DPOS discrepancies

avoidthelloyd

Active Member
We need a DPOS section in this forum. I have a 2+ year old mountain subdivision project that I have on going. I have built this project from the 'hybrid rtk' method over the years as my poor radio performance required 3 base positions. I always DPOS my Bases and points associated as soon as my first day of work is done.

Now it is 2+ years later, I have one base from 2020, one from early 2021 and a new one in late 2021. I periodically tie into these to maintain positional integrity. What is a normal tolerance for your check shot when you tie jobs together?

Today, I set up on the newest base and tied some pins I set from another base 2 years ago and they are 0.25' different. I ran DPOS again on this base (about a year later) and got a 0.09' positional difference.

What do you consider 'good enough'. I think I'm well within my states minimum standards for this very rough rural terrain, but still.

What factors are playing into this variance? I do check my bubble regularly! Thanks.
 

Nate The Surveyor

Well-Known Member
My explanation will of necessity be incomplete, and not totally the last say on this... Ok?
The fact is that continents are in motion. It's even expressed in velocities. And parts of continents are also shifting, rotating, going up, going down, and are quite simply, not altogether stationary.
The satellites are flying at high speeds. There are tracking stations, all over the world. Some of these stations are mounted on plates that are in motion.
Your survey is also in motion.
An epoch date is where it was, at a particular time.
So, if I use an epoch date of 2010, that's an adjusted position, keeping that coord tied to the ground under it. If I start a job in 2010, and then change the epoch date to 2020, now we can see the movement of the plate in question, over a 10 year period. Some places are quite large, like 1 foot in 10 yrs.
Quite simply, it means that coord has an applied adjustment to it, so that it artificially is attached to that dirt. Think of 100 postage stamps, on a basket ball, with all of them in slight motion. Do we keep generating new coords, for all our surveys on the postage stamp, or do we tie that postage stamp to the ground, so that the grounds motion is invisible to the user. If we change epoch date, then we find crustal motion.
If we KEEP a single epoch date, crustal motion continues, but it's corrected for, so that our surveys stay intact.
Nate
 

Michael

Active Member
I see what you are saying. Would localization using the base coordinates of the original base bring it more in line.
 

Nate The Surveyor

Well-Known Member
As surveyors, we are mostly interested in local coords. Meaning ground scale, and 100% compatibility when integrating jobs.
The latest system now makes it easy to COPY localization.
If I had it to do over, from 5 yrs ago, I'd come up with 1 single scale for ALL my surveys, within a radius of 75 miles of my house. It'd be about 95 ppm. Or add 0.095' per 1000', to go from GRID to GROUND. Then all my surveys would be very close to GROUND scale, and on the same coord system. And all surveys on 2010 epoch date.

Nate
 

Nate The Surveyor

Well-Known Member
But, back to your question. If you tie into all 3 bases, from one setup, and use their local coord, you eliminate the error between the DPOS bases. You can post process if radio does not reach.
N
 

avoidthelloyd

Active Member
All that being said, where can I research my data to find that information? The unit? The PDF ? Thanks.

PS. You didn’t answer my question. What is “close enough” for you in rough terrain?
 

Adam Plumley

Active Member
JAVAD GNSS
Tap points, highlight a point, tap the blue box. You will see the base/rover statistics. Out to the right of the coordinates is the epoch date
 

Nate The Surveyor

Well-Known Member
PS. You didn’t answer my question. What is “close enough” for you in rough terrain?
OK, let me add some parameters.
For: Land Boundary corners, set, and found, how close do I feel they should be set? Well, my goal is always + - 0.05'. Because that is what the unit is capable of. In most environments. Truth, is, some of those are more like 0.10' and occasionally 0.15' . I try to use only ONE base for most work. I feel changing bases adds about 0.03' of error. If you use RTPK, or Post Processed RTPK, you can keep it small. To achieve higher accuracy, just takes a little more time. And, sometimes a multiple shot.
A trick to set a deep woods corner, during poor GPS time, is to get a point within a foot of the corner. Set the Javad up, and shoot it 3x. And, then use average, and a compass and box tape to set the corner. This gives you a high accuracy control point to work off of.

Nate
 

Shawn Billings - Javad

Active Member
JAVAD GNSS
Precision will depend on the duration of the set up mostly. The longer the base setup the more precise it will be, generally. Precision will also depend on your specific CORS geometry. Some parts of the country have CORS more closely spaced, some further. I would usually want 4-5 hour sessions for determining a base coordinate. I don't always get it, but I feel confident in the horizontal and vertical results with sessions that long. I would expect 0.03' horizontal and 0.05' vertical (in my area) with a duration that long. At two hours or less, I would expect 0.05' horizontal and 0.15' vertical. I would only expect 0.25' horizontal with a session duration of 15 minutes.

Also, there is local ground movement. HTDP, for example, tries to compensate for local ground movement, but it's a model and based on the CORS. Like in a contour map, the contours will be more reliable in the places where a point was collected. So it's possible but perhaps not likely, that the area shifted over time.

If I understand, you now have three coordinates on this base? I'd be curious if you see the coordinates (A, B, and C let's call them) moving in a single direction, basically that B is between A and C. You've done well to note that you are comparing not just the base point but other points as well, as it could just be that a dozer track went by your base point 6 months ago and slightly moved it. But seeing that you have rover points around that corroborate the relative location of the base, we know that we're talking about a translation of the project network. So I wonder about some local velocity.
 
Top