Geocachers' Gpsrs

[+Kitty Hawk]

Only started reading this most interesting thread just now. I have a question, most likely aimed at Forester or Barry.

 

If the initial baseline was at Heathrow - what is the 'Liddington Baseline' near Swindon. I had understood this was the original, but perhaps I am wrong. My memory then goes on to say that all the trigs were all measured out using tapes and that the accuracy obtained using this method was no worse than 15 inches out.

 

Dino- irl Averaging.

 

My understanding is that you could walk right up to the coordinates. If you then put your gps down and came back 5 minutes later you'd probably find the gps had changed it's mind and the the spot was away off. If you let it a while longer, the gps may change it's mind some more. However, the longer it stays still the more readings it takes and the more accurate the measurements become, ie the longer it has stood still the more likely it is that it is right.

 

Stands back with interest and waits to be corrected on all aspects of this post.

[barryhunter]

the GPS lags behind your horizontal movement across the planet: standing still helps and some GPSs need several minutes

Can anyone explain why this is?

Not from knowledge, but I would say...

 

The satalite movement is predictable, as its continuouslly monitored and updated, and thefore can be figured in. Whereas the ground movement (by the user carrying the GPSr around) is not predictable and unknown.

 

I remember reading something somewhere along the line of that the GPSr software can spot you have stopped apprent movement, so can then start an automatic averageing process to give a better precieved position.

 

(so basically what Flyfishermanbob has just said )

[barryhunter]

If the initial baseline was at Heathrow - what is the 'Liddington Baseline' near Swindon. I had understood this was the original, but perhaps I am wrong.

 

I beleive the 'Liddington Baseline' was the baseline used for the Primary Re-Triangulation which was a line choosen because it was a line observed during the Principal Triangulation, which did use the Heathrow baseline. (see Timeline)

The Re-Triangulation then had a second baseline measured in Northern Scotland to check the accuacy of the network, and if memory serves the measured distance was within 24cm of the calculated distance, over a 20mile gap. (edited to add: the Ridgeway baseline, beucase it was actully very hilly and needed platforms etc building to bridge the gaps for a nice level line to measure the distance!)

 

My memory then goes on to say that all the trigs were all measured out using tapes and that the accuracy obtained using this method was no worse than 15 inches out.

trigs positions where all by observed lines (both optical and radio), a few where backed up with astronomical observations. However trigpoint heights where mostly (some used maths from the trigangulation to calculate hieght) measured by normal levelling which involved staffs and levels. AFAIK tapes where only used for measuring the few baselines

 

 

My understanding is that you could walk right up to the coordinates. If you then put your gps down and came back 5 minutes later you'd probably find the gps had changed it's mind and the the spot was away off. If you let it a while longer, the gps may change it's mind some more. However, the longer it stays still the more readings it takes and the more accurate the measurements become, ie the longer it has stood still the more likely it is that it is right.

That's my understanding of it (see my previous post), but not sure if all GPSr's do it, some have a specific averaging function. (see SAwatch - posted earier in this thread - for a screenshot map, to illistrate this.

Edited October 16, 2005 by barryhunter

[+Kitty Hawk]

Thanks Barry, hazy details are clearer now.

[+kewfriend]

the GPS lags behind your horizontal movement across the planet: standing still helps and some GPSs need several minutes 

 

Can anyone explain why this is?

 

This is nothing more than the time taken to compute and update displays. When the gps perceives you are static then it will continually improve and resolve its accuracy. In practice the effect will mean that your displayed coordinates on arrival at a cache will take a moment to 'catch up' - so allow for it.

[+Snosrap]

I seem to recall from a previous thread that there is a difference between how Magellan and Garmin units take 'average' readings.

 

When I save a waypoint on my Garmin V there is a separate function labelled 'average location' which then starts taking one second readings and updating the position from these with ever-improving accuracy until I say stop. I believe Magellan units somehow do this without user input but I'm not sure how.

[+kbootb]

the GPS lags behind your horizontal movement across the planet: standing still helps and some GPSs need several minutes 

 

Can anyone explain why this is?

 

This is nothing more than the time taken to compute and update displays. When the gps perceives you are static then it will continually improve and resolve its accuracy. In practice the effect will mean that your displayed coordinates on arrival at a cache will take a moment to 'catch up' - so allow for it.

That's why Marnie finds most of the caches. I'm standing still waiting for the GPS to settle, and she is off like a bloodhound using cacher's instinct to find the box.

 

Still. I get to go home with clean hands and no scratches.

