Altitude Accuracy

Both of my Garmins usually can't settle on an altitude. If you place them on a stationary object, the altitude number will go up and down. In general, it reads higher than the actual altitude. It all depends on the current satellite accuracy, which changes as the satellites move across the sky.

+/- 400 feet not uncommon for altitude

https://support.garmin.com/support/searchSupport/case.faces?caseId={66f1b0a0-4cd6-11dc-4733-000000000000}

I think the issue with this is that there is no way to calibrate the altimeter for any more accuracy. I have a trail map app on my phone that is excellent and has an altimeter, but in order for it to be good, you have to calibrate it by entering in a point of known altitude.

GPS elevation accuracy with consumer grade GPS equipment is notoriously inaccurate. This is particularly true at sea level with a partially obstructed horizon. A 2001 article at gps information.net explains part of the problem. Most is still true today:

http://gpsinformation.net/main/altitude.htm

As ddeerrff points out, Garmin claims a +/- 400’ elevation error with their GPSr’s:

https://support.garmin.com/support/searchSupport/case.faces?...

The next generation of GPS satellites will hopefully improve this woefully inaccurate situation.

I used my GPS on a flight from NY to FL on JetBlue.
One of the TV stations shows the location,speed and altitude of the plane. My GPS only had very little difference with the one showing.
I attributed this to the location of the sensors. I had the same reading difference while on the ground.
My speed reading was very close to the one showing on the TV.

Don't altimeters use air pressure to gauge altitude? If that's the case, I would think that the GPS would be more accurate.

The altimeter uses air pressure. When flying you get the current altimeter setting (air pressure) from your weather briefing and update it along the flight. It is given on the automated weather broadcasts at airports.

An altimeter does use air pressure to calculate their reading. Pilot's need to know the current pressure at their location to set the altimeter in the aircraft before takeoff. Failure to do this before flight can result in a significant error in the reading. (high or low). Automated recorded information at airports provide this information for pilots and it is also a standard piece of critical information provided by air traffic controllers and the tower controller at the airport on approach and prior to takeoff. It's especially important on approach/takeoff in mountainous terrain to get correct! An error of only 100' is serious! Even wrist altimeter hikers use will thrown off due to pressure changes caused by approaching weather fronts, high or low pressure areas moving in, and need to be calibrated periodically en-route against a known reading from a countour map.

There are, or could be, two altitude indications. One is determined from the GPS satellite system and the other is determined from atmospheric pressure.

It is important to know that the atmospheric pressure means must be calibrated frequently. My GPS devices can do constant calibration of the atmospheric altitude indication against the GPS derived altitude indication.

When I leave the GPS sitting outside, the two will agree.

The atmospheric altitude just isn't trustworthy if it hasn't been calibrated recently.

The atmospheric altitude is affected by slight air pressure changes. Opening/closing windows in my car as I'm driving makes HUGE changes in "altitude". Activating fresh air/recirculate in the car changes the "altitude".

Whenever I am driving I know my altitude without looking at a GPS. If it isn't 0 AGL there is a problem.

My Garmin Oregon 550 has a barometric altimiter. But interestingly, you calibrate the barometer by setting the correct elevation at a known point. From contour maps I know the elevation of my home, at least within 10'. When I calibrate my Oregon using this, the barometric reading is usually very close to what the local weather stations are reporting. So I can then be pretty certain that the Oregon is reporting fairly accurate elevation readings ... at least for a while. But has been said, you need to calibrate often, especially if barometric pressure is changing.

My nuvi's which of course just use satellites to report elevation are usually within 50 or 60 feet of what the Oregon shows, but sometimes can be way off by several hundred feet.

Altitude in GPS is hard.

The main hard part is, "Altitude with respect to what?"

The Earth's surface? Sea level? What does "sea level" mean in Omaha or Edmonton?

The Earth's surface isn't a sphere. It's a lumpy ellipsoid, and the lumps move (tides, normal continental drift, and non-normal continental drift called earthquakes). Even "sea level" at the coast is an approximation based on mean tides.

GPS uses an ellipsoid approximation, WGS-84, for the Earth's surface. "Altitude" to a GPS box is the height above this ellipsoid at a given point. WGS-84 is a committee thing, and you know how good committee decisions are.

So yeah, plus or minus a few hundred feet is pretty good.

And the future? Difference between accuracy and repeatability.

Repeatability is getting the same (or close to the same) result time after time. Accuracy is getting a measurement that's close to a known standard or reference.

