GPSr Accuracy test

[nikongod]

Hello all

 

I was pretty big into geocaching in 2002-2004, but the bug died off. This was well before any affordable cell-phones/tablets had a GPSr built in, so it was all dedicated GPSr. I recently got back into the hobby, and one of my first questions was "is my tablet/phone really good enough for geocaching?" A few searches on the forums came back some info from 2010-2012, but in terms of computers its about as relevant as something written about GPS in 2004...

 

Anyways.

 

I am interested in starting a "GPS accuracy test" for personal knowledge and the betterment of all mankind (lol, yea right).

 

My test would work something like this:

 

Place a virtual geocache somewhere in a pretty sizable city (Manhattan/NYC, in my case) and ideally away from signal-bounce-generators. Ideally on top of a hill or next to a river away from tall buildings or cliffs.

 

The coordinates given would only be approximate - about 50ft from the actual test site so that people could not just return my coordinates but the description would make the test site obvious. For example: there are 3 benches in a semi-circle - sit in the middle of the middle bench and test.

 

In order to claim the cache the finder would need to submit their best coordinates for the cache, how they were obtained (single point VS averaging), and what GPSr they were using. I would of course accept multiple points from the same person, and would hope that people would send multiple coordinates if they have multiple devices.

 

After that, put everything in a spreadsheet and graph away.

 

Has anyone done something like this recently?

Are there any glaring flaws in my plan?

Would anyone be willing to run a similar test in other cities?

[+geodarts]

I have seen an old virtual or two that require you to submit coordinates from your GPSr or measure it against a benchmark - but none that I can remember that are tied to the type of GPSR being used. I have compared readings of different models and phones to see the differences, and it is interesting. Unfortunately, I don't know on any way that such a thing can be listed on this site, given the prohibition against virtuals or ALRs, but perhaps there would be a way to do it through Waymarking.

Edited November 23, 2014 by geodarts

[nikongod]

I did not realize that you could no longer post Virtual/ALR caches. I'm only just getting back into this.

 

Any idea who I would need to petition for an exemption? Or a few exemptions, if I can find some people to help around the country/world?

 

I don't think Waymarking will generate enough samples for a meaningful data set. The popular waymarks in the area only seem to get hit every other month.

[+wmpastor]

Along the same lines, are there are locations available to the public where exact coordinates are published? People could go to such places and see what their own devices say.

 

(& where these coords were established by the best gpsrs available)

Edited November 24, 2014 by wmpastor

[+Walts Hunting]

You could put a cache on a benchmark and ask finders to take a reading and include it intheir log.

[+Gitchee-Gummee]

I would suggest the same as WH did... use a Benchmark. A point marked by trained professionals adhering to a strict set of "rules" dealing with marking that point.

 

But, let's also keep one major thing (flaw, if you wish) in mind. Consumer-grade GPSr devices are not constructed to deliver absolute pin-point accuracy. The thousands or millions of us wouldn't be able to afford such units. Expect nothing better (on average) of 20' degree of accuracy*.

 

*Edit for clarity: within about 20 feet (on average).

Edited November 24, 2014 by Gitchee-Gummee

[+DukeOfURL01]

If you use a benchmark to determine location, you must use one that the location is "corrected" and not "scaled" - in order to get a precise position. Scaled is not precise, as it was determined by somebody looking at a map 50 years ago. If corrected, it is by a professional surveyor using a professional surveying GPS, and then the coordinates will be to I think 8 places, which is way overkill for most people, but it will be in the Degrees Minutes Seconds coordinate system, and will need to be converted into the Degrees Decimal-minutes system that Geocachers use.

Edited November 24, 2014 by DukeOfURL01

[+SageTracey]

Why not host an event and have everyone attending log their coordinates at a specific location at the event - maybe place/tether the logbook for the duration of the event and make that your location.

[+Timpat]

We in upstate NY have been conducting just this sort of experiment at our annual local geocaching picnics, and last year at the Mega Berkshire BeoBash. A survey grade GPSr is used to set a control point for what we term a "closest to the pin" contest. Each hand held GPSr user gets a pin flag with their name on it. The precise coordinates are published and each user places their flag where they feel the control point is (which is driven below grade- not visible of course).

 

In the past we used different colored flags for different GPSr brands (Garmin, Magellan, Lowrance, etc) and it was interesting to note how each brand clustered. You could do this for your study and create the spreadsheet based of the user input.

[nikongod]

Thanks to everyone for input on this. I only quoted a few people below.

I wish there were a way for moderators/admins to override the ALR/Virtual cache thing but apparently not, from the email I got.

 

I agree that virtual caches were getting out of hand in the mid 2000s, and I can only imagine what it would be like now with the smartphone crowd, but it would be nice if they could make just 3 or 4...

 

You could put a cache on a benchmark and ask finders to take a reading and include it intheir log.

This was what I wanted to do. Not with a USGS benchmark, but with a specific rock. Sadly ALR & Virtuals are banned

 

I am not so concerned with absolute accuracy but how GPSr commonly used for Geocaching compare to each other.

