How accurate is Elevation supposed to be

 

Recently during our travels from Arizona to Montana we have gone over several routes where there are elevation signs. The posted elevation and the Garmin Nuvi display are usually off by 30' to 50'. Is this normal? I know that the GPS lags as it moves but I am talking about the max displayed on the GPS vs. the posted elevation on the road sign.

Jerry

Yes, normal

For consumer grade GPS devices, satellite based elevation calculation is not as accurate as the satellite based coordinates. I have read techincal explanations for this on other sites, so you can probably find more information with a little googling.

--
Alan - Android Auto, DriveLuxe 51LMT-S, DriveLuxe 50LMTHD, Nuvi 3597LMTHD, Oregon 550T, Nuvi 855, Nuvi 755T, Lowrance Endura Sierra, Bosch Nyon

That's pretty good--

GPS altitudes are also calculated as distance above a reference shape (WGS-84 usually). SO things may not be exactly on, but at least they're consistent!

--
Nuvi 2460, 680, DATUM Tymserve 2100, Trimble Thunderbolt, Ham radio, Macintosh, Linux, Windows

Vertical References

k6rtm wrote:

GPS altitudes are also calculated as distance above a reference shape (WGS-84 usually). SO things may not be exactly on, but at least they're consistent!

World wide, there can be up to a 100 meter difference between MSL and the WGS84 ellipsoid (sometimes above & sometimes below). If the posted elevation is relative to MSL and the GPS elevation is relative to the WGS84 ellipsoid there can be up to a 100 meter difference depending upon where you are. My recollection is that the vertical error of a GPS is typically 1.5 times the horizontal error.

For a technical explanation of the difference between GPS ellipsoid height and MSL (Mean Sea Level) see:
http://www.esri.com/news/arcuser/0703/geoid1of3.html

NGA publishes a mathematical model for the Geoid. Note that the Geoid approximates MSL:
http://earth-info.nga.mil/GandG/wgs84/gravitymod/egm96/egm96...

For a world map showing the difference between the EGM96 geoid and the WGS84 ellipsoid see:
http://cddis.gsfc.nasa.gov/926/egm96/egm96.html

Mark

Other probable reasons

There are two other probable reasons.

First. GPS satellites effectively broadcast their orbital parameters and a very accurate clock. This allows the precise location of the satellite to be computed at any point in time. The GPS receiver effectively measures the difference in time between the reception of the different clock ticks. From knowledge of the satellite location and the time differences, with enough satellites it is possible to compute the receiver location. Every parameter has some ambiguity which results in loss of accuracy in the resultant calculation. For a given ambiguity how much accuracy is lost depends on the satellite geometry. The computed loss in accuracy is usually referred to as Dilution Of Precision (DOP). DOP assumes constant errors. Real errors are unknown and may exceed the assumption. DOP is therefore a computation of “typical” error for the current constellation, not a measurement of the current errors. DOP can be computed in both horizontal and vertical directions although normally horizontal is displayed. DOP is at its least when there are large angles between satellites. Horizontal DOP is virtually always less than vertical. With satellites at the horizon, the difference in horizontal angle of arrival can almost reach 180 degrees while the difference in vertical angle of arrival would be limited to 90 degrees. I would expect vertical DOP to be consistent with the errors you are observing.

Second. Most automotive GPS receivers include software that (almost) always displays the car on the road. So even if the GPS thinks you are 30 feet off the road, you may never know. The receiver has these errors all the time but you need to be paying attention under the correct circumstances to notice. The road tracking software only works horizontally so vertical inconsistencies can be seen whenever you look for them.

a few summers ago

I was traversing the Blue Ridge Pkwy in VA. They have frequent posts/markers that show the elevation as you motor along. My Nuvi was always within 1' or 2' of the quoted elevation.

--
(2) Nuvi 1450LMT + 3597LMTHD + 2557LMT + DS61LMT-S Boston MA

When?

My first thought was when I saw the discrepencies that some of those signs were put up many years ago before GPS or at its infancy. I remember that when it became available to the public it was advertised that the accuracy was within 50 feet. Today it is more accurate but the states are not going to waste money checking how accurate their signs are for it does not matter if you are 30 feet higher or lower than what the sign reads.

Generally, most of those

Generally, most of those signs are derived from Geodesic Surveys.

