We’ve just completed our first year using GPS for flight verification. It’s been a steep learning curve, and a painful one for several individuals. A number of people lost points in France because of carelessness in flying through sectors, and some of the GPS turnpoints seemed to be in the wrong place relative to the ground features. One pilot in Wales lost his start gate because it seemed to be in a different place from everyone else’s - even though he had put the same numbers into his GPS.
How can this happen, using a system which is claimed to be accurate to within ten metres or so? To know the answer you need to understand the meaning of the Position Format and the Map Datum which you select in your GPS, and a little bit of background on map projection: the science of representing features from the curved surface of the earth on a flat piece of paper or GPS screen.
(To skip the technicalities, just read the summary and guidance notes at the bottom of the article.)
The position format simply defines the way you write down the position which the GPS has calculated. John always asks for positions in degrees, minutes, seconds and decimal seconds (or hddd°mm’ss.s”). You can also set your GPS to use degrees, minutes and decimal minutes (hddd°mm.mmm’), degrees and decimal degrees (hddd.ddddd°), and a number of regionally-specific grid systems: British National Grid, German, Taiwanese and so on.
You can switch between position formats with impunity. Think about the phrase "ten and a half minutes". You could also write this as "ten point five minutes", or "ten minutes thirty seconds". They all mean the same thing. In the same way, N54°45’0.0” W3°15’30.0” (hddd°mm’ss.s”) is the same as N54°45.0’ W3°15.5’ (hddd°mm.mmm’) and the same as N54.75° W3.25833° (hddd.ddddd°). (That last number looks a bit odd, but it’s correct. If you’ve got a calculator with degrees/minutes/seconds you’ll be able to work out why.) So long as you are using the same map datum, all these numbers relate to precisely the same place on the surface of the earth. If that map datum happens to be OSGB, then this position could also be expressed as NY 19017 40187 using the British National Grid. You can measure both the British National Grid reference and, with some difficulty, the longitude and latitude of a position from the marginal information
around the outside of any Ordnance Survey 1:50,000 map.
It should be obvious when somebody gives you a position in a different format to the one you have set on your GPS, because the numbers just won’t fit into the spaces available. You know then that you need to change to a different position format.
It is also obvious that you can' t use the British National Grid if you are in France. Your GPS will just show blank spaces instead of numbers. The British National Grid extends just far enough to cover Great Britain; beyond that it has no meaning. The same applies to other local formats.
This is trickier, and harder to tell if you get it wrong. In mapping terms, a datum is a fixed referencing system in relation to which the rest of the map is drawn. So that’s easy, isn’t it? Longitude and latitude are expressed in degrees, minutes and seconds east or west from the Greenwich Meridian, and north or south from the Equator, respectively. Sadly it isn’t that simple.
The problem relates to the difficulty of representing ("projecting") real-world objects from the curved surface of the earth onto a flat plane and, more fundamentally, to the fact that the earth isn’t a perfect sphere: it’s flattened at the poles and bulges at the equator. In fact it’s not even that simple, being a rather irregular lump of rock with odd lumps and hollows here and there. Unfortunately, to do the business of projection, the cartographers need to assume that the earth is a mathematically regular shape (an ellipsoid), because projection is a mathematical process which has to have a consistent basis. An ellipsoid is an artificial mathematical construct which is produced by a process of calculation. Many different ellipsoids have been calculated for different purposes, because an ellipsoid which fits well to the real world in one place may not
fit well in another.
Cartographers use whichever ellipsoid happens to fit best with the bit of the world they happen to be mapping. The British Ordnance Survey map datum (OSGB), for example, uses the Transverse Mercator Projection and the Airy Ellipsoid, because they give the best combination of fit and function for general purpose maps of Great Britain. The WGS84 map datum (the GPS default) uses its own GRS80 ellipsoid, because it needs a good general purpose fit for the whole globe. The Airy and GRS80 ellipsoids are significantly different in the region of Great Britain. This, in turn, means that a ground feature projected onto a map will have a different latitude and longitude depending on the datum which was used. In Llandinam, mid-Wales, the difference between OSGB and GS84 is about 480m. Elsewhere, the difference may be even larger.
