How Long Between GPS Control/Boundary Corner Occupations?
I am trying to establish a minimum time "policy" between the occupation of points requiring accuracy and redundancy for confirmation. Things like existing corner monuments and control - not the random topo "x" shot.
There are other checks and procedures I require, but this post is solely about the time between occupations to ensure a sufficiently "new" constellation.
Currently, my instructions are:
1. If this is a multiple day survey, get the shots requiring redundancy on different days and preferably different times of day.
2. If this is a single day (or last day of the) survey, get me the longest time between occupations that leaves time for me to (a) download the observations from the field; (b) do some analysis/calcs; (c) push the corners to be set back to you; and (d) leaves time for you to set them before leaving; BUT
3. No sooner than 2 hours - even if this means you're going back tomorrow - with certain exceptions - not an absolute.
With 4 GPS/Glonass/Beidou/Galileo available to us, we don't really time/plan the GPS surveys anymore. The 2 hours is just my gut answer. I am wondering...what is the real minimum time specified in #3? I think I could go with 1 hour.
I would like opinions/input on the following:
I have heard some say it only needs to be 20 mins - seems short to me - what do you think?
Is anyone aware of any studies that have considered this (Midwest, North America)?
What are your policies for field crews regarding this issue and how did you arrive at them?
What are your experiences?
If you are measuring a boundary, only concerned with local relative positions the main thing is to let the satellites move a little.
If you are setting control using distant CORS and need later measurements to match, the iono/tropo propagation is important and that changes slowly.
The recommendation for time displacement are all over the map. I've seen as little as 20 minutes being recommended. The fact is, with multiple constellations and modern algorithms, the time displaced multiple shot thing is not as important as it used to be. These days I'm more concerned with other sources of error, ie/centering.
Who can legally determine the length of time required to get the "best" shot is my question.
Seems like this should fall under a minimum standards protocol and boy what a can of worms that's going to be.
NSPS has guidelines for monuments and the like, but who actually gets to determine what time matters, great question!!
Base/Rover? Open sky?
Time isn't important for constellation geometry anymore that I've seen with new units. When there were 5-6 satellites then sure it was, but 20-30 not really.
If it's a concern use some of the options with your GPS to change it real time. The easiest way is to change the elevation mask. Saves all the time of returning to the corner. Of course, rotate the unit, refix, all that good stuff. If the base is a concern run PPK as you survey and check in to a nearby CORS point.
Maybe you want to change atmospheric conditions if the control is far-flung.
In canopy? That's a case by case basis, the new units should really guide you there.
To further complicate the issue, many of the standards out there were written when there was only 1 or 2 constellations available, when rapid orbits were not that great, when single-baseline processors were not common, when RTNs were of greatly varying reliability and quality, when RTK was considered evil and impossible to mix with static work...
...when whoever was writing the standards drew solely upon their personal anecdotal experience rather than controlled testing and research/white papers...
Let me add a little based on recent replies.
(1) we will have short base lines (<1000 feet typically)
(2) you should assume that some points will not be ideal (canopy)
(3) Base/Rover setup - GPS/Glonass/Beideio/Galileo typically 22 minimum sats in solution, sometimes 24-28.
(4) good software (multiple fix technology, auto fix resets between sessions, blunder checking, point analysis, averaging, least squares adjustment, cool stuff) - but this question is not about the software other than it should give you some comfort
I need a policy boiled down to a single number to tell field guys..."This is my minimum threshold, don't go under it unless you have a reason and you should call to discuss if possible."
The nerd in me would be interested in some study or written documentation based on modern equipment and GPS, but I've never seen any. I would like to hear what you tell your field guys.
The time required will be highly dependent on your equipment and your knowledge of it.
With full GNSS you have some units (Javad) that can create thier own 'constellation change', running four different configurations at once. With other multi-constellation units you can enable or disable birds or fleets of birds to simulate a changed constellation.
If you're running GPS only (which is still useful in some cases) aim for a two hour spread.
Good luck, Tom
Honestly, your single number is going to depend entirely on what type of work you do, where you work, what gear and software you use, what methods you use, and what kind of results you want.
The nerd in me would be interested in some study or written documentation based on modern equipment and GPS, but I've never seen any.