The modern generation of GPS receivers give more repeatable results (by using data from multiple birds and better models for combining that data) than early GPS receivers.

But because altitude is the height off a calculated (fictional) surface, don't expect altitude calculations to be more accurate with newer generations of GPS receivers alone. The extra (civilian) frequencies will help correct for ionospheric errors, which will help with 2D positioning, and give results that are more repeatable (and 2D accurate), but that altitude thing? It's still dependent on WGS-84.

So yeah, altitude is hard.

Well said.

The altitude indicated on my 3597 near my house is pretty much ~ +-20ft, pretty good as far as I am am concerned. As for absolute accuracy, I do not know. I typically use the altitude just for some fun and casual information.

When I drive along Hwy 101 near SFO, the indicated altitude is ~+-10ft, not bad compare to my visual of the SF bay.

Driving through the hills, I sometime use the increment/decrement of the displayed altitude to compare with my senses of if I am going uphill or downhill. And it is comforting to see that they agree usually.

Geoid is an equipotential surface of the Earth's gravity field, which approximates the mean sea level. Some information and a visual map of the earth’s geoid can be found at http://www.ncaor.gov.in/pages/researchview/11 as well as a number of other locations. A relatively simple table can relate the WGS-84 ellipsoid to the geoid with an accuracy of several meters or better for any location on earth. I don’t know which if any gps units incorporate geoid correction. Aircraft units probably care a lot. Not really of much value for automotive receivers which assume you are on the earth’s surface.

All gps location calculations are done by triangulation solution from the satellites. Most incorporate some form of “best fit” and averaging approach to reduced apparent error and jitter. For example every Nuvi I have used will never show your position as near a road. Once you are sufficiently close, the unit assumes you are actually on the road and shows your position there. In addition if your position has changed very slightly, the screen may not be updated. That makes it hard to see the instantaneous error effects, making the unit look more accurate than it truly is. Handheld units (as you might use for Geocaching) typically show the instantaneous position, allowing the apparent position to drift around on the screen.

Position computation typically includes the computation of the Dilution of Precision (DOP). DOP indicates how the satellite geometry affects know errors. Handheld units often show an “error circle”. It is not a known error but rather an indication of expected errors based on the Horizontal DOP. Because of the typical geometry, the Vertical DOP and hence the vertical accuracy observed is noticeably worse than horizontal.

Basically:
• The apparent horizontal accuracy of an automotive gps receiver is better than it truly is.
• The vertical accuracy is noticeably worse than the true horizontal accuracy.
• Automotive receivers probably ignore the geoid because they don’t need it.
So one would expect the computed elevation to show a lot of error.

People who compare what their GPS receiver shows in cruise flight to what the flight crew is seeing will get another whole category of error which has not been mentioned in this thread yet so far as I can see.

Once an airplane is above the transition level for a given jurisdiction (it varies somewhat from place to place but 18,000 feet in the USA, and often less elsewhere) all the altimeters are supposed to be set to a standardized assumed pressure level (1013 millibars). This avoids the problem of different airplanes having different settings, which would make the vertical component of separation assurance more problematic. It also comes closer to standardizing aircraft performance than would a more accurate pressure setting.

However it can easily give a 1000 foot discrepancy from true altitude in cruise flight, so that even with the geometric errors arising from a rather poor collection of satellites visible to your receiver through the window (and perhaps some in cabin reflection going on) quite often in cruise flight your consumer GPS receiver will show a more accurate altitude (understood as feet above sea level) than the one the airplane is being controlled to.

Take away wisdom from this point, when the airplane you are riding is being controlled to flight level 360, that is actually a pressure reference, not an altitude reference.

That is a REAL low-rider!

Depending upon which vehicle I am driving, and whether I am referencing my head or my fundament, my altitude is generally between two and six feet AGL.

- Tom -

I didn't expect anyone to notice that. My tires are usually at 0 AGL but of course that changes when you go over a bridge.

Never really thought of this feature...good idea for a new product release.

My Garmin GPSMAP 64s has a Data Field option called "Elevation Above Ground."

I'm going watch it to see what happens.

Or on a ferry.

Which raises the question - when in a tunnel, what is your AGL altitude ?

EVIL

- Tom -

[quote=-et-

Which raises the question - when in a tunnel, what is your AGL altitude ?

EVIL

- Tom -

Appreciate the detailed explanation...

Thanks for all the GPS education. Never even thought of some of the points brought up. I was thinking of elevation above sea level but didn't think about that not even a constant.