The following illustration shows the variation in position of the horizontal benchmark (a surveyor’s reference point) at the Capitol Dome in Austin, Texas.
Using WGS84, the latitude and longitude of the feature are 30°16' 28.82" N, 97°44' 25.19" W. The red crosses show where you would end up if you entered exactl these same numbers into your GPS using a different map datum. For example, if you used the Ordnance Survey GB map datum, which would be pretty silly in Texas, you would find yourself about 560m south-west of the actual location of the benchmark.
You can try it yourself. Enter Waypoint A with this latitude and longitude into your GPS under the WGS84 map datum, then change to the OSGB map datum and enter Waypoint B with exactly the same co- ordinates. Now set up a route between A and B and check the distance between the two. Note that when you changed to the new datum, your GPS changed The displayed co-ordinates of Waypoint A so that it is still correct in the new datum. If you change back to WGS84 it will revert to the original numbers, and re- calculate Waypoint B’s displayed co-ordinates.
How Garmin GPS deal internally with position references
Garmin GPS units do all their collection, internal storage, processing, input and output of data using the WGS84 map datum, changing the numbers only for the purposes of user interface. This explains why the units sometimes try to round the numbers you enter, because it is trying to get them to fit into the WGS84 datum regardless of the datum displayed on the screen. There are some important things to remember here about exchanging data with other GPS or computers:
• Your GPS always uploads your task record to the scorer' s computer in WGS84, regardless of what map datum it is set to, so the scoring software needs to be aware of this fact;
• Your GPS always assumes that data downloaded to it (as in St André this year) are in WGS84, regardless of what map datum the data were actually surveyed in – so you' d better make sure they actually are WGS84;
• It doesn' t matter what the screen says when you survey a point (e.g. by standing in the middle of a prospective goalfield and pressing the "mark" button) – the data will always be stored internally correct to WGS84;
• BUT IT DOES MATTER if you read off the number displayed on your GPS, and then manually enter it into another GPS set to a different map datum – the position will then be wrong.
How these issues can affect us: three cases Wrongly located turnpoints
- In St André we were told that the turnpoint co-ordinates had been collected by a person physically standing at the feature and pressing the “mark” button on a GPS. The data were then delivered to us by download direct into our GPS. We know that GPS units both store and upload data in WGS84, and that they assume that downloaded data are also in WGS84. If, therefore, this had beena completely digital process from start to finish there should have been no problem. I am prepared to bet that at some stage in the process somebody wrote down the latitudes and longitudes from the survey GPS before they were entered into another unit which was set to a different datum. This might have happened, for example, when the data collected by several surveyors were collated. This would explain why the GPS waypoints appeared to be significantly removed relative to the actual ground features. It would be interesting to resurvey these points in a more organised way, and see whether the results are any better. Maybe we’ll get an opportunity next year.
- Missing the start gate. On the day we flew from Llandinam at the League final one pilot‘s GPS failed to record entering the start sector even though he flew to the GPS "virtual startgate" on his screen and pressed "mark". Now, John had taken the antenna’s grid reference from the OS 1:50,000 map, so it was correct to the OSGB datum. The pilot concerned entered the co-ordinates into his GPS while it was set to WGS84 datum. Therefore, when he flew to his personal start gate, he was actually about 480m south of the correct position, and this is the information it uploaded into the scorer’s computer. Sorry, Mike – nul points! Note: OS maps are high quality and the grid references are easy to read, so if you find yourself flying nearly half a kilometre away from the ground turnpoint in Britain it may be a clue that you' ve done something wrong!
- Mixing your datums : On the last day in Wales the task was Bache to Llandinam via Gladestry, with the race section starting at Gladestry church. Again, John took the grid reference of the church from the OS map, so it was correct to the OS datum. However, the goalfield co-ordinates were collected the previous day by a pilot standing in the field and pressing the “mark” button on his GPS. He then read off the numbers from his unit set to WGS84, but most of us entered them into our units set to OSGB. The numbers we were given were SO 02300 89900. This should first have been converted into OSGB, in which the numbers would be SO 02397 89429. Gladestry church is at SO 23050 55150 (again, to OSGB datum).