 

Why not host an event and have everyone attending log their coordinates at a specific location at the event - maybe place/tether the logbook for the duration of the event and make that your location.

We in upstate NY have been conducting just this sort of experiment at our annual local geocaching picnics, and last year at the Mega Berkshire BeoBash. A survey grade GPSr is used to set a control point for what we term a "closest to the pin" contest. Each hand held GPSr user gets a pin flag with their name on it. The precise coordinates are published and each user places their flag where they feel the control point is (which is driven below grade- not visible of course).

 

In the past we used different colored flags for different GPSr brands (Garmin, Magellan, Lowrance, etc) and it was interesting to note how each brand clustered. You could do this for your study and create the spreadsheet based of the user input.

 

I nested your quotations, they go together so nicely.

 

I think the comment about how various brands of GPS "cluster" is very interesting, and something I am curious about.

 

I like the idea of going to a meet and having people give me data but I am concerned that at a smaller meet there wont be enough data to form any sort of meaningful consensus. A virtual cache in a large city could easily get hundreds or possibly thousands of hits! Lots of data! Om nom nom.

 

I am going to a meet in NYC this Saturday. Hopefully I can get a bunch of people there to play along.

 

I don't know if anyone is interested, but here is the data I have been gathering on my 2 (functional) GPSr units - a Google Nexus 7 (2013) and a Garmin Etrex-10

I wanted to see how the compared for consistency, and relative position.

As a "benchmark" the GPSrs were placed on a crack in my driveway and run simultaneously. The Etrex 10 has a few rows missing because I only got it a couple weeks ago.

The numbers shown are the last 3 digits of (degrees) (minutes).(3 digits)

 

The columns are SERIOUSLY shifted because I fail at the internet.

The left 2 columns of 3-digits are for my 2013 Nexus 7 tablet, waypoint averaging through some free app. I let it run a bit more than 100 samples each day.

The 2 right columns of 3-digits are for my Etrex-10. Waypoint averaging, all of the satellites & WASS turned on.

 

date N7 tablet Etrex 10 Etrex 10

2014-11-02 413 473

2014-11-06 412 463

2014-11-23 417 471 417 472

2014-11-24 414 474 418 470

2014-11-25 410 476 417 474

2014-11-26 420 468 417 471

2014-11-27 416 473 417 472

2014-11-30 419 474 418 472

2014-12-01 417 470 419 470

 

Average: 415.33 471.33 417.57 471.57

Standard Dev 3.34 2.75 0.79 1.40

 

As you can see they both return a pretty similar position for the long term average, but the ETREX 10 is much more consistent.

I don't have any way to measure it, and it may be due to the software I am using, but I feel that the Etrex-10 is also "faster" than the tablet. By that I mean that as I approach the cache, the Etrex-10 gives precise distance information down to a few ft, at a normal walking pace. With the tablet I find I often overshoot GZ by a pretty wide margin before it even drops into "fine tooth comb" range.

Edited December 2, 2014 by nikongod

[jimlarkey]

@nikongod, Sometime ago I had a surveyor from a neighboring constr proj plant a PK nail in my yard with claimed centimeter accuracy (cost me a 6-pack of IPA). Here's the result of logging 1070 data points over 18 hours, concurrently, comparing the "mother of all GPSrs", 60Csx to my Oregon 600 (which I finally dumped after 16 months).

 

You'll note that I used the track length as a measure of the scatter in the comparison. The 600 was tighter by 40%.

 

I'm suspicious of the accuracy of the WGS coords that the surveyor gave me. I've noticed the W'ly bias in subsequent tests. I've worked with surveyors in my previous lifetime, and know that errors can be made.

 

I am a believer in using benchmarks for valid positional (vert as well) comparisons. I've used county BMs with success. Not many USGS convenient for me.

 

Good luck with your tests, look forward to the results.

[+Zuuk]

I didn't do any "testing", but whenever I'm out geocaching, I use my Samsung Galaxy Mega simply for ease of use and on the spot logging. I always carry my GPS unit, which is an ETrex Venture HC, because sometimes the phone option won't work. As long as I'm in the open or light tree coverage, the phone works great. If I get into heavy overhead coverage, it starts to lose its accuracy. My GPS will work in the heavy coverage where the phone starts to fail. The phone also has another drawback, which is that if you get into a place where there is no internet coverage, the maps won't update if you try and zoom, move them around, etc. I like having the maps available, so my GPS unit works in those places.

 

As for accuracy, in the open I found no difference between the two. Recently, I relocated a cache for someone, and I got curious if my phone would be accurate enough to get the coordinates to send in. Both my phone and GPS showed the exact same coordinates. Earlier in the day, under some dense tree coverage, I had an 18 yard difference in my location between the two, and it was the phone that was off.

 

A bit off topic, but useful info for accuracy when you start thinking about signal strengths. GPS still wins.