How the Kiwi OSD unit determines coordinates from GPS

Aug 04, 2005

From a posting by Kiwi Geoff to the Kiwi-OSD group on Yahoogroups .

Thanks to the careful observations of Sam Herchak and resulting caveats for using the Garmin "18" (in averaging mode), I thought it would be constructive to provide some guidelines on how to use a GPS to give position coordinates.

The following comments are (for the most part) general in nature and apply to most consumer GPS receivers. However I will also detail how to use the information displayed by KIWI OSD to allow users to be more confident about obtaining improved accuracy of their geo coordinates.

VTI's (Video Time Inserters) do not turn your GPS into a survey grade "positioning" device. Their focus is timestamping video, and report (as a bonus) what the GPS currently says the geo position is.

I think an analogy is helpful. We are used to seeing the path maps created by WinOccult, where there is a path and 1 sigma lines. If we knew the exact position of the star, and exact shape and orbit of the asteroid, there would be no need for 1 sigma lines. We would know the perfect path of the asteroid shadow. The 1 sigma lines give us a measure of the "probability" that the shadow will be contained within a certain width.

When we switch on a handheld GPS, it gives the impression that it knows exactly where we are, the LCD display is like a calculator with lots of significant digits - that imply precision. However the displayed coordinates should not be thought of as an exact dot on a map, but simply the centre of a "sigma circle", and that your position is "probably" within that circle.

When KIWI OSD is first switched on, it examines the data from the GPS until it knows the 'time' (to within 1 millisecond). It then displays the coordinates and other data, and then the time until the user presses a switch - to invoke the last remaining integrity check. One more press of the button starts the process from scratch again.

After initial powering on, the OSD is only concerned with the relevant data pertaining to time display. So the initial coordinates displayed can have a large sigma circle, and that 1 millisecond of timing precision = 300km position precision.

So how can a user be more confident about knowing the size of their "sigma circle"?

After the coordinates are displayed another line of numbers is displayed, careful attention to the first three numbers tells us how big our sigma circle is.

The first number is "Fix Status".

0 = Fix not available
1 = Non Differential Fix available.
2 = Differential Fix available (i.e. WAAS or Egnos).
6 = Estimated position

The second number is "Number of satellites in fix".

The more the merrier, but it depends where they are. If they are clumped together in the sky, a high number can still give a large sigma circle.

The third number is "HDOP = Horizontal Dilution of Precision".

This tells us the relative "size" of the sigma circle, the smaller the number the better. Anything above 2.0 should be treated with suspicion. We are looking for a number as close as possible to 1.

In theory it can go to 0.9, but as long as it is close to 1 you can be confident of a small sigma circle.

Important things to remember:

For those interested in how complex position and averaging can be, here are some good links: