GPS altitude
 

Amongst other things, my old Garmin GPS III (non-pilot) tells me my current altitude - assuming it has found a certain number of satellites; if not, it switches to 2-D mode and the altitude display disappears. Sitting on the ground it seems to do a good job - both the figure in Pooley's and the altimeter (with the aerodrome QNH set, of course) are within the estimated position error on the GPS (around 15ft).

Now assume I'm at 2500ft (on the altimeter) doing 80kts. The GPS is still tracking a reasonable number of satellites, the EPE is not significantly different to the value on the ground and we're still in 3-D mode. Would you expect the GPS to still agree with the altimeter. If not, why not - and if the difference is significant which is more likely to be in error?

In terms of actual distance above nominal sea level, I'd expect the GPS to be more accurate most of the time - because it is not affected by all the things that make altimeters vary - changes in baromatric pressure, air temperature etc.

But two major caveats:

1) When the GPS gets height wrong, it can get it really wrong, and my GPS III Pilot sitting on the ground at home quite often cycles through a range of +- 60 feet over a period of a few minutes. Funnily enough my old Garmin GPS-12 doesn't do that.

2) Everyone else in the world willbe using altimeters, and air space boundaries, e.g. the bases of airways, the height of a MATZ etc., are based on altimetric standards not geometirc accuracy. So it's usually more important to hav an altimeter reading that is the same (inaccurate) figure as everyone else's rather than know your actual height above a nominal sea level (real sea level, of course, varies according to tides among other factors!)

You can see an example of the differences between GPS and barometric altitude (recorded during a glider flight) here (In French)

Red curve is the barometric altitude (metres)
Blue curve is the GPS altitude.

Bre901 - that's a very interesting plot. For most of the time they are almost identical, but for one ten-minute period there is a difference of up to 300 metres. I wonder why? I think it is explained, but my schoolboy French didn't get the meaning...

Lowtimer - I see the altitude vary on my GPS III as well, however the variation is always (or at least seems to always be) inside the EPE quoted by the GPS (i.e. with a 15 ft EPE, you'll get 85-115 ft when you're 100 ft AMSL). I understand your second caveat as well, but i'm trying to work out if a GPS is reliable enough to highlight problems with an altimeter, or if any difference could just be a GPS wobble. I'm not planning on using it as a replacement :)

A person who writes software for GPS devices told me yesterday that their ability to give you a sensible height is basically crap and that this is inherent in the technology.

(IFR approved GPS devices might do very much better than others, he was maybe not talking about them.)

Here's an example statement from a GPS manual:

"WARNING: The KLN 94 displays GPS-derived altitude on the AUX 1
page. Due to many factors, GPS altitude may typically be several
hundred feet in error. Do not use the GPS-derived altitude for navigation."

GPS altitude references the geodetic altitude which is a mathematical model of the equipotential surface of the Earth gravity field that most closely approximates the mean sea level. As this changes from around -106m to +85m around the world, your GPS could tell you you're below sea leavel even when standing on the sea shore. GPS altitude errors also tends to be in the order of 1.5x the horizontal errors, due to satellite geometry. Assuming a horizontal accuracy of 75', this leads to an altitude error of 112'. So in theory you GPS altitude could be in error by 685'.

Using differential GPS (WAAS etc) you can virtually eliminate these errors, and could use GPS altitudes for vertical navigation, on say a precision approach (virtual ILS). IFR approved units tend to also have a barometric altimeter built into the GPS.
EA

Evo

The reason for that difference is because during that period the glider was spiraling (climbing & turning very tightly) so the GPS lost view of some satellites and reverted to 2D mode, the altitude being wrong or "guessed".

I'm sure I watched a programme on Discovery (or somesuch) where a group of scientific bods were figuring out the exact height of Everest and K2 blah blah - wasn't really paying a great deal of attention until they mentioned that it was being measured by GPS!

This was about the time that the good ole CAA were telling me that NDBs were far more reliable for non-precision approaches than that dodgy box of magic.

I've been using a Skymap 2 and a KLN94B for 4 years, and have never (once it has picked up enough satellites) seen its displayed altitude to be more than 50ft different from the elevation of a known place on the ground.

Once one gets airborne, the GPS will normally be more accurate, due to simple altimetry errors due to the atmosphere deviating from ISA. At 10,000 ft the altimeter error could easily be 500ft whereas the GPS will still be perhaps 50ft out.

I always check both altimeters against the GPS altitude when doing an IAP in IMC, and if the error was more than about 200ft (which I have never seen, in 350hrs of flying) I would go somewhere with an ILS. There is no point in risking one's life just because a large group of PPLs distrust technology :O

GTW - the person you spoke to doesn't have a clue. The altitude computation is part of the lat/long computation and is only slightly less accurate because of the angle of the satellites. I don't think he actually writes the software in question - very few people do that bit because most GPS receivers are purchased ready-made from 2 or 3 suppliers.

There are situations where the altitude figure can be wildly out, and these are the same as why the lat/long figures can be wildly out: ancient GPS receiver with buggy software, poor reception due to not using a rooftop aerial in a metal cockpit.

Obviously altimeters are used as the primary reference for good reasons, but that isn't the question we are discussing here.

