Accuracy or otherwise of GPS

Bob,

Thanks for the response on WAAS.

Yes, we had a few pluggers at work under lock-and-key way back when. Far more accurate or timely than our other nav systems at the time.

We use only COTS GPS receivers now, many with RTK.

Night-and-day difference from the old days of dead-reckoning, OMEGA, LORAN, and occasional jumps when we got TRANSIT fixes.

But I really digress (and date myself).

— Rex

http://blog.oplopanax.ca/2012/11/your-gps-is-lying/

from a local SAR blog …

;)

Nearly everything that we've discussed here pertains to GPS accuracy, as in the accuracy of a lat/long display output. What has not been discussed is the accuracy of an underlying base map. Most modern GPS receivers have some kind of a base map, whether it is streets and roads, topography, ocean depths, airspace restrictions, or whatever.

I've seen many places where my GPS receiver would place me perfectly at some road intersection shown on the map. I've also seen many places where it was routinely off by 80-100 yards. This is not a "GPS accuracy" problem. It is a map registration accuracy problem. If my GPS receiver showed me various distances from the intersection and various directions N-S-E-W, then that would be a random error. But, there is a map registration problem in some areas, so the map appears to be shifted by that fixed distance from true. I also know this because there is a USGS survey marker in my neighborhood, and the small inaccuracies from perfect are listed online. When I go there with my GPS receiver and it agrees with the survey marker, and then when I go to the intersection and it shows me to be 85 yards off, then I know something is fishy in the map registration. The map datum might be off. And then, of course, some streets and roads and backpacker trails are changed and rerouted over time, and maps showing this are not immediately updated.

–B.G.–

Some of this is bad software. I'm a software engineer and repairing bad data is something I do for a living (for 10-15 year now).

There are no error bars anywhere for example. They could highlight areas in the route where the signal jumped rapidly.

If they KNOW you're on foot (which is true with some of the backpacking models) then you can't be going 25MPH… so if you jump a half mile all of a sudden and then jump back you can safely discard those data points.

Maps: I installed Garmin maps into my device which were made before the Tahoe rim trail was completed. So the TRT is not on my maps. More: the TRT joins together a lot of already existing trails, but it also establishes new trails and sections of trails. In winter, this has to be remembered. So far, not a big deal but it has led me down a now abandoned snow covered trail that had all of the hallmarks of being the 'correct' trail because of sawed logs, etc. On my gps map, it WAS the correct trail.

"but it has led me down a now abandoned snow covered trail…"

yeah, you just have to know that the USGS data a lot of topo maps are based on often has the trails at an old location. Many times re-routed or abandoned. Still useful though. Usually correct. Sometimes fun to investigate an abandoned trail.

Some of these newer maps like switchbacks.com or gpsfiledepot are free and is more likely to have the trails correctly placed.

"I'm a software engineer and repairing bad data is something I do for a living (for 10-15 year now)…"

What I hate is the odometer. Sometimes the stated mileage is larger than reality and sometimes smaller. It can be 10% or more.

Over stating mileage makes sense because of the error in position calculated. For example, if you were stationary, it would keep reading a slightly different position, and if it used these to calculate distance traveled, it would gradually increase even though you were sitting still.

I think some software engineer figured out to just ignore this if the position didn't change enough, because usually the odometer doesn't increase if I'm sitting still.

Under stating mileage has to just be a software bug. GPS error in calculating position can only make odometer bigger.

Maybe the software that ignores small position changes has a bug. If the new position is too close to the old position, it just ignores it, but it should keep the original old position. If it uses this new one as the old position for the next iteration, then you could lose small, real position changes so odometer could under state mileage.

If you understood that then…

I know little about GPS. Do they measure mileage? If they do, seems they would need to sample quite frequently unless the trail is perfectly straight. So much for battery life…

Ha, ha, ha Nick

Playing dumb just to entertain us : )

Samples few times a minute but you can set it to sample more or less often

Mine runs for maybe 20 hours on a set of batteries. That's like 40 miles for me. That's about as far as I go most trips but I take spare set of batteries.

When I learn how to program the remote control for the TV, I might try something complicated like a GPS. My wife gave me one a few years ago and just turning it on was difficult enough.

Once a GPS receiver has enough satellite data to get a position fix, it actually has a PVT fix. That stands for Position, Velocity, and Time. It can calculate new fixes pretty rapidly, but one of the lower priority functions is to update the display, so it might be slow to show you what it knows.

–B.G.–