GNSS benefits to having 4+ hours between observations

The material is a little dated, but I don't know that that will impact the topic at hand.?ÿ This is from "Guidance Notes for GNSS Network RTK Surveying in Great Britain", which is still available for download with a quick Google search.?ÿ Even though this pertains to network RTK, I would think you would see similar results from single base RTK too.

@alan-chyko  Thank you. That supports my experiences as well. I have heard that a separation of 30 mins is really all you need - anything else is just overkill and yields no significant benefits. 

A legitimate study and white paper would be useful--sadly don't have one at my fingertips.?ÿ But it seems intuitive by evaluating the sources of error.?ÿ Setting aside problems with just being over the station (level bubble, antenna height, offsets, and such), the typical GNSS-specific errors would seem to be: 1) satellite geometry, 2) local obstructions, 3) reflected signals (multi-path), and 4), initialization errors. So based on your scenarios;

A) Changes initialization

B) Changes initialization, geometry, local effects, and maybe set-up

C) Doesn't change much (maybe more observations over limited time)

Redundancy is our primary check.?ÿ But evaluating the data can be a little tricky and potentially misleading?ÿ Redundant observations that don't have changes may look precise but may lack accuracy.?ÿ Observations with lots of changes may lack precision but ultimately could be more accurate.

Ultimately, we rely on education, experience, and professional judgment applied to the specific conditions--but good research and write-up on the topic would be helpful.

The 4 hours is from the days of fewer satellites, especially with GPS only. The intent was to get a big change in the actual satellites used and get them to cover various sky positions. With todays 12 plus GPS visible at times, plus 30 or more other constellations that rule no longer applies to GNSS. What you do want is multiple short observations separated by some period of time. A second observation is not really a good check without a third observation. Much more important is if your need is for good elevations. Especially under canopy you want to vary the satellite sky positions as much as possible.

Paul in PA

@paul-in-pa  In this situation, its gonna be boundary related, no vertical. Using R-10 units, capable of seeing a majority of available constellations. I am quite confident that 3 observations for 30 epochs each separated by 15 mins each and then processed through Trimble's weighted average will yield results acceptable for a rural survey classification.  

I don't see the 4 hour rule doing much of anything. If you are looking for .1' h then get multiple locations, change your multipath environment by re-fixing and shifting your receiver (raise or lower the rod +-1'), you will be good as gold with the R-10's.

I can't give you a paper study but we are seeing less than .05'h consistently in moderate canopy with the R-10's. We have yet to find a bad point. If it says it's good then it's good (I guarantee it-where's that sarcasm button), you have probably been frustrated waiting for the unit to say it's ready under canopy, I trust it so far, it hasn't allowed me to locate a point that didn't work at least a second time and they always check. I know a poster related a bad shot he found, back in the early days I didn't go a day without finding one or more bad shots.?ÿ

Heck if you are going to do 4 hour wait times it often works to lay out a boundary and on the way back to the base pick them up a second time, but if that isn't workable you should find that the second shot a few minutes apart will confirm the point. Satellite geometry today removes the reason for waiting 4 hours, the sky is randomly covered and if your base is close there isn't a reason for atmospheric conditions to do it.

If you do want different geometry it's simple to create it by turning off half the satellites for one shot, then turning those back on and the other half off, takes a bit of time but that changes the geometry completely. As we got more and more satellites we did that for a while but stopped, it didn't ever seem to do much of anything but take up some valuable time.?ÿ

@mightymoe I couldn't agree more. I am always trying to convince co-workers that the 4-hour window is no longer necessary. Of course though without some form of scientific proof it's like talking to a wall.?ÿ

?ÿ

@stlsurveyor

It's a bit like a carbureted 289 V-8 engine in a 67 mustang compared to a fuel injected turbo four mustang today. 100 horsepower vs 300.

Things change fast in the GPS world, I used to sit on a point in open country for 1/2 hr waiting for the dang thing to fix with 4 satellites and finally a 5th would creep over the horizon and I would be so excited, those days are long gone.?ÿ

?ÿ

Big part of the picture now is getting the final orbitals.?ÿ This comes a week or two depending how busy the head shed geodesists and people are to get them out the door.?ÿ Also depends on the accuracy and precision you require.?ÿ Gravity is a real PITA with elevation.?ÿ we always waited until we obtained these final orbitals to process the last data for reporting to the clients.?ÿ My $0.02

