If you collect quality data, you do not need to use averaging.
Genuinely asking for you to elaborate. Are you talking about conventional survey? I'm not clear in what way you could collect 'quality' GNSS data without averaging or applying some other statistical analysis to derive a most likely position from a given set of observations.
Seeing some discussion on using the GNSS Analysis routine in SurvPC, as shown in the Mark Silver video previously posted. Perhaps this is common knowledge, but it was a learning lesson for me so thought I would share:
GNSS Analysis and GNSS Averaging (Raw Data Averaging) are two separate and distinct routines present in both SurvPC 6 and SurvPC 7. However, in operation, both are implemented from the 'Average' menu button. It is a setting which determines which type of average you will use when 'averaging', either GNSS anaylsis, or a plain-old average.
From my understanding, GNSS Analysis looks at each of the vectors in each of your observations and performs some type of least squares adjustment. If you have at least 3 separate observations, after running it will give your position a 'quality' rating of 1, 2, 3a, or 3b. The documentation is very vague on this, but evidently those ratings are supposed to correlated some unknown jurisdiction's classification of survey precision requirements, (1=City Center, 2=Rural, 3a=Farm Land, 3b=Forest, 4=Unproductive Land). You could also export a fairly extensive report detailing the results of the analysis.
GNSS Averaging, in the software called a Raw Data Average, is just a plain old add-them-up-and-divide average.
In SurvPC6, I used the GNSS Analysis routine as my main method of storing boundary or other key points. What I've found is that the new workflows of SurvPC7 do not seem to play well with using GNSS Analysis. Most of the new features seem to be expecting you to use the standard Raw Data Average, including the new Least Squares Adjustment routine. I'm also not sure that you're able to take multiple observations on the same Point ID and using GNSS analysis to get a coordinate anymore, you would need to take each observation separate, then use GNSS Analysis to store a final point.
Overall, the updates to SurvPC 7 are so extensive and useful, it's definitely worth the tradeoff.
If you collect quality data, you do not need to use averaging.
what way you could collect 'quality' GNSS data without averaging
I'm guessing he means to let the receiver and/or data collector do the averaging before storing one point, rather than storing multiple points in the collector and then telling it to average them.
@bill93 Yes - that is what I meant. I have observed multiple sets of data on the same point in the same near time and generally do not find any reason to not accept that first point stored.
Errors in positioning - if sensor on a pole and not bi or tripod, then there is error in manually centering. And if your rod is not plumb, you have error. These errors influence positioning and from my experience, adding another observation has not increased the accuracy of the survey.
If you want to do averaging for a geodetic position, then come back later with a new constellation and observe same points again. Relative positioning to other points in the survey are best done with same/similar constellations and averaging generally does not improve the accuracy. Just my 2cts.
We used to have a really screwy procedure for GNSS work - crews would have to observe six ten-second (or ten six-second?) observations one after the other, and would never come back to the point.
No re-centering, no turning the rod 90/180, no RTK reinitializing, no dropping & reacquiring SVs...
Even worse, the office would simply look at the spread of the points and arbitrarily decide which ones were "outliers", and toss them from the average. We definitely got some skewed data, because after we switched manufacturers and changed up our procedures to better align with best practices, we were noticing some very loose check shots when going back to legacy projects.
My preference (assuming decent conditions) is to take two 1-minute observations, re-initializing RTK and re-centering the rod in between, and then return to the point no less than half an hour later and do the same thing once again. This provides a check on the rod each time as well as a check on the initialization solution (even though we don't have fixed/float receivers and rarely see a "bad" solution).
The only reason I don't observe a single 2-3 minute observation each time is that I don't want to return to a point later on in the day, tie it an again, see a large delta, and have to wonder whether the rod or the RTK solution was off on the previous session.
@toeknee I'm 99% sure you can average over multiple sessions with survpc 6. You just enter the existing point number in store points and you can average additional observations.
I enabled this feature on the options before, it works with TS shots as well. I expected that since I was doing highway work, we would be using this for control points. As control lines are well established already, that is not the case.
Now I can not find how to disable it! I like overwriting point numbers if a mistake was made, rod height usually, and with averaging enabled it does not allow that.
Now I can not find how to disable it! I like overwriting point numbers if a mistake was made, rod height usually, and with averaging enabled it does not allow that.
Overwriting points with Carlson can create issues. The raw data for the first shot does not get erased, so if you process from the raw data, it can make things weird.
I was doing this in stakeout mode, so no raw data processing. But I will heed your advice when setting control points and the like.
Now I can not find how to disable it! I like overwriting point numbers if a mistake was made, rod height usually, and with averaging enabled it does not allow that.
Overwriting points with Carlson can create issues. The raw data for the first shot does not get erased, so if you process from the raw data, it can make things weird.
I'm OK with @dmyhill, I also try to avoid to reuse the same point number, especially for total station survey. For stakeout or control, I'm using suffixes.
I didn't see this mentioned, but the biggest improvement with SurvPC 7 is that the program is 64-bit....Every tablet I am running it on runs noticeably faster.
I didn't see this mentioned, but the biggest improvement with SurvPC 7 is that the program is 64-bit....Every tablet I am running it on runs noticeably faster.
Seconded. Very peppy now.
The Carlson RT4 has twice the memory which is also twice as fast and twice as fast a processor. Huge improvement over the previous equipment. Little surprised it needed more than 4GB but I guess when using DXF files the footprint can get large.