It depends on your required accuracy per type of survey you are preparing and what equipment combination and method you are using. I can provide my 2 cents worth on Static and Network RTK, I have stopped using Base/Rover RTK. With Network RTK, I see a couple of hundredth per session, 3 to 5 in 20 minutes, 5 to 7 in 2 hours and the same 5 to 7 hundredth the next day. So I would say your one hour difference might work, but I know two hours is useful for a redundant shot. I do not see long observations, like 5 minutes per session, to be useful with Network RTK. After about 20 seconds it just continues to wallow around the same circle. Doing control runs though the same control points throughout the day gives you a nice pattern to work with. Then if you wish to adjust a weird group of shot, you have a basis to work with. Yes, of course, doing a control run through the points the next day is even better.
I don’t really have an error circle with Static, I am sure there is some, but the other errors I have cover them up. I’ve been hitting the same points over and over for almost twenty years and the year to year movement in not perceptible. I am down to a twenty minute observation duration. Mark Silver says I could go to ten, but it takes me twenty minutes to move the other units, so I am staying at 20. (Earlier this week I had a 14 minute observation ‘cause I was tired, I was dark and I wanted to go home, it had the same numbers as the rest.) I don’t have a redundant observation time difference with Static. My redundancy is occupying the same point with Network RTK or Robotic Total Station.
If I will be setting rear corners with Network RTK, I leave two stadles. That way the corner is setback with the same alignment that the boundary was calculated with. I have only gone back once to bump a corner over. That still beat cutting line or using Static. Good Luck
I would start with the RTK class guidelines by ngs. Now saying that it was written for base rover and gps. If you are collecting more that gps and glonass. I would do a hour minimum if i moved my base that helps with redundancy and allows for a least squares if set up correctly. I am torn on this as GPS only And truly having an understanding of it from a survey and orbital determination standpoint. In that case 4 hours. I have not studied or kin folk from the other side of the ponds nor chinas system in depth. Gps +glonass 4 hours if I want a better than a tenth with confidence. I don’t have a lot of data points but from using base and rover in the wide open full gnss I think an hour is good in most situations. The thing is if your crews are taught the hz vt precision and can do a little estimating in their heads from the difference in that time frame from a base move. Pdop gdop. Plays a role. 8 am today and 8 am tomorrow is not a constellation change. I did a job recently that i was able to test this. I had two days in field canapy and wide open. Three observations to each corner and control point and base moved. The hour gap and two obs were not much different in final coords than the using 3 observations and one having 4 hours. In relative aspects. Now in canapy two obs 1 hour apart had enough difference in final coords from using all 3 that i used all 3. Now my survey would not have been bad or wrong i still landed on a 1 inch pipe but using 2 obs and comparing to one 4 hour had about tge same results as holding all 3. This is not enough test to help or say this is gospel yet. I am building that. I would say in any multipath environment 4 hours gap wide open 1 hour minimum but that’s moving the base. My 3 observation 1 to 4 hours gaps moving base makes it negligible error when checking direct ties with robot as well i mean it is about flat . That is reducing my ground distances to ellipsoid and back to grid. I have found the geometry of the obs from base to control or corner helps in final adjustment. So i tie everything a minimum of twice from different base set ups. And shoot for 2 hrs in my planning.
I think you guys are over thinking my request (but thanks!).
I want my guys to quit giving me two 5 minute sessions 2 minutes apart.
I am curious how long it takes to get a different "satellite configuration" in the sky. I mean enough different satellites and/or different locations from one occupation (with a few 120 sec sessions) to the next occupation (with a few 120 sec sessions) to provide some "meaningful redundancy".
I think I'll just go watch a skyplot in the DC and watch the time. Maybe I'll wait for a warm Friday afternoon in the back yard with a cooler?
If you can find it makes a difference with RTK lets us know. But it doesn't.
I don't see it, crews don't see it.
In canopy time can be important, but that's not cause of the general configuration of satellites, that's cause of multi-path and of course changing satellite location will change the multi-path.
It has been my experience that I don't get a much of a difference either. I wonder if I am going to far by seeking to separate occupations by more than some required minimum.
If you want different constellations and you want them quickly, enable only half of the visible satellites and collect for your standard time and then disable the ones you used and enable the ones that were not used. You have redundant data from different constellations and it will process fine since you have 4 different systems to make use of.