Edited October 16, 2005 by kbootb

[+The Forester]

I've recalculated the spreadsheet, using the full precision co-ords kindly provided by Barry Hunter, and adding a few more logs.

 

There are now 77 valid logs, averaging 53° 04.1092'N 4° 04.5741'W, which is just 0.37m from the precise true value. That's almost exactly one foot. Amazing! That's actually less than the radius of the brass indicator plate which shows the directions and names of other mountains.

 

The average error is 3.28m and 57% of fixes are more accurate than that.

 

33.8% of fixes are within 2m

75.3% of fixes are within 4m

90.9% of fixes are within 6m

96.1% of fixes are within 8m

100% of fixes are within 10m

 

So, in summary, when someone asks you how accurate is a geocaching GPSr, a reasonably correct answer would be to say something like 3 or 4 metres.

 

Cheers, and thanks to Gamma Boo for creating this interesting dataset.

[+Haggis Hunter]

So, in summary, when someone asks you how accurate is a geocaching GPSr, a reasonably correct answer would be to say something like 3 or 4 metres.

I'll remember that the next time I am asked, I usually say about 5 metres but can be as accurate as 1.

[+The Forester]

Before I did that little exercise, I would have guessed that the averaged co-ords would be no better than 1 or 2 metres from the true value. I'm astonished at the actual accuracy which has been achieved by what is probably a very representative sample of 'typical' geocachers in typical conditions.

 

All that accuracy achieved by little gizmos which mostly cost one or two hundred Pounds or less! The first NavStar GPS receiver I used was in the early-mid 1980s. It cost $98,000, was the size of a picnic hamper, and because the satellite constellation was so incomplete you were lucky if you had coverage for a total of two or three hours day.

 

It's predecessor, the US Navy Transit system, was about the same size, though much cheaper, but it typically took two or three days of datalogging to obtain a single stationary fix to an accuracy of about a metre. Just processing the translocation data took 1½ hours!

 

For well under a hundred Quid you can now buy a GPSr which requires no skill to operate and takes only a few minutes for a raw beginner to start using. Amazing!

[+The Forester]

When I save a waypoint on my Garmin V there is a separate function labelled 'average location' which then starts taking one second readings and updating the position from these with ever-improving accuracy until I say stop. I believe Magellan units somehow do this without user input but I'm not sure how.

When the Magellan detects a speed of less than half a knot, it immediately goes into averaging mode, whether it's receiving WAAS data or not.

 

If you want to reset the averaging, so as to log a series of averages, just wave the GPSr about for a couple of seconds and it will come out of averaging mode until you leave it stationary again.

[+The Forester]

What does Airy's spheroid mean?

For co-ordinates to make any sense, they must be referenced to a stated spheroid.

 

The spheroid to which OS co-ordinates refer is the Airy 1849 one. The spheroid to which GPS co-ords refer is the WGS84 one. There are quite literally hundreds of spheroids to choose from.

 

People have been measuring the shape and size of the Earth for many hundreds of years. The first was Eratosthenes who measured the sun angle into two wells in Egypt. He then measured the distance between the two wells and calculated the distance from the surface of the two wells to the intersection point of the extended vertical centreline axes of the two wells, which he hypothesised would meet at the centre of the Earth. From this he calculated the radius of the Earth which he presumed to be spherical.

 

We now know that the Earth approximates to a spheroid (also known as an ellipsod as it is a ellipse of rotation), with a fatter radius at the equator than at the poles. Every geodesist who has devised a national mapping agency's standard spheroid and geodetic datum has measured the spheroid for himself by astronomical observations and has optimised the dimensions of his adopted spheroid to make a best approximation over as much of his area of interest as possible. That's what Sir George Airy did in 1847 for mapping the UK. In the US they adopted Clarke's 1866 spheroid and applied their North Ameriscan Datum 1927 (NAD27) to it. In post-war Europe a Europe wide datum called ED50 was applied to Hayford's "International" spheroid of 1909.

 

When tourists visit the Greenwich observatory, many of them like to photograph eachother standing astride a line in the ground which marks the Grenwich Meridian (0° Longitude). Some who expect to see 0° on their GPSrs are puzzled why it shows 100 metres or 5 metres or some other distance East or West. That's because they have selected the wrong datum/and spheroid in their GPSr and is further complicated by the fact that the OS has actually adopted the transit (a single axis telescope/theodolite) next door, some 5 metres away as the 0° meridian for OS work.

 

Selecting the appropriate spheroid in a GPSr is essential to make sense of the coordinates!

