GPS accuracy

[phiggins]

To check the the accuracy of my GPS I created a chart of the area around my house from GE using GE2KAP then displayed the chart on OpenCPN. It first showed the position of the GPS 3 meters away from its actual location. I left it for a while with the track enabled to see how it moved around. After a few hours OpenCPN created an oval pattern (see attachment) with the center of the oval at the actual GPS position.

My question is what does the pattern represent?

[transmitterdan]

It probably doesn't mean too much. The shape could be due to phase noise in your GPS receiver or some other issue. Does the GPS receiver have wide area augmentation system (WAAS) capability and is it turned on?

What are you specifically concerned about?

[phiggins]

Not sure about the WAAS its just one of those small BU-353S4 pucks.

I am just wondering why OpenCPN is creating this highlighted pattern. Its actually a good way of checking the accuracy of the GPS. if the center is where the GPS actually is after a few hours you can be sure its accurate.

[hpeer]

Atmospheric changes cpeffect GPS accuracy. There are fixed sites that measure the differential from what the known signal should be and send out a differential signal for correction. But you do not experience exactly the same atmospheric change, so there is a little error still there.

I have done the same thing away from a differential signal and have gotten a much larger pattern.

The take away is usually GPS is very accurate, but there can be errors, sometimes significant.

Oops, above exchange appeared as I posted.

Correct, taking a signal over time assures accuracy. Don't know about Open CPN but many systems have a way of using time to calculate your likely error. Probably just as you say.

That said, the bigger issue is that charts are not always update to the same coordinate system as the GPS. in most U.S. waters it is not much of an issue. However I have seen smart/GPS discrepancies approaching a half mile. My track showed me sailing through a 500' headland. It would have been funny except for the pea soup fog.

Nigel Calder has an excellent book that deals with all of this nicely. How To Read a Nautical Chart.

Oops, above exchange showed up after I posted. Yes time averaging increases accuracy. Not familiar with CPN, probably just as you say.

[CarCode]

The GPS satellits are not fixed at the sky. They are always changing from your point of view and traveling. So there are sometimes better and stronger signals and sometimes less. I have remarked the more satellits are used for a fix the better is the accuracy of the fix. I guess the minimum is 3 sats and the maximum is 12 sats.
You may use the dashboard plugin to check this. The "GPS Status" item might help. However the "Sats in View" item is useless because the number shown has nothing to do with the sats number used for a fix. This number is in field 7 of GGA sentence which is sadly not used by the dashboard plugin.

Gerhard

[Snowpetrel]

Look up HDOP and UERE for a basic starting point on GPS accuracy and errors.

Sent from my HTC_0PCV2 using Cruisers Sailing Forum mobile app

[P_Dub]

Originally Posted by phiggins

To check the the accuracy of my GPS I created a chart of the area around my house from GE using GE2KAP then displayed the chart on OpenCPN. It first showed the position of the GPS 3 meters away from its actual location. I left it for a while with the track enabled to see how it moved around. After a few hours OpenCPN created an oval pattern (see attachment) with the center of the oval at the actual GPS position.

My question is what does the pattern represent?

Paul, are you aware of the fact you have trees in the way? this may affect partly the accuracy of the position. Secondly, for the position accuracy of the antenna must be, as much as possible, in the open sky.
Regards

[bcn]

Looks that what we see is the track over time filling up every free space due to the stochastic character of the errors.
Will you get the value for the real center or the real position over time?
You will get quite close. How much? Depends on the receiver and location but 1m should be doable.
Try to run the procedure several times and look for dispersion of the center.

What is very surprising is the sharp delimitation. I would expect to see spikes going beyond the borders. Few given the size of the pattern.
Your GPS might use some filtering techniques while stationary.

The specs of the GPS chip sets are working with a "CEP"number to define the precision. Indicates a circle of (non)precision it will achieve 90% of the time.
3m is here a common value. Using SBAS (WASS, EGNOS...) 1.8m is normal.
With differential corrections in real time and receivers using the L1 and L2 frequencies values of 1cm precision for a moving object are feasible.

[CarCode]

Originally Posted by Snowpetrel

Look up HDOP and UERE for a basic starting point on GPS accuracy and errors.

Correct, the graphic display of sats in the "GPS Status" item of dashboard plugin gives a hint about HDOP.

Gerhard

[phiggins]

I am aware of the many things that can affect the GPS accuracy. But what I really want to know is what the highlighted area created over a few hours by OpenCPN represents. I assume it has taken the limits of the track points and produced the pattern. The actual track produced after two hours has only two points, I assume OpenCPN has pruned away most of the points.

I guess the bigger the oval the less accurate the GPS is. Mine is about 18 meters across at the widest side and 14 meters at the shortest. but that the center is right where the GPS is an important fact.

BTW the other thing this test showed me is that the chart produced from GE for my area is very accurate.

[DotDun]

Originally Posted by phiggins

To check the the accuracy of my GPS I created a chart of the area around my house from GE using GE2KAP then displayed the chart on OpenCPN. It first showed the position of the GPS 3 meters away from its actual location. I left it for a while with the track enabled to see how it moved around. After a few hours OpenCPN created an oval pattern (see attachment) with the center of the oval at the actual GPS position.

My question is what does the pattern represent?

1) It appears that OCPN is converting the HDOP value produced by the GPS into an uncertainty area and displaying it. It's basically stating you are within the displayed area.

2) You must also realize that map/chart registration affects the display. As charts are drawn and satellite pictures are created, they must be scaled/aspect ratioed accurately and then register (anchor) them to the geo frame. When people report their chartplotter is showing them 500' from where they actually are, this registration process is normally the largest contributor to that error.

3) GPS with a clear view of the sky/horizon is typically accurate within 2-3 meters, less accurate with obstructions blocking it's view.

[phiggins]

This is what that pattern looks like when I first turn on tracking. Its a circle about 15 meters in diameter. Can't find anything in the Options that sets this value.

[bcn]

If you turn off "highlight track"?

[Snowpetrel]

Ha, I wonder if that is just the thickness of the track line, shown big when you Zoom right in?

Sent from my HTC_0PCV2 using Cruisers Sailing Forum mobile app

[Jon Hacking]

You're absolutely right that GE (& GE2KAP) produce EXTREMELY accurate charts. Not sure how GE gets such excellent geo-referencing, but it usually seems at least as accurate as our WAAS GPSs. Makes using GE2KAP a no-brainer, especially here in Indonesia.

But I'm not sure about the oval. What's its orientation? If it's NS (or EW) then I'd suspect a (normal) satellite anomaly.

Or could it be just interference from your house? It seems well aligned to your house.

BTW, I ran a similar test of the old Navy Transit (SatNav) system in the mid-80s. The average was pretty accurate, but outliers were up to 400m away. But that was in Venezuela, so pretty close to the equator (those satellites were in polar orbits). Not sure about the orbits of the GPS birds...