GPS accuracy

[phiggins]

After the results of my previous test where the track on OpenCPN showed me that sometimes the GPS was off by 9 meters, I wondered how often the GPS was off that much.

So I wrote a program to analyze the GPS RMC messages and plot how far they were away from the actual GPS position. It plots the number <1 meter, <3 meters, <5, <10, <30 and anything > 30.

Below is the track produced over a 20 hour period and the output of my GPSTest program for the same period.

[phiggins]

I cut off the totals in the previous screenshot. Here is what they should be

[bcn]

Originally Posted by phiggins

After the results of my previous test where the track on OpenCPN showed me that sometimes the GPS was off by 9 meters, I wondered how often the GPS was off that much.

So I wrote a program to analyze the GPS RMC messages and plot how far they were away from the actual GPS position. It plots the number <1 meter, <3 meters, <5, <10, <30 and anything > 30.

Below is the track produced over a 20 hour period and the output of my GPSTest program for the same period.

If you run this statistics during one month on a daily base you will get something like a GPS "weather report".

Some more material about the issue:
GPS.gov: GPS Accuracy

and a report for June 2015
http://www.nstb.tc.faa.gov/reports/R.../2015%2006.pdf

[Snowpetrel]

Originally Posted by phiggins

I cut off the totals in the previous screenshot. Here is what they should be

Nice work. It would be interesting to correlate the results with HDOP values at the same time to see if the bigger errors were due to geometry or other issues.

Does the antenna have a clear view of the sky?

Is it running waas? Cheers

Sent from my HTC_0PCV2 using Cruisers Sailing Forum mobile app

[phiggins]

No as I pointed out before the GPS is in the house and close to high voltage power lines. Would like to try it out on my sailboat but its in Mexico.

The specs for the BU353 S4 says it supports WAAS but not sure its using it.

[DotDun]

Originally Posted by phiggins

After the results of my previous test where the track on OpenCPN showed me that sometimes the GPS was off by 9 meters, I wondered how often the GPS was off that much.

So I wrote a program to analyze the GPS RMC messages and plot how far they were away from the actual GPS position. It plots the number <1 meter, <3 meters, <5, <10, <30 and anything > 30.

Below is the track produced over a 20 hour period and the output of my GPSTest program for the same period.

Note that "actual GPS position" is subjective. Your reference, Google Earth, includes errors. So your "9 meters" could actually be 6, or 3, or 12.

Google Earth has errors, the process you used to import it to OCPN has errors, it all adds up!

The good news is for the application of navigating a boat in open water, GPS serves the purpose.

[CarCode]

Originally Posted by DotDun

Note that "actual GPS position" is subjective. Your reference, Google Earth, includes errors. So your "9 meters" could actually be 6, or 3, or 12.

Google Earth has errors, the process you used to import it to OCPN has errors, it all adds up!

The good news is for the application of navigating a boat in open water, GPS serves the purpose.

But that was not the question. The question is that the fix shifts several meters during some time and why it does so. Paul has made some good observations for this. It has nothing to do with Google Earth errors nor with data imports to OCPN.

Gerhard

[zeehag]

speaking of the accuracy of gps.. i been watching as folks come into the lagoon here, which has notoriously shallow areas and shoaling. those who use the visual aids obvious to the naked eye make it thru unscathed. the folks using gps to guide them in, ALWAYS go aground some 20-30 yards off the channel and into the short water spots which are obvious with naked eyes.. has been most entertaining.
my spot will show me in a channel when i am in a marina slip, and it shows me on shoal s when i use it at anchor.
my garmin shows about 50 ft off, when i am in a slip it shows me in the casino they were building here at the posh hotel, which is duuh on land.. ok.. we all have known gps is not exactly accurate, and we have known for years this info. the best i have gotten as yet has been in the channel, 50 ft away from actual location, while at anchor.
the funnest one had me in ana bananas in la cruz when i was safely snugged into a summer slip for cane season 2012..... anas is about 50 yards from the slip i used in marina la cruz...

[transmitterdan]

Zee, it' not the GPS that's inaccurate. As someone else pointed out it's more likely the charts. Many charts in some parts of the world are not that accurate. Also, some people don't know how to set the datum of their chart plotter. When everything is off by 50 feet the first thing to check is the chart datum.

But you are right, using GPS to navigate a 50 foot wide channel with markers is just stupid.

[zeehag]

Originally Posted by transmitterdan

Zee, it' not the GPS that's inaccurate. As someone else pointed out it's more likely the charts. Many charts in some parts of the world are not that accurate. Also, some people don't know how to set the datum of their chart plotter. When everything is off by 50 feet the first thing to check is the chart datum.

But you are right, using GPS to navigate a 50 foot wide channel with markers is just stupid.