 

Cheers, The Forester

[+Snosrap]

Before I did that little exercise, I would have guessed that the averaged co-ords would be no better than 1 or 2 metres from the true value. I'm astonished at the actual accuracy which has been achieved by what is probably a very representative sample of 'typical' geocachers in typical conditions.

 

 

I'm surprised you're surprised by this result Forester. I would have thought that given a large enough number of readings taken over a significant period of time the 'average' inaccuracy overall would tend towards zero. In fact it would not surprise me if the average co-ordinates you have calculated are more accurate that the OS ones.

 

As far as the practical value of these readings is concerned I think it gives us 2 vital pointers: 1. on any given reading you've only got a 50% chance of getting accuracy of better then around 3m; and 2. The more readings you take and the longer the period of time between them the more accurate they are likely to be when averaged.

[+The Forester]

In fact it would not surprise me if the average co-ordinates you have calculated are more accurate that the OS ones.

Yes, I'm beginning to think that they might be.

 

After my first computation, which was based on OS co-ords which had an unknown rounding error which might have been anything up to 0.5m in either/both Eastings and Northings, I said:

I suspect that the accuracy of our averaged fix is probably better than the accuracy of the co-ords I adopted in this exercise.

 

If the rounding error had been 50cm in each axis then it could have been as much as 0.71m Given that our averaged co-ords were 0.38m from the rounded figure of the given OS co-ords, clearly there was a possibility that the geocachers' co-ords were actually better than the OS ones.

 

Once I had the full precision co-ords, I found that the actual rounding error had been 36.5cm, which is very close to the error which was computed from the slightly smaller sample and with the rounded co-ords as the true value.

 

I'm surprised to learn that the Snowdon trig is only third order. Perhaps the OS decided that as it is 'muggled' so regularly that they don't want to have to occupy it any more than was necessary to create the thing in the first place.

 

What's the accuracy of the trig co-ords? I dunno. Third order is normally the standard applied to surveying prominent landmarks such as church steeples and lighthouses etc. You would normally only bother with 4 rounds of angles with the theodolite (16 rounds for second order work and 64 rounds for first order) and would look for angular accuracy of a bit better than 5". Given that there's a first order station only about half a mile away, and knowing thet the OS tends to work up to a standard when doing their triangulations rather than down to a lower standard, I would expect that the accuracy of our given precise co-ords (the ones supplied to/by Barry Hunter) are probably good to a decimetre or better, but I can't be sure without inspecting the original Surveyor's notes.

 

on any given reading you've only got a 50% chance of getting accuracy of better then around 3m

You've got a 56% chance of getting better accuracy than 3m, on this sample (in other words, you will probably get better accuracy than that)-- and a much better than 95% chance of better than 3m if you are using WAAS.

 

The more readings you take and the longer the period of time between them the more accurate they are likely to be when averaged.

Yes.

 

I would have thought that given a large enough number of readings taken over a significant period of time the 'average' inaccuracy overall would tend towards zero.

Yes, but it will never reach zero and the decay rate is very slow at the low end of the quasi-exponential decay. Here's a graph of three sets of 86,400 fixes with a non-WAAS GPSr

 

Edited October 20, 2005 by The Forester

[barryhunter]

 

I'm surprised to learn that the Snowdon trig is only third order. Perhaps the OS decided that as it is 'muggled' so regularly that they don't want to have to occupy it any more than was necessary to create the thing in the first place.

 

What's the accuracy of the trig co-ords? I dunno. Third order is normally the standard applied to surveying prominent landmarks such as church steeples and lighthouses etc. You would normally only bother with 4 rounds of angles with the theodolite (16 rounds for second order work and 64 rounds for first order) and would look for angular accuracy of a bit better than 5". Given that there's a first order station only about half a mile away, and knowing thet the OS tends to work up to a standard when doing their triangulations rather than down to a lower standard, I would expect that the accuracy of our given precise co-ords (the ones supplied to/by Barry Hunter) are probably good to a decimetre or better, but I can't be sure without inspecting the original Surveyor's notes.

It's my understanding that Snowdon didnt even have a trigpoint early on, they choose the nearby Garnedd Ugain instead, to lessen vandalism. They only added the trig later (hence the high Flush Bracket number), beucase of of public constantly asking why Snowdon had no trig. I mentioned somewhere that the trig may of been replaced, which I think was due to detriouation, but will check with the OS.

 

As for the source of my coords, its my understanding that they are corrected values, where the whole network was updated in about '79 with new technology.