--
Frank DriveSmart55 37.322760, -79.511267

The distance markers however

phranc wrote:

Generally, most of those signs are derived from Geodesic Surveys.

Those signs telling you its so far to the next exit, city or whatever however are not always correct. The highway engineer specifies which signs go where but when they are installed, they are placed near the exact point, but where they can be safely installed. Some are as much as a thousand feet or more off.

--
Illiterate? Write for free help.

not showing WGS84

I think there is a wide mistaken impression that current consumer GPS units typically report WGS84 elevation directly. If that were true, many of us would see much larger discrepancies to well-referenced benchmarks than we do. I think the truth is that quite generally current consumer GPS receivers contain a model of the position-dependent correction from WGS84 to Mean Sea Level, and report to the display a value adjusted by the local difference as estimated by that model.

I don't think this model has been the same on all Garmin receivers within my (long) period of ownership. In particular I've noticed that readings averaged over a large number of samples of a specific location on my property have differed by something like twenty feet or so between models. My current models are close enough to the elevation shown on official documentation filed with the authorities for my lot as to leave me unsure of the sign of the error, and it appears to be within ten feet of agreement.

--
personal GPS user since 1992

Delta Difference

From what has been said here can I assume the relative difference in two spots a few hundred feet apart are close to right. Was trying to use the GPS to see how much elevation change I had from the base of my driveway to the house. Don't care if the true elevation is exact but it would be nice to know how much change in elevation I have fairly accurately.

Have seen a lot better

Recently going past El Centro, CA on I-8, I noticed a roadside elevation sign I had not seen before reading Elevation 0 MSL. I found that intriguing, and I wondered how the GPS would compare. In the time it took me to navigate the menu to the "Where Am I" page where I could see the GPS elevation readout, I guessed it would probably increase a bit, since I was headed west towards the hills. Sure enough, it was reading 5 ft. Pretty darn close to right on the money, I remember thinking!

--
Garmin Nuvi 760, Drivesmart 55; Retired Nuvi 765T ><> Dave <>< "He is no fool, who gives up what he cannot keep, to gain what he cannot lose." - Jim Elliot

some other errors

JebNY wrote:

From what has been said here can I assume the relative difference in two spots a few hundred feet apart are close to right.

The error in that case due to the issues we have been discussing would be quite small. However, unless you contrive to take the measurements simultaneously, using two receivers each viewing the same satellites, you'll get a not entirely negligible component of random error in each, so yet more in the difference.

If you have a unit that lets you use WAAS, that will help this particular measurement quite a lot. Also if you have a unit that lets you average measurement over several tens of minutes while sitting in one place, that also will help. I think currently these capabilities are more likely offered in models meant for hiking or sailing than in the automotive market machines. My eTrex Vista HCx does both, and would be much better for this purpose than my 3790LMT. On the other hand, our family has his and hers 3790LMTs, so for you situation we could set up both receivers and record the reported elevations at the same moment, though I don't think the 3790LMT lets you constrain which visible satellites are used in the calculation.

These random errors I mention can quite easily be on the order of a couple of tens of feet in elevation. I've made many trips from Albuquerque to Santa Fe, and have a habit of watching the peak elevation shown on passing a hilltop nearly the same elevation as my house. The reported elevation maximum at that point has varied from about 6200 to 6250 feet.

--
personal GPS user since 1992

its been a long time

Pointer.SD wrote:

Recently going past El Centro, CA on I-8, I noticed a roadside elevation sign I had not seen before reading Elevation 0 MSL.

Since I was in El Centro but I recall there being a water tower with a line on it saying sea level... And at the bottom of the tower a sign saying -118 feet below sea level.. Is it still there?

--
Never argue with a pig. It makes you look foolish and it anoys the hell out of the pig!

its been a long time

Pointer.SD wrote:

Recently going past El Centro, CA on I-8, I noticed a roadside elevation sign I had not seen before reading Elevation 0 MSL.

Since I was in El Centro but I recall there being a water tower with a line on it saying sea level... And at the bottom of the tower a sign saying -118 feet below sea level.. Is it still there?

--
Never argue with a pig. It makes you look foolish and it anoys the hell out of the pig!

Thanks

archae86 wrote:
JebNY wrote:

From what has been said here can I assume the relative difference in two spots a few hundred feet apart are close to right.