AS this is happening with NOT JUST my garmin, includes all other methods of positioning i have used since and including san diego, usa, i would say it IS the gps system giving that loss of accuracy.
remember this is a military thing, not meant for our use. as the military says, it will NOT be exactly accurate. i read a lot, and remember what i read.
some info says CANNOT GUARANTEE POSITION WITHIN 3 MILES.... this is not the manufacturing company stating these facts, it is the folks who program, place and maintain the satellites.
the numbers were a lot farther off in coronado, sin diego, kalifornikasha, than here in beautiful downtown rural fishing villageville, mexico. figger that one and try to splain it in grown up terms...SPOT, garmin, android x 3, computers x 2.....
channel into the lagoon is only 20 ft or less in width.

[DotDun]

Originally Posted by CarCode

But that was not the question. The question is that the fix shifts several meters during some time and why it does so. Paul has made some good observations for this. It has nothing to do with Google Earth errors nor with data imports to OCPN.

Gerhard

Hmm, I didn't see any question. I was attempting to explain that one cannot depend on Google Earth as the holy grail of references. The experiment has only determined the difference between the GPS measurements and the GE reference, it does not prove which is more accurate.

This whole thread is nothing more than the discovery and learning curve for some on the aspects of trilateration calculations on timing signals from satellites that are 12,500 miles above earth. Once you understand that, you gain an appreciation for the accuracy produced.

[phiggins]

DotDun,

To prove that the chart and the GPS (when it has a clear shot of the sky) are accurate, I put the GPS on top of my van created another chart of the parking lot a few blocks further from the power lines and drove over to the parking lot and parked right in the corner where I could measure the distance from the curbs.

As you can see from my attachments, OpenCPN shows it exactly in the correct spot and my GPStest never varied from <1 meter accuracy.

I know GE can be off in VERY rare circumstances but the charts produced by GE2KAP are accurate if the source is good. There are lot of sailors on the cruisers forum who depend on them in places where the charts are bad or non-existent.

[HappySeagull]

Jumps beyond the basic errors (How does the Global Positioning System (GPS) work ?) might be due to software. The rejection filter it uses is no doubt a compromise. Compare this to an automotive gps which seldom makes these jumps. One reason would be the software which mostly limits signals to a vector road map, but even off the road, they are generally "better" this way.
Again, I say Including heights ( a 3d table similar to your 2d ) is interesting..

Perhaps "dropping" one satellite and recalculating another sat' into the "matrix" gives the worst discrepancy.
A time scale with satellite azimuths might indicate when satellites are dropped and if this is why the worst jumps occur.

[DotDun]

Originally Posted by phiggins

DotDun,

To prove that the chart and the GPS (when it has a clear shot of the sky) are accurate, I put the GPS on top of my van created another chart of the parking lot a few blocks further from the power lines and drove over to the parking lot and parked right in the corner where I could measure the distance from the curbs.

As you can see from my attachments, OpenCPN shows it exactly in the correct spot and my GPStest never varied from <1 meter accuracy.

I know GE can be off in VERY rare circumstances but the charts produced by GE2KAP are accurate if the source is good. There are lot of sailors on the cruisers forum who depend on them in places where the charts are bad or non-existent.

What you have to realize is accuracy is important only to the use case. If it's accurate enough for the use case, then it's accurate. Your parking lot test only shows that the GPS can get you within the same parking spot as GE. If that is your definition of accurate so be it.

GPS.gov: GPS Accuracy

Try another test. Set your GPS on the ground at a corner of your house. Let it set for 1 hour. Record the lat/long. Move the GPS to the next corner of your house. Let it set for 1 hour. Record the lat/long. Find a distance between waypoints calculator on-line and plug in the 2 waypoints. Compare that number with the actual measurement between the 2 corners of your house using a tape measure. Would you use a GPS to lay out the foundation of a house at construction time?

Next, zoom in with GE as far as possible on the same corners of you house. Compare the waypoint reported by GE to what your GPS told you. There will be a difference.

My point is that GPS is very accurate for the application of marine navigation. What you have figured out is your GPS performs better with a clear view of the sky and GE is accurate enough for the applications you tested for.

[bcn]

Originally Posted by zeehag

AS this is happening with NOT JUST my garmin, includes all other methods of positioning i have used since and including san diego, usa, i would say it IS the gps system giving that loss of accuracy.
remember this is a military thing, not meant for our use. as the military says, it will NOT be exactly accurate. i read a lot, and remember what i read.
some info says CANNOT GUARANTEE POSITION WITHIN 3 MILES.... this is not the manufacturing company stating these facts, it is the folks who program, place and maintain the satellites.
the numbers were a lot farther off in coronado, sin diego, kalifornikasha, than here in beautiful downtown rural fishing villageville, mexico. figger that one and try to splain it in grown up terms...SPOT, garmin, android x 3, computers x 2.....
channel into the lagoon is only 20 ft or less in width.

Not correct.
Again here the basic values:
GPS.gov: GPS Accuracy

Availability (% <4.5m (historical 3- sigma)) 99.71%

Selective Availability (S/A) for civil GPS has been turned off many years ago.

Chart plotters being systematically off are result of a wrong datum of the underlying charts or errors geo-referencing charts (old material). That several charts/plotters for the same region are showing the error is simply due to the identical underlying chart material.

And for sure going GPS based into a 15m channel is just unreasonable.