How To Get An Average

[PCFrog]

I read some place a few months back on how to get an average of GPS coordinates in order to accurately give coordinates when placing a cache. I unfortunately cannot locate this reference.

 

First off, my GPS unit does not do averaging so that’s not an option.

 

Is there a tool on a website that will allow me to enter the coordinates (3-5 times) and will give me an average coordinate to utilize for where the cache is?

 

Now the next important question, if there is a tool, is this a good way to get a more precise coordinate?

[+Renegade Knight]

The best way is to have a good constelation of satalites. If you own a magellan let it sit in one spot for a while before you take a coordinate (though magellans tend to average automaticly).

 

If you have an eTrex make sure you have a good lock then take your coordinate. You should be good to go. If you don't trust that walk away come back, take anohter and repeat from a few different directions. Then use a calculator and average them together.

[Keystone]

My post in the Frequently Asked Questions Thread, pinned at the top of this forum, offers some advice on this.

 

I don't know of a computer tool that does the averaging. If the numbers are so far apart that you can't average them easily in your head or with a few calculator button pushes, you've got problems!

[+DaveA]

First off, my GPS unit does not do averaging so that’s not an option.

what unit do you have?

[+Prime Suspect]

Unless you've got a whole day (or two) to dedicate to it, don't bother. You're not going to significantly improve the accuracy, and could actually make it worse. Interesting reading.

[+Thot]

I wrote a utility for this. It is one of three utilities in the Geocaching Combo set. It's called (surprise) Averaging Coordinates -- Click Here.

[+briansnat]

I personally find averaging to a waste of time. I averaged for my first few caches and haven't for my last 110 or so. The coords were no better for the caches that I averaged for than they were for the unaveraged ones.

 

First, averaging, unless its done for an extended time over a number of days, can actually make your coordinates worse. Second, even if it does make your coords better,it will only be by a matter of a few feet. Is the extra time worth 8 additonal feet of accuracy?

 

If you have a unit that auto averages, go for it. It won't hurt (much) and may help (a little), but if you have to manually take waypoints and average them, don't bother. As long as you have a good signal lock, you're set.

[+Thot]

First, averaging, unless its done for an extended time over a number of days, can actually make your coordinates worse.  Second, even if it does make your coords better,it will only be by a matter of a few feet.  Is the extra time worth 8 additonal feet of accuracy?[emphasis added]

I see this kind of comment made here frequently, but I don’t understand why people think that. From my Coordinate Averaging utility these are the coordinates I took for a cache I placed recently.

 

As you can see I took 5 readings (at two times). I used the procedure described in the utility documentation. For each I allowed my SporTrak Pro to average for one to two minutes, and longer if the last digit was still changing.

 

The distance between the most widely separated points is about 20 feet. If I had taken only one reading and left it at that, my posted coordinates could have been either of these two. It seems to me it’s hard to argue that either of those two results is as likely to be as accurate as the average of all 5. I probably threw out the outlier and got 470, 046 as the average of the 4 remaining.. This is about 17’ from the outlier.

 

The scatter in the above example is typical for what I get, but I sometimes get wider ranging outliers.

 

Here’s another example.

In this case the distance between the most widely separated points is about 35 feet. Again, it seems to me hard to argue that the outlier is as accurate as the cluster of points. In my opinion, if I had taken only one reading and it had been the most outlier I would have been much farther off than necessary.

 

Edited to add another example:

Edited May 25, 2005 by Thot

[+Prime Suspect]

First, averaging, unless its done for an extended time over a number of days, can actually make your coordinates worse.  Second, even if it does make your coords better,it will only be by a matter of a few feet.  Is the extra time worth 8 additonal feet of accuracy?[emphasis added]

I see this kind of comment made here frequently, but I don’t understand why people think that. From my Coordinate Averaging utility these are the coordinates I took for a cache I placed recently. [yada yada snipped]

You've made a few unwarranted assumptions, the worse being that simply averaging data will somehow improve it. If you're averaging bad data, it won't necessarily make it better.

 

Averaging harks back to the days when SA was turned on. SA introduces a random "fuzz factor" into the satellite signal. An averaging routine could actually have a slight positive effect, because of SA's random nature.

 

But SA is gone. If you're GPS is giving you a less than accurate reading, it's probably NOT caused by something of a mathematically random nature.

[+Thot]

You've made a few unwarranted assumptions, the worse being that simply averaging data will somehow improve it. If you're averaging bad data, it won't necessarily make it better.

If you think the outlier point in the set of measurements below is as likely to be near the correct location/reading as the rest of the cluster then I'm sure you can't be convinced taking multiple readings helps.

Edited May 26, 2005 by Thot

[+JohnnyVegas]

Every time I have found a cache listed in which the cache owner has stated that they took several coordinates and then average them out those caches have had very bad coordinates. Just take a reading and be done with it.

[+briansnat]

Every time I have found a cache listed in which the cache owner has stated that they took several coordinates and then average them out those caches have had very bad coordinates. Just take a reading and be done with it.

Of my 120 caches, I have 4 that have garnered a few complaints about coordinates. I averaged the coordinates for about my first 10 caches and haven't since. The one that has generated the most complaints was one that I averaged. So roughly 1 out of 10 (10%) of my averaged caches has bad coords, while three out of 110 (less than 3%) of my unaveraged caches had bad coords. In all of the cases the coords were good enough that the searchers were able to find the cache. I'm not trying to present this as a scientific sample. Just telln' y'all what my experience has been.

 

BTW, if you go through my cache's logs, you'll see many logs complimenting the accuracy of my coordinates.

[+KG7JE]

The combination I have will do the average automatically (GPS Tuner on my PPC). I can watch the scatter diagram as it is taking the readings. My conclusion is simple, if the dots are jumping all over the place with some of them 20-30 feet away from others then it doesn't really matter what number you pick. The average is as likely to be 20-30 feet off as any of the other points.

 

On the other hand, if all 100 points are within a 3 foot circle then the average is likely to be as close as any single other point.

 

So, do what is easiest for you. But be aware that if the numbers are jumping a lot, then they might will be doing the same for the hunter that follows you.

 

My tightest scatter diagrams happen with the unit is using 6-8 sats and has a couple of others that it sees. Perhaps WAAS is also active.

[PCFrog]

Thanks for the information that all you have provided.

 

This one the reasons I like forums they let me get some information from a wide source quickly.

Edited May 27, 2005 by pcfrog