Can I assume the relative altitude change is more accurate than the absolute? I measure 280 feet of rise from the main highway about 1/2 mile away and my parking spot beside the house. Just curious if that is about right. Keep telling myself I'm going to walk to the main road and back for exercise, not sure my knees will take the climb.

I did compare the GPS numbers with Google Earth and the numbers are close. Of course I don't know where Google gets their data from or what it is relative too.

Jim...

On average, the relative elevation readings taken by your GPS will be more accurate that the absolute. The magnitude of this will depend on the horizontal distance, length of time elapsed between readings and the number of satellites in view at each location.

The altitude datum used in Google earth is based mainly on the SRTM data (Shuttle Radar Topographic Mission) made in Feb, 2000, which mapped approximately 80% of the earth’s surface. Errors of +/- 5 meters are not uncommon but usually average < 3 meters. Radar measurements were made relative to the earth's surface and adjusted using the appropriate geoid model datum (WGS84, EGM96, NAVD88, etc).

It’s interesting to note that this vertical error is far less that the average 60 meter horizontal position error of Google Earth in North America. This is just the opposite of a consumer grade GPS unit.

A topo map “should” be reasonably close but not perfect. Probably best when the gradient is relatively flat. Somehow they interpolate the contour lines from known elevations. It used to be a real live person drew them in by hand based on aerial photographs. (That was clear when you found a real boundary in the maps. Sometimes they misaligned by several contour lines.) One could hope that satellite based topography is now better.

Errors based on GPS satellite geometry are not short term stable. If I set my handheld GPS down, I can sometimes watch the apparent location walk around by maybe 30 feet in just a couple of minutes. (It also has a time averaging mode for better accuracy.) One would expect the elevation inconsistencies to be noticeably worse. Using two GPS’s simultaneously probably doesn’t help that much.

Our little GPS boxes do a lot of computation to figure out where they are.

One way they seek to minimize errors is to cheat.

They cheat by making assumptions about what reasonable motion (changes in position from one GPS solution to the next in 3 dimensions) over time is going to be.

Let's rank situations from min to max motion, answering the question, I'm a GPS antenna mounted to a:

house -- pedestrian -- bike -- car -- Cessna 150 -- F-22 Raptor

A house isn't moving (nominally). Changes in 3D position have to be stable (e.g. moving the antenna).

Pedestrian -- bike -- car: these show mostly 2D motion, with limited/slower changes in altitude.

Aircraft solutions allow for the validity of much more rapid changes in altitude, and in all 3 dimensions.

What's signal and what's noise change.

One approach to developing a better solution in a moving platform is to incorporate more data into the solution -- data from more sensors, such as accelerometers, speed sensors, magnetic compass, and the like. We can also incorporate external information on the local accuracy of our GPS solution -- differential GPS, where a local transmitter sends out a stream of GPS corrections.

But altitude is still hard! Relative changes we can measure pretty well. It's that relative to what question that's difficult.

We're not flying a plane, where altitude is critical.

I just find it's interesting to see how much the elevation has changed, especially when you feel your ears pop, as you go through a mountainous area.

In many cases, the actual elevation at the peak is marked on a highway sign. Generally this has proven my 2555 to be as accurate as I will ever need.

Let me chime in with a couple observations I have made over the years.

1) I took my Garmin Nuvi 1490 up in a private plane years ago. The altitude was all over the map. The pilot had a "made for aircraft" Garmin unit and it was dead on. Moral of the story: never use an auto GPS in an aircraft. The same goes for water -- lake or ocean.

2) Went to the top of Pikes Peak in Colorado back in 2010. The same Nuvi was with me at that time. Pikes Peak is registered at 14,110 feet above sea level. The parking lot is at ground zero on Pikes Peak. My Nuvi read 14,103 feet mounted on the windshield of my parked car. A 10 foot difference -- not to shabby.

3) Last week, with the very same Nuvi, I was parked at Bad Water Basin in Death Valley, CA. The altitude there is 282 feet below sea level (-282 feet). My Nuvi registered 280 feet below sea level (-280 feet). Being 3 feet up off the basin floor, 1 foot off is really good.

On drives through the mountains of Nevada and California, the altitude lags behind the posted altitude signs all the time. When I stop and the Nuvi has a chance to catch it's breath, the altitude is more accurate.

It is a guide and nothing more. That is how I look at this feature and adjust accordingly.