The error in that case due to the issues we have been discussing would be quite small. However, unless you contrive to take the measurements simultaneously, using two receivers each viewing the same satellites, you'll get a not entirely negligible component of random error in each, so yet more in the difference.

If you have a unit that lets you use WAAS, that will help this particular measurement quite a lot. Also if you have a unit that lets you average measurement over several tens of minutes while sitting in one place, that also will help. I think currently these capabilities are more likely offered in models meant for hiking or sailing than in the automotive market machines. My eTrex Vista HCx does both, and would be much better for this purpose than my 3790LMT. On the other hand, our family has his and hers 3790LMTs, so for you situation we could set up both receivers and record the reported elevations at the same moment, though I don't think the 3790LMT lets you constrain which visible satellites are used in the calculation.

These random errors I mention can quite easily be on the order of a couple of tens of feet in elevation. I've made many trips from Albuquerque to Santa Fe, and have a habit of watching the peak elevation shown on passing a hilltop nearly the same elevation as my house. The reported elevation maximum at that point has varied from about 6200 to 6250 feet.

Thanks, I may try it with the 2460 in the wife's car to see if it agrees. I should also download an altimeter app for my iPhone and see if it agrees with anyone.

And I forgot

BarneyBadass wrote:
Pointer.SD wrote:

Recently going past El Centro, CA on I-8, I noticed a roadside elevation sign I had not seen before reading Elevation 0 MSL.

Since I was in El Centro but I recall there being a water tower with a line on it saying sea level... And at the bottom of the tower a sign saying -118 feet below sea level.. Is it still there?

to use this feature when I was at Badwater, Death Valley.

--
nuvi 1390 LT, nuvi 1450 LMT, Win 10

ancient garmin 38

When flying my ultralight the Garmin 38 is very close that what my altimeter reads on the instrument panel, can't vouch for the Nuvi line.

--
Garmin 38 - Magellan Gold - Garmin Yellow eTrex - Nuvi 260 - Nuvi 2460LMT - Google Nexus 7 - Toyota Entune NAV

50' to 100' off the mark isn't bad

Garmin GPS is quite a bit closer to actual elevation than my wristwatch's digital display of elevation, which is routinely off by hundreds of feet unless it's been calibrated within a few recent hours to a known accurate number of feet above sea level. For most peoples' purposes, being off by 50' or 100' as Garmin is is probably close enough.

--
JMoo On

Interesting, I'll have to watch it next road trip

I thought it was a pressure sensor, never really paid that much attention.

Triangulation Method

On a flight to Florida, I fired up my Garmin NUVI 2555 LMT in the plane and was surprised to see the elevation show what I believed to be a reasonably good elevation data.

I contacted Garmin and asked if they could confirm that the elevation is derived by triangulating from the satellites. Garmin did confirm this and although, I am unable to track down the original response, I recall that they stated the accuracy to be about +/-50 feet.

Considering the calculation method, I would consider this to be pretty good and I feel that comparison to elavation markers, shows, it's closer than that.

One of the functions that I include on my Dashboard, is elevation. It's fascinating to check when you're traveling through a hilly/mountainaous area, especially as you feel your ears pop.

--
DriveSmart 65, NUVI2555LMT, (NUVI350 is Now Retired)

Trianulation

GPSgeek wrote:

I contacted Garmin and asked if they could confirm that the elevation is derived by triangulating from the satellites.

All position is triangulation based. But generally the satellite positions are much less favorable for accurate elevation computation. Therefore, on average, elevation will be less accurate than horizontal position. Might be pretty hard to check either of those.

Quadrangulation

This article in TIME says a consumer GPS device needs to see three satellites for map location information and a fourth satellite for altitude information.
http://time.com/3952770/gps-how-works/

--
JMoo On

elevation error compared to horizontal position error

Have read from several sources that elevation error typically can be about 3 times the horizontal position error...

4 satellites required

dagarmin wrote:

This article in TIME says a consumer GPS device needs to see three satellites for map location information and a fourth satellite for altitude information.
http://time.com/3952770/gps-how-works/

I would disagree with that a bit. Four satellites are required at minimum for location, and location is established in space (3 dimensions). Three satellites determine location in the 3 spacial dimensions and the fourth satellite is for time correction, (the 4th dimension) because your GPSr clock is not very accurate. (None of them have internal atomic clocks.) Each additional satellite you receive increases the accuracy due to angular differences of the received signals.