The effect here was that the scorer’s computer, and most people' s GPS, thought that the race leg was about 480m longer than it actually was. This will have had some interesting consequences:
- Final glide calculations made by GPS-linked flight computers will have been wrong (getting you to goal too high);
- Points allocated by the scoring system for the distance element of the task will be wrong;
- Pilots who landed short will have lost out. I landed at SO 05223 80477. This is actually 9.39 km from goal, but the scorer thinks I was 9.87 km short. At 3.7 points per kilometre for this task, I lost 1.8 points! I was robbed. In a higher scoring task, of course, the difference might be more significant.
All this applies in the specific case of that particular task. Of course, if the task had finished in a southerly,rather than a northerly, direction, the problems would have been reversed: your final glide calculations would have decked you before the goal line and those of us who landed short would have gained, not lost points.
The solution to this problem would have been to make sure that everyone' s GPS was set to the same datum at the briefing, including that of the pilot who did the goalfield survey. Since John had taken the Gladestry grid reference from the OS map, it would make sense to use the OSGB datum for the whole briefing
1. There is no such thing as an “absolute” location reference, even latitude/longitude. Any given
position reference (in latitude and longitude, British national grid or any other format) is valid only in terms of the particular map datum in which it was calculated.
2. Using the wrong map datum to find a position reference can result in errors up to 1km.
3. You can switch between position formats without danger – they are simply different ways of formatting the description of a single point on the earth’s surface and do not affect the internal GPS record of the position.
4. OS National Grid references physically read from an OS paper map must be entered into your GPS with the unit set to the OSGB map datum – or they’ll be wrong.
5. Data uploaded from a Garmin GPS are always sent in WGS84, regardless of what datum the unit is set to.
6. Data downloaded into a Garmin GPS must be in WGS84, because that is what the GPS unit
7. Position references entered in one map datum will be displayed differently if you reset your GPS to another map datum. The displayed figures will, however, still be correct in the context of the new datum – so if you "goto" the point you will still get to the right place.
8. Setting a route based on waypoints collected in different map datums is fine, so long as the waypoints themselves were entered into the GPS in the correct individual datums.
9. If any data are manually entered, whether on GPS or computer, or written down, or given over the phone, or read off the GPS screen and told to someone else, then the correct datum must also be used or given – or errors will result.
Things for the Meet Director to remember:
• Brief the appropriate datum, make sure that people set it and realise how important it is;
• Use OSGB in Britain, WGS84 elsewhere (unless you know that locally-supplied data have been surveyed correct to some local datum);
• As a minimum requirement everyone should use the same datum, even if it's the wrong one;
• Don't confuse pilots by telling them to switch datums and position formats before uploading flight logs or downloading turnpoint data; it makes no difference;
• Make sure that locally-sourced data to be downloaded into pilots' GPS are correct in WGS84.
Things for the scorers to remember:
• Always make sure pilots are talking WGS84 when they phone in their positions;
• Make sure your software knows that Garmin uploads and downloads WGS84 (Check'in does).
Things for the pilots to remember:
• If the flight is scored using GPS, then fly to the GPS "virtual" feature not the ground feature!
• If you land out and phone in your position, switch your GPS to WGS84 before you read the numbers Otherwise you will be out of step with everyone who uploads the data directly from their GPS to the computer in the scoring room.
• After the briefing check that your task length agrees with the task board to within 0.1km. Question it if it is even slightly wrong.
Things for everyone to remember:
• When communicating GPS position references, make sure you specify the relevant datum.
• When collecting position references from someone else, be sure to find out in which map datum the originated, and set your unit to that datum before you enter them.
Thanks to Prof. Clifford Mugnier (Professor of Surveying, Geodesy and Photogrammetry, Louisiana Stat
University), Russell Fox (Chief Librarian, Ordnance Survey), Jeff Butterfield (User Support, Garmin), John Aldridge, Clive Belbin, Gordon Rigg and Mike Stephens for proofreading this article and suggesting alterations.
(C) Oliver Moffatt 2000