Occupation Time and Precision

[Matt Johnson]

There have been some questions about what the expected accuracy of TRIUMPH-LS should be and how occupation time affects accuracy. I am posting the results of some precision test conducted by John Evers in this thread. These test were conducted with a TRIUMPH-2 base station and TRIUMPH-LS rover in an open sky environment.

The settings used were:

Verify without V6 Reset
Consistency Level: 0
Accept Fixed Only, RMS: All, DOP: All
Accept Number of Fixed RTK Engines: At least 2
Confidence Guard: 0.164 ft (5 cm)

Auto Accept and Auto Re-Start were used to automatically collect many points at the same physical point on the ground. The number of epochs to stop after was varied for each data set. These spreadsheets show the frequency distributions of the points' horizontal and vertical distances from the average position for each data set:

2 EPOCH

5 EPOCH

10 EPOCH

300 EPOCH

600 EPOCH

As you can see precision increases with number of epochs collected for the point. These results were very good, in the last set 600 epochs were collected for each point (a 10 minute occupation time) and a total of 62 points were collected. Of those points, 83.9% of them fell within 0.01' horizontally of the average position and all fell within 0.02'.

 

[Aaron S]

I'd like to try and recreate this test myself using the Minnesota RTN (which is what I use for the majority of my work). Since this looks like it was done in Excel, could you or John tell me how you get it to calculate clusters like that? Or was that done outside of Excel?

 

[Matt Johnson]

I have attached the spreadsheet I created to do this. Paste the points into columns A to E of the Data sheet in the format of P,N,E,Z,D.

An option to view these type of statistics was also added to J-Field. To use it, average the results of the points and select it as a reference point. Next select that point as a reference point for the Distance Filter:

Then select a point in the points list and choose "Statistics" from the menu. You will be presented with a screen like this:

 

[Aaron S]

Thanks - I'll give it a try.

 

[Darren Clemons]

Great stuff Matt. Thanks for posting. I’ll add my two cents, for what it’s worth.

We did nothing to this extent when we first started using the Javad equipment, but rather we quickly saw the results in real time on a day to day basis while we worked. We’ve been “around the block” a few times and have had close to a dozen GPS systems since our first Novatel backpack system (heavy and HOT) back around 1998.
This Javad system didn’t require massive testing amounts of points like you list for us to see the difference! We started seeing it every day.

As soon as beast mode was enabled I felt this machine was, without question, the most accurate GPS system there has ever been. I’ve seen and dealt with the +- error of GPS for years and this system almost totally eliminates it - with the proper time and settings.

We would, and still do, always check a point or two on each days work on the same job and for years using the yellow brand, we were used to 0.06’ to 0.12’ or so on some places and days. That just doesn’t happen anymore. Anything over 0.03’ as Aaron said in the other thread is odd now.

We’ve returned to previous jobs now that we’ve done with this system and it’s just remarkable the consistency. If the points are off over 0.06’, you had a bubble out or the pole wasn’t plumb - it’s not the Javad!!

If ya need it closer than 0.06’, well hmmmm, not really sure what you need to use.

 

[James Suttles]

I know this is an old post. So with that in mind, using a 5hz recording rate, you can obtain 600 epochs in roughly 2 minutes, not 10 minutes as before. With that in mind, is it the time the that is the determining factor or the number of epochs?

I am truly amazed with the LS, and I too have some yellow brand boxes, and the LS eats them for breakfast, its quicker and way more forgiving in bad coverage areas.

 

[Matt Johnson]

Time is the determining factor.

 

[Adam]

I know this is an old post. So with that in mind, using a 5hz recording rate, you can obtain 600 epochs in roughly 2 minutes, not 10 minutes as before. With that in mind, is it the time the that is the determining factor or the number of epochs?

I am truly amazed with the LS, and I too have some yellow brand boxes, and the LS eats them for breakfast, its quicker and way more forgiving in bad coverage areas.

Time on point is by far more important than epochs.

 

[Matthew D. Sibole]

Time on point is by far more important than epochs.

100% agree!

 

[Nate The Surveyor]

I second Matt's comment.
It's a statistical measuring tool. Small (time) samples can be good, and they can be skewed. Large (time) samples tend to remove outliers, and give a better idea of the real error elipse. Time alone can allow enough change in environment, to allow a realistic sample.