Good questions Tony,
(1.) I am both a little bored and a little drunk... but, hey, it's 39 degrees Celsius here in Perth today and there is stuff all else to do (kite-wise) because of the weather system.
(2.a.) The filtering programs do not make the measurements more accurate, the only way to do that is invest in more accurate equipment... and that is too expensive for recreational use.
(2.b.) The filtering software does/can, however, remove any data that is obviously erroneous... that is, a reading that would appear as a dramatic spike on a graph... once removed that data does not ill effect the other data in the set. That does have the effect of making the data set more accurate, but not the individual measurements themselves. It sorts the seed from the chaff, if you like.
(2.c.) The other thing the filtering software might do is give you a mean/average of the filtered data... that is, it gives you a precise answer from a group of readings... it will give you precision.
(2.et al.) This link gives a good definition/analogy of accuracy vs precision.
http://honolulu.hawaii.edu/distance/sci122/SciLab/L5/accprec.htmlWhat this analogy doesn't cover well is the issue of measurements that are not necessarily within a clearly defined field, like a dart board. Accuracy applied to GPS is more akin to throwing the board at the dart, to use the same analogy... now, obviously, sometimes you will miss completely and GPS does too. So filtering software removes those misses from the final equation. Still using the same analogy, for one position, if we have a group of readings we must overlay the area of the dartboard over the bullseye of each reading... so for two readings you will get some kind of figure "8" pattern and for five readings it will be something like a demented Olympic rings symbol. The area covered by the dartboards is your area of accuracy, the software will average those positions to give you a precise "mean position" and the accuracy is the whole area covered by the dartboards. (It's messy, I know.)
The Garmin Geko 201 manual states the accuracy as "<15m RMS", and it says "subject to accuracy degradation 100m 2DRMS under the U.S. DoD-imposed
Selective
Availability program." The
S.A. program has now been terminated, but don't expect full compliance with "<15m RMS" all the times, due to PRN outages/maintenance and DoD usage... yes, they still like to spy on people and blow sh1t up etc. So less than 15m RMS is the best you will ever get from one reading, and speed is calculated from at least two readings and a time factor... so less than 30m is the best you will get for a speed calculation based on only two readings. (Now that is purely accuracy not precision.) Introduce three readings and your error is now between less than 30 and less than 90m, dependant upon where you third reading is.... now add four, five, etc etc.
Like I said before,
Accuracy costs money, how precise can you live with?For a couple of hundred dollars, not thousands or tens of, you can have lots of fun zipping around and going fast and being able to say "I did XXknts within this much reasonable certainty." (Of course you can leave out the "reasonable certainty" clause because it is largely moot within our degree of fun... isn't it?
)
any mean of the measurements carries the sum of all inaccuracies within the set. So the larger the set the more precision we can get but at the cost of less accuracy. The accuracy remains inherent in the system, we can't change that. Precision we can manipulate by statistical analysis, we can do whatever we like to shape that aspect of the data.