 

If the coordinates where converted by using the definitive transformation from the OS (which I assume you did), then the accuacy should be very good. They quote a 7cm accuacy (i think), becuase the transformation is defined using the Passive GPS network, which closely matches OSGB36 to ESRS89 (effectivly WGS84). [or something to that effect I dont fully understand the full terms!]

[+Teasel]

In fact it would not surprise me if the average co-ordinates you have calculated are more accurate that the OS ones.

Yes, I'm beginning to think that they might be.

<snip>

I would expect that the accuracy of our given precise co-ords (the ones supplied to/by Barry Hunter) are probably good to a decimetre or better

Now hang on a minute! In your first posting, you said that your sample standard deviation was 3.41m, and that you had 74 samples. So that makes me 95% certain that your sample mean is within 1.10m of the "actual" position, and 99% sure that it's within 1.44m.

 

So, if you are telling me that the Ordnance Survey are, say, 95% sure that their figures are within 10cm of the "actual" position, then I'm afraid they win hands down!

 

And that's before we worry that we have implicitly assumed that all of our uncertainties are due to random error, and that there are no systematic errors in our GPS measurements. Presumably the OS include both types of error in their quoted accuracies.

 

Above you say that you think your average is within 0.37m of the true value. I'm only 53% sure that it's not just a fluke!

Edited October 20, 2005 by Teasel

[+The Northumbrian]

Having read most of this thread, not knowing what half the words mean. or even understanding it I only have one question to ask ,If I was to learn all about this , does it mean I can go geocaching without using the clue ?

Nige

[+John Stead]

If I was to learn all about this , does it mean I can go geocaching without using the clue ?

Nige

No, Nige, I think you would be able to find all caches without even reading the webpages!

Edited October 21, 2005 by John Stead

[+wildlifewriter]

Having read most of this thread, not knowing what half the words mean. or even understanding it I only have one question to ask ,If I was to learn all about this , does it mean I can go geocaching without using the clue ?

Nige

If you were to learn all about it, you might be as surprised at some of the assertions in this thread as I am...

 

... and it might cause some amusement down at the Ordnance Survey offices as well, were they to read it.

 

-Wlw.

[markandlynn]

does it mean I can go geocaching without using the clue ?

Nige

You should try that one day. We have. Just get the co ords (make sure they are traditionals ) in your GPS put it in nearest mode and drive. Good fun if your confident.

[+The Northumbrian]

I think I will carry on the way I have always done it, send Marg on ahead to find a safe route for me to follow ,and if anybody going to get wet feet or nettled, it wont be me I read a bit more of the thread this morning,( the long bits) And I have to admit, why do I need to read all this boring stuff when My little yellow etrex has never let me down in the past, the only time I never find a cache is cos I havn't looked hard enough or someones hidden it in a dark dense wood

Nige

[+Haggis Hunter]

Please take note everyone I am saying nothing

[Bassrock]

""Can anyone explain why this is? The GPS satellites are all in low orbits, so the relative velocities between each satellite and the GPSr will be quite large. How is it that the software on the GPSr can cope with satellites flying around the sky, but can't cope with the receiver crawling around the earth?"

 

My response:-

 

Altitude of orbits (low or high) is probably irrellevant in respect of relative velocity to us on ground, relative velocity is still fast/large whatever.. (as each satellite is 12,000 miles up travelling at 2000mph,) depends on how high and type of orbit.

 

Sensitive gps receivers will pick anything up these days, but the ionosphere, cloud, rain, ice and virtually every single factor at every altitude from the satellite to us, that can effect/influence a GPS signal will do so in some manner, whether delaying, attenuating or bending the signal. The software has been adapted over the years to cope with all this, so we joe public switch on and expect perfection. Thats OK if we are out in a field with clear sky everything should work. But we want to walk with it, cycle with it, drive with it, fly with it and some people even want to geocache with it !!.

 

As far as I am aware the antenna design on most GPS receivers dictates the antenna has to point up or towards the satellites (e.g. the antenna cannot "see" 360 degrees, maybe 180 at a push (I don't know). So if you drop it, cover it for a couple of seconds the GPS update rate of 1 second may not be able to cope with the outage and therefore give a jumpy/staggered type update on display.

Then don't forget as with all radio type signals there will be areas that just don't get a GPS signal or there may well be local interference of signal e.g. has anyone had problems picking up a signal when near 30Kv pilons ??

 

Don't know if I've answered the question, but at least it may stimulate discussion.

 

Bassrock

(PS clue to my location is in my name)