock,
putting a large PVC casing in, then putting a second steel pipe inside
the large casing that actually carried the antenna, to decouple the
surface movement effects (soil moisture making the soil expand and
contract, for instance).

Try here:
https://kb.unavco.org/kb/article/permanent-gnss-gps-station-planning-technology-equipment-costs-55.html

and
https://kb.unavco.org/kb/article.php?id=104

Thank you, everyone, for your enthusiastic guidance and observations to my quirky question.

Quite a few mentioned the difficulties in measuring rotation over a short baseline. In response to the question of “is there another measurement point 10 miles away”, the quick answer is yes: NIST is on the opposite side of Boulder City from me.

The question of a sturdy — i.e., dimensionally stable — antenna mount brought to mind something I learned during my home inspection. My house is built at the foothills of the Front Range in North Boulder. Soils there include a kind of clay that swells significantly when exposed to water. As a result, house foundations are build on a system of screw pilings that go down to bedroom. The house’s cellar floor is a concrete slab poured on corrugated steel plates supported by cross-web girders that sit on these pilings. The cellar walls (to which the higher-precision pendulum clocks are mounted) are poured concrete that also rests on these pilings. It seems the house foundation is probably a better reference point for antennas than something sitting on the ground at the corner of my (tiny) lot.

Of course, my house and the neighbors’ houses are obstructions to signals for an antenna attached directly to the foundation walls.

The plan was to install two antennas (for two GPSDOs) mounted on short roof-penetrating metal tubes secured to the roof framing, just high enough to clear the neighborhood’s rooflines. This will give full sky access (except for the portion obscured by the foothills of the Front Range, a problem that plagues NIST as well). For time purposes, my understanding is that this should be fine.

For millimeter-scale position determination, this sounds like a more difficult situation. The house is generally wood framing with some structural steel elements (not in useful locations). Position measurements would contain noise from the diurnal/seasonal changes of the house framing. Maybe that could be averaged out?

— Eric

This is partly[1] why I'm doing the same exercise[2]:  I just built
a new house and it seems that nobody in Denmark knows how much the
wooden roof construction moves over a climatic year.

Poul-Henning

[1] The other part is that I cannot imagine a more relaxing hobby than
"Watching continental drift in real time" :-)

[2] But only one antenna because Scandinavia has no noticeable rotation.

--
Poul-Henning Kamp      | UNIX since Zilog Zeus 3.20
phk@FreeBSD.ORG        | TCP/IP since RFC 956
FreeBSD committer      | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.

On 11/22/19 10:17 AM, Eric Scace wrote:

Interesting problem.. You could put a steel (or Invar?) pipe down the
side of the house to hit the foundation, but then as the house moves
from side to side (wind, temperature, humidity) it would move the
antenna. Some form of rigid spaceframe around the house would result in
aesthetic problems. And drilling a hole all the way from the roof to the
foundation is probably a non-starter (although, in my house, there is a
sort of utility chase that does go from top of 2nd floor to 1st floor),
but there's probably something in the way (HVAC vents, most likely).

I'm not that much of a GPS nut (yet) (because you know, living in
southern California, there's not much in the way of geodetic measurement
infrastructure I can leverage <grin> - Self reliance - at the end of
civilization as we know it, at least you'd be able to measure
continental drift without depending on others...)

Maybe, but if the uncertainties are big enough over time, then no amount
of averaging helps.

I lived near Lyons, CO for about 20 years. ISTR that there is a fault line in CO, I think on the western slope.

DaveD

Sent from a small flat thingy

Hi

If NIST puts their real time data on any of the networks, you may be able to shortcut
the process. Goddard puts up data so indeed NIST may also do so. The trick is to
stream the data real time into the F9P or into a PC. You effectively solve for the distance
and bearing to the “reference” station. This is what a typical survey crew does on
a daily basis. Not much use for continental drift, may be a way to see if the house is
sliding off into the gulch ….

Bob