No question here, just a discussion.

I 'm currently toying with a GPS unit, through an Arduino Duemilanove.

I read its datasheet and manual, and I noticed the Datum parameter, which reconciles the GPS measurements with map topography of a certain area.
From what I gathered, a map's Datum is "linearizing" the flat map over a certain area, to minimize the fault due to the earth's spherical shape.

So I though it would be a good idea to change the GPS's Datum to Greece. BAM! 40m error on the measurements. I returned to the default Datum and I hardly get any error, maybe the expected 1.8m one.

That makes me think, what's the point of the Datum parameter for my GPS unit?

 

There is a "Greece" datum??

Datums should be WGS84, NAD27, NAD83, OSGB36 etc.

 

How much is this introducing an error?

from wiki WGS-84
Longitudes on WGS 84

WGS 84 uses the IERS Reference Meridian as defined by the Bureau International de l'Heure, which was defined by compilation of star observations in different countries. The mean of this data caused a shift of about 100 metres east away from the Prime Meridian at Greenwich, UK.

Also, as the agreed upon standards shift due to improved measurements, we will shift again.

There is a North American Datum 1927 ... as you indicated. There are datums all over the world.

Is there a known datum mark near you where you can visit with your GPS? I'd say the airport but their reference is typically the center of the runway .... and you'd look pretty funny dodging jets.

 

@spinnaker

Appendix A of this pdf says there is.

@JoeJester

I 'm not sure I get what you 're trying to say. I don't think the modern GPS system uses a 1927 standard.

 

The datum is actually European 1950 or ED 50.


http://en.wikipedia.org/wiki/ED50


As I suspected, as you move further from your location the difference will get greater. So it is really just coincidence that it is only 40 meters off at your location.


You would need to check the map you are using to determine which Datum to use.

 

BTW a position error of 40 meters is a fairly significant error.

 

The process I followed was the following:

I set my GPS unit to Greek Datum #62. I read the NMEA coordinates, from both this custom viewer and the proprietary free viewer from Fastrax and they gave the same coordinates, 30-40m NE from my location, using Goolge maps and also the GPS map from the Fastrax viewer. The second one is a different map from the one Google uses and resembles a GPS one, with street numbers and traffic lights.

Then, I reset the GPS unit to default settings and it returned my location spot-on.

I still don't know how the system works here. I think the devices' default Datum is WGS84 but I cannot confirm it.

 

BTW a position error of 40 meters is a fairly significant error.

I know, that's why I 'm trying to account it to a bad axis reference.

 

The process I followed was the following:

I set my GPS unit to Greek Datum #62. I read the NMEA coordinates, from both this custom viewer and the proprietary free viewer from Fastrax and they gave the same coordinates, 30-40m NE from my location, using Goolge maps and also the GPS map from the Fastrax viewer. The second one is a different map from the one Google uses and resembles a GPS one, with street numbers and traffic lights.

Then, I reset the GPS unit to default settings and it returned my location spot-on.

I still don't know how the system works here. I think the devices' default Datum is WGS84 but I cannot confirm it.

Your location is not going to matter. What matters is what maps you use. The map should list what datum it uses.

Some software like Mapsource will allow you to change the datum.

The other setting you need to be concerned about it the grid. This is how the position it self is represented.

Three common grids are:

degrees decimal
degrees / minutes decimal
degrees / minutes / seconds


More than likely the default datum for your GPS is WGS84. This is now pretty much the international standard. Google Earth uses this datum.

 

I see. I assumed that Google Earth and the Fastrax maps were switching Datums depending on the area the map was drawn.

It makes sense that there would be deviations if the map used one and one Datum only for the whole globe.

Thanks!

 

How long did you let it run after changing the datum? Most GPS units store a lot of averaged data and use that to process new satellite data. If you change one factor it can make some funny results until enough data is averaged and it settles down.

 

I was sure to unpower and restart the GPS again. On power down it returns to factory settings.

 

I see. I assumed that Google Earth and the Fastrax maps were switching Datums depending on the area the map was drawn.

It makes sense that there would be deviations if the map used one and one Datum only for the whole globe.

Thanks!

NOTE: This is a rather naive response coming from someone that is only passingly familiar with this stuff from limited exposure many years ago, so take what I say with a big grain of salt.