Just to clarify myself ;)

Should be noted that post selective availability most of the time GPS is highly accurate, and that:-

For horizontal accuracy you can expect:

within 2.5m 50% of the time
within 3.8m 68.27% of the time
within within 7.0m 95.45% of the time
within 9.8m 99.73% of the time

For vertical nav:

within 5.6m 50% of the time
within 7.4m 68.27% of the time
within 14.4m 95.45% of the time
within 21.3m 99.73% of the time

Around the UK the difference between the geodetic altitude and MSL is not very great, though some places like the Indian Ocean show much greater deviation (-105m). So most of the time, GPS altitude will closely resemble pressure altitude (+/- 50 feet or so). Add the max error of 21.3m into it and you could possibly show a deviation of 100' or so.

Adding in ground based reference stations, and altitude accuracy increases to:

68.27% confidence: 2.7 meters
95.45% confidence: 6.4 meters
99.73% confidence: 12.0 meters

which is probably better than a pressure altimeter.

Cheers
EA

You don't mean pressure altitude, at least not using the conventional definitions, which is flight level (indicated alt with 1013 set). You mean indicated barometric altitude.

Nice analysis, BTW. It's never been clear to me if aviation GPS receivers use a Geoid model to convert height above the standard ellipsoid. It doesn't seem hard to do. Yet geoid height gets the blame for a lot of the vertical errors.

Sounds much better:D

Apparently - despite the oblate-ness of our spheroid - the difference between polar and equitorial circumferences is something like 22 miles.

I can't believe I had to learn the phrase oblate spheroid for that!

Isn't there also an issue over the inherent errors being greater in altitude measurements due to the poorer intersection angles of the pseudo ranges vertically?

I seem to recall measuring all this many years ago, and concluding that because the vertical distance from the receiver to the SVs was so much smaller than the horizontal distance, due to the relatively low orbital height of the SVs, there were bound to be bigger spherical geometry errors in that plane.

IIRC I concluded then that the vertical accuracy could never be better than around 5 or 6 times the horizontal accuracy, unless using a stationary, phase measuring, survey type GPS receiver.

Vertical accuracy will also degrade faster, due to the need for really good visible SV geometry to maintain a reasonably good vertical pseudo range intersect angles.

Yea correct. The error is in the region of 1.5x the horizontal accuracy though.

EA

Depends how close you mean by close. It only has to be about 15 degrees colder than ISA and you'll have a difference of almost 300 ft at 5,000 ft, or just over 100 ft at 2,000. Allow a good distance from any controlled airspace if your altimeter fails and you have to use the GPS altitude...

Mark

SSD don't forget that the most familiar oblate spheroid is a rugby ball ;)

Remember also that GPS altitude is based on a model of the earth's surface, not an actual reading .... and we all know how good the US is at reading maps :)

No offence meant :}

GPS altitude does work if used in the correct circumstances. For something like £20,000 you can buy a "GPWS" system for GA which consists entirely of a GPS and a contour database of the region. There is no radar altimeter. I've seen a demo of it.

The GPS altitude is compared with the terrain map and it takes into account your flight path, warning you if you are going to hit something. It's a certified system which you can buy today from Honeywell and probably others.

Personally I would go for a radar altimeter, at about 1/3 of the price. That would be actually useful as a final safeguard on instrument approaches.

I am sure that if aviation was invented today, a lot of things would be done differently....

IO540 (nice engine, btw)

That is a EGPWS that you are talking about, and they actually don't include the GPS, at least in the Honeywell set (I do some consulting for Freeflight Systems, who bought the aviation GPS licences from Trimble, and who Honeywell asked to program a basic unit as a cheap input for their EGPWS for any aircraft with no convenient position output available). They use whatever position information is available, FMS position being the most accurate which takes into account much more than GPS position.

We (Freeflight that is, not BCFT or Solent who I also work for!) can also sell you the radalt you prefer, although that has no forward prediction.

I see no need to complicate the cockpit by ever refering to GPS altitude. Keeping MSAs 1000 feet clear of terrain, 2000 in mountainous areas (due to venturi effect) is perfectly adequate to avoid terrain. Baralts have to be used for aircraft separation, and at the levels at which we have terrain in the UK are very accurate anyway (you pointed out youself that the high errors are at high altitudes). There is no need for GPS altitude in aviation. It is a lot less accurate than horizontal position.

M14P

GPS is not entirely reliable, and when it goes, it all goes! You can't change to a different system, except to go back to traditional radio navigation. Occasionally it even goes in such a way as you don't know for perhaps 2 minutes, which can be bad on approach. With an NDB, as long as you can hear the beeps it is working, crude as the system may be. In aviation known inaccuracy is much more acceptible than extreme accuracy with any question of reliability.

Reliability has nothing to do with measuring K2. That is all about accuracy, and if you leave a GPS there long enough it will give you an answer to the nearest millimetre or so (they are used by geophysicists to track plates).

The CAA is a bit over the top on its concerns about GPS, but the basic philospohy is right, and the US government and military are now moving toward that position. The FAA are in fact looking at combining GPS and Loran (yes, you read it right, Loran!) for a more reliable system.