How accurate is Elevation

Unless you are flying an airplane, why would you want to know what the elevation is when all four wheels of you vehicle is touching the ground.

--
3790LMT; 2595LMT; 3590LMT, 60LMTHD

Knowledge is Good

rthibodaux wrote:

Unless you are flying an airplane, why would you want to know what the elevation is when all four wheels of you vehicle is touching the ground.

So you know how high you have climbed up a mountain, or how much higher you need to go, or how your engine/body will perform in elevated air, how low you have descended into a valley, or just because you like to know how high the ground is where you are and be informed about the world around you.

The earlier explanations are correct. GPS needs augmentation to get precise elevation information. Also the reference data may differ than the sign installers and the signs themselves can be off too.

Eggs

telecomdigest2 wrote:
rthibodaux wrote:

Unless you are flying an airplane, why would you want to know what the elevation is when all four wheels of you vehicle is touching the ground.

So you know how high you have climbed up a mountain, or how much higher you need to go, or how your engine/body will perform in elevated air, how low you have descended into a valley, or just because you like to know how high the ground is where you are and be informed about the world around you.

The earlier explanations are correct. GPS needs augmentation to get precise elevation information. Also the reference data may differ than the sign installers and the signs themselves can be off too.

Don't forget calculating the time it takes to hard boil an egg. Water doesn't always boil at 212 (100 to our Neighbors to the North).

Altitude.

It reminds me of when I was a kid and my brother drove our Dad's '53 Plymouth flathead 6-banger to the Estes Park area of Colo. for a vacation. Once we got above about 8000' the dang thing would boil over and about every 20 miles we would have to stop, let it cool off and add water.

--
Alan - Android Auto, DriveLuxe 51LMT-S, DriveLuxe 50LMTHD, Nuvi 3597LMTHD, Oregon 550T, Nuvi 855, Nuvi 755T, Lowrance Endura Sierra, Bosch Nyon

I10 CA

When traveling from LA to Phoenix, AZ, east of Palm Springs I10 dips below sea level just before you drive up the mountain for 10 miles going east The grandchildren love it.

No way /heh

avandyke wrote:
dagarmin wrote:

This article in TIME says a consumer GPS device needs to see three satellites for map location information and a fourth satellite for altitude information.
http://time.com/3952770/gps-how-works/

I would disagree with that a bit. Four satellites are required at minimum for location, and location is established in space (3 dimensions). Three satellites determine location in the 3 spacial dimensions and the fourth satellite is for time correction, (the 4th dimension) because your GPSr clock is not very accurate. (None of them have internal atomic clocks.) Each additional satellite you receive increases the accuracy due to angular differences of the received signals.

Are you saying a TIME reporter didn't quite get the science right??

/Cancel all my appointments... Someone on the Internet is wrong!

--
JMoo On

If one dimension is supplied, 3 is enough

My very first GPS (an almost shoebox sized Magellan) had an optional operation mode in which you supplied an elevation, which it assumed was correct, which permitted it to use 3 satellites to get your latitude and longitude.

At the time there were only about 13 active birds, and the unit could only try to listen to 4 of them at one time, so the chances that it could only "hear" 3 of them at any given moment were actually rather substantial, so the elevation supplied mode was nothing like so useless as it might seem in modern times, with a constellation of active satellites twice as big, and receivers that can try to listen to a dozen satellites at a time.

Maybe some here should mock Time magazine a bit less?

--
personal GPS user since 1992

Nuvi's Vs Handhelds

I find it interesting that Garmin uses barometric sensors to determine elevation in some of their handheld devices rather than rely on data derived from satellite positioning. Comparison tests done between my Rino 530HCX with barometric sensor and several Nuvi’s would often show discrepancies of 100 feet or more.

While barometric sensors aren’t perfect, I found them to be more accurate in determining elevation differences between points. The trick is to calibrate the unit using an accurate barometer or elevation reference point.

Garmin publishes a barometric readout accuracy of +/- 10’ which is considerably better than measurements taken using consumer grade satellite based equipment.

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

A five year old internet article gives a pretty good explanation of the errors involved with both barometric and satellite based equipment.

http://www.dcrainmaker.com/2010/05/understanding-sport-devic...