At one time datums were designed to be accurate to a desired level within a specific area and then were allowed to be whatever they were outside that area. But at some point it was realized that we needed to be able to use a single datum that had a specified level of accuracy over the entire world. All a datum really is is a parametric model of the shape of "sea level" (or possibly some other reference surface). I believe the WGS84 models the shape of this surface as an oblate spheroid with the parameters chosen to achieve minimum RMS error over the entire surface, though it is very possible that the errors are geographically weighted in order to reduce errors even further of regions of interest at the expense of accepting greater errors elsewhere.

 

I 'm not sure I get what you 're trying to say. I don't think the modern GPS system uses a 1927 standard.

No ... the NAD 1927 is a datum, just like WGS-84 is a datum. Why can't Greece have their own datum. After all WGS is an agreement between all the signatory nations on the location of the prime meridian ... and it's "only" about 100 meters east of where we were taught the prime meridian is located.

Whenever you see a latitude/longitude published, it's published based on a datum. From that information you can vary ... as you did when you shifted to the Greece datum ... without the aid of movement. The inverse is equally true.

Also, as the human's ability to measure improves, the datum's prime meridian will shift.

The positional shift you observed, while being stationary, was all about the reference datum you used. WGS-84 replaced WGS-72.

Suppose you found some coordinates in a eighty year old document. Using GPS' WGS-84 would get you close. Greece datum could get you closer, assuming the coordinates were in Greece. Still you would be on a good search for the location in the document.

Using just the lat/long, I managed to find the location of the Loran transmitting locations in Southeast Asia that were built in the 60s and abandoned in the 1970s. The reminents of the buildings and other structures are still visible on google earth. Some of the short lived WWII island hopping units are more difficult to determine, however the one near Bonavista Canada is visible, mostly the fuel tank cradles.

 

The history of standards and conventions is fascinating. Do you think your post holds effectively lost knowledge right there or is this common knowledge in topography?

 

When I was looking into map projections about a decade ago to better prepare of some short classes on map reading that I was going to be teaching to CAP cadets, I came across the fascinating history of datums and some of the practical and some of the political drivers behind how it evolbed. While I'm sure a number of peices of valuable information have been lost to time, there is a lot of stuff still available and out there.

On a somewhat related note, the book "Longitude", which is a short book easily readable in one sitting, tells the fascinating tale of the development of a means of determining a ship's within some prescribed tolerance. It reveals not only the practical importance of this, but ties it into many historical events including the Bounty and Cook and piracy in general and political intrigues between England and France. It also talks about the many different approaches that were proposed, both serious efforts and snake oil proposals.

 

Longitude ... I hadn't read the book but the movie was quite interesting.

In the early days of NAVSTAR, when there were few satellites up (circa 1978/1979), there were enough satellites in view once every 12 hours. It took days to get a sub-meter fix. That accuracy was needed for a project to measure overland phase shift in a signal.

That marker resides in a town in Southern Italy that has over 180 degree view of the Med and is approximately 38° 2'18.90"N 16° 5'16.20"E. The thing I remember the most about that location is thei local wine tasted very good ... and the ride back down to SS106 along the coast,was exciting as guard rails didn't exist on the road and there was some serious drops.

 

I was sure to unpower and restart the GPS again. On power down it returns to factory settings.

Not all GPS are that simple. How many minutes did it take to get the new fix after power up? Normally they have flash memory that retains data from the last fix, so "power up" is faster.

It can take maybe 15- 20 minutes on some models to lock in to a totally new location.

I'm not saying that is absolutely the cause of your error issue, but please be aware that just powering down most GPS units won't completely reset their fix.

 

I would think that the fix from the satellites is independent of the datum used. The datum is used to translate from a generic x,y,z position fix relative to the satellites to a geographical fix in terms of the chosen datum. If you change the datum, you should see an immediate change in the reported location.

 

The manual states that:

4.3 Reset
Reset can be initiated by switching off VDD supply for >150 ms.

I 'll interpret this as cold reset, as there is a separate power pin for a backup battery, in order to maintain the flash data on power down.

The fix takes less than two minutes in the balcony of my 6th floor penthouse.

The fix problems have been dealt with now. Google maps uses WGS84, my receiver uses WGS84, and the world is green. The measurement error is now never more than a few meters.

I have even set DGPS tracking via EGNOS satelites.

However, I have the impression that the initial measurement is less than a meter from my actual location (very good accuracy) but then it progressively drifts towards one side over the course of an hour, reaching up to 10m W-SW.

I 'll investigate further on this.