?ÿ

Even 180 epochs is too long, especially for the sought after precision. I'm having trouble justifying much over 30.?ÿ I worry more about the plumbness of the pole than I do the positioning. All these receivers are constantly reacquiring fixes so there is no big need to dump the solution.?ÿ My usual procedure is to occupy 30-60 seconds, rotate the rod 180?ø, and occupy another 30-60 seconds. Average the result. I like to hit some points a second time if the opportunity presents itself, but that is more to prove the process I'm using than it is to improve precision.?ÿ?ÿ

I think that JitterBoogie touched on what Accuracy and Precision you are trying to achieve will dictate. If you take 30 sec or 180 sec or epochs close together even 15 minutes a part you have precision. The constellation change of the satellites in there PRN slot will tell you why. Do this little test. Go out and plot the satellites by sat number and the PRN number at say 0800 then 1200 then 1600 etc.etc. for 24 hours Then wait 30 days do it again. You don't need to observe a point just look at there azimuth and elevation at those times in the same spot. What you will find is there is a huge change in which satellite is in which PRN at a giving time. The SN Sat number your data collector shows you is like a serial number it never changes. It is attached to that specific satellite. However not often but sometimes and especially when they launch a new satellite the PRN can change that is its reference to its slot in the constellation. This is GPS I am not speaking to others. Also every satellite is broadcasting messages Simplified= ( What Time it is , Its speed Velocity, Where it was)?ÿ So new messages get sent up to correct for many reasons but each satellite gets updated at minimum once per day and or every 2 hrs. I am not saying you should place a 4 hour gap in time but if you are trying to achieve accuracy then at least every 2 hours or your just achieving precision. NGS has a sample data set showing this every 2 hours. I still like 4 hours for anything that needs to be good accuracy because of my understanding of what goes on in the constellation and realizing that some of those satellites are very old and not all of them are carrying the new messages. NGS also has a User Guidelines for Single Base Real Time GNSS positioning which is very similar to the old style of Horizontal and vertical class orders and procedures. Yes there use to be less satellites and that was a big reason for 4 hours. Also there are what the surveyor side of me use to say its an anomaly that is why the point is bad. But the Orbital Scientist side of me joined to the surveyor side of me really understands that there are so many error sources to GPS and or GNSS. The manufactures really have done a great job of simplifying it for us and making it very easy to accept the positions. I am going to try and attach the User Guidelines for single base rtk ....?ÿ But if you wish to achieve accuracy you have no other choice but to at least place a 2 hour gap in time no way around it at this point in time. I believe when Galileo is fully operational and all the kinks are worked out then maybe not.?ÿ Today at 0800 and tomorrow at 0800 is not a two hour gap in time. You will have the same geometry roughly on both days..... And over 30 days well. just look at the numbers over time. The PRN and SAT numbers with Azimuth and elevation.... I hope this helps... P.S. The file is to big to attach so go to NGS documents / publications..?ÿ?ÿ

While responses to the original post seem centered on the changes in SV constellations, few mention the changes in meteorological conditions one experiences when following this guidance. Those performing precise geodetic leveling should remember the requirement for morning and afternoon runnings due to the same concern.

Geodetic surveys aim to obtain accuracy and not repeatability.

?ÿ

Relevant links follow:

?ÿ

https://www.ngs.noaa.gov/heightmod/MiamiTropDelayCorrections.ppt

?ÿ

https://www.researchgate.net/profile/Mehmet_Ugur15/publication/287113432_The_effect_of_tropospheric_delay_modeling_on_the_determination_of_GPS-derived_ellipsoidal_height_in_permanent_GNSS_networks_using_OPUS-projects/links/59c9a15145851556e97a7260/The-effect-of-tropospheric-delay-modeling-on-the-determination-of-GPS-derived-ellipsoidal-height-in-permanent-GNSS-networks-using-OPUS-projects.pdf

?ÿ

?ÿ

?ÿ

First off, a 15 minute RTK observation will not give you any better result than 30 to 60 seconds.?ÿ RTK isn't static data, it's just a position relative to your base station.?ÿ

Your key point was "under medium canopy". If that's your situation, it would be good to let time pass because of your obstructions and the RTK fix potentially being compromised due to "stop and go" signals.?ÿ If you are adamant about not coming back to the point, you should reset satellite tracking between your 2 RTK observations.?ÿ If it's a hike to get to these points, you might as well do 3 30-60 second RTK observations with reset of satellite tracking between them to confirm your observations.

Although if you are going go come back to the point later, it'd be good to move your RTK base so you can perform an adjustment provided you have software for it such as Trimble Business Center if you use that brand.