There are a couple more big cheats that automotive GPS uses:

Assuming you are on a roadway and indicating that you are even if the computation only places you “close”.

Knowing that cars have limited acceleration and turn rates and smoothing the position indication accordingly.

I wonder if they apply those cheats in 3D. If so, that would allow a unit to indicate more accurate elevation than actually computed.

BTW, the only time I ever actually checked elevation was with an eTrex while biking in the hills of Ireland. Reported elevation was something like -180 feet. Since I was on a hill where I could see the ocean in the distance, I didn’t believe it.

That's the easiest, unlike Pilots we didn't care about true altitude, just how many feet above ground. I compared those altimeter's to my Garmin, Garmin all over the place.

Some of the old Garmin eTrex models used to include modes for high altitude skydiving. You needed to preload landing elevation, chute opening elevation, a winds aloft table, and maybe some other things. It was intended for soldiers who needed to jump high and really needed to hit the landing zone. I suspect the government paid to have the feature added and in return purchased them by the thousands. Don’t know if any current model have that feature.

I knew I came to the right place for my question.

good information here

Always great info here

I have a Garmin GPSMAP 64s and before that an eTrex 30. Both models have a built in altimeter which uses air pressure in addition to the other methods of triangulation. I've never activated that, since you have to calibrate it frequently to compensate for weather-related air pressure changes.

Besides, when hiking, my knees, thighs and heart rate will far more accurately warn me when I am exceeded the maximum elevation climb for a man of my age. No need for any special technology there!

Getting older isn't as much fun as it used to be?

I've seen fluctuations below sea level.

As have I, goboymd. Altitude is not a concern normally for me.
Several of the previous comments are very disingenuous. A Garmin navigator is intended for use while driving, cycling, even walking.

For all activities done while one is basically firmly planted on the earth, altitude should mean above sea-level, and sea-level should be a standard mean tide or sea height. The fact that I'm driving on hard ground in Florida, and going slightly uphill, while my Garmin tells me my altitude is going down (-3, -4, -5) is certainly a concern that the thing isn't working right.

There are numerous places on this earth that are below sea level and you're still on dry land. Sometimes very dry.

Jim...

A few years ago I purchased a professional GPS designed for boat use. I set it up in my home near a window and hooked it up to my computer.

It came with a few different software packages, and each one showed a different altitiude value.

There's something strange with GPS altitude readings.

Yesterday, I had both Garmins on the windshield to go shoot some pictures (i have not consolidated all of my waypoints). The one closer to the redlight that i was stopped at showed 90 ft a right turn whereas the other (closer to me) showed 80 ft to the same turn.

But as the name implies (Bad Water Basin), you don't want to drink it!

Yea that when gets me to, sometimes I'm around the corner and it is still telling me 50 feet to the corner. This can be confusing when multiple roads are coming together, panic hits and I wonder if I turned to soon.

Driving on Hwy 19 about 12 miles north of Summersville, WV at 70+ mph.

Saw elevation sign that said elevation was 2276 ft. My Garmin 2555 said 692 meters, or 2270 ft.

I'd say that was damned fine accuracy.

When I used to skydive, I opened my parachute at the altitude where I could make out individual trees.

Air pressure didn't matter.

Air temperature mattered more. Cold air, in winter, gave MUCH softer landings.

Has anyone taken alternate readings at BGL?

Agreed. My wristwatch altimeter readings can vary widely. Also, the manufacturer's manual indicates it's advisable to be stationary when taking readings and that it measures relative and not absolute altitude, not sure how?

I bought a 3D relief map of the general area we live, something like 2'x3'. Was interesting to see the Mountain we live on the side of (lower level, only a few hundred feet up) and the surrounding area that still seems very hilly and it is almost flat in comparison. Kind of amazing how a few hundred feet of roll in the terrain can seem very rugged but when scaled to a few mountains, are just ripples.

Taking a trip tomorrow through some new mountain terrain, I will be watching for elevation signs to compare to.

Wikipedia says my town at 5,332' MSL. A topo map I dug up while looking for some land, the contour lines show about 5,330' to 5,335'.

On the other hand my town is proud, so they put up this BIG sign near the stop light (so you CAN'T miss it) stating it's at 5,340'. At the light my Nuvi shows 5,335' in one direction and 5,334' in the other. Close enough for me.

I had no opportunity to test new spots on my trip last week, the signs that I always thought were altitude on the interstate must be something else because they were all way off from my GPS and common sense in a few places.