That is a very crude measurement. It may be great for airplanes and such, but it isn't going to be of much use at measuring different elevations on the ground.

 

A crude measurement, "Good for airplanes?". Of course it works on the ground. I've used it for years on my bike. Garmin uses BP in its GPS. I can't say if it is more accurate than whatever other method might be used, but I certainly, after asking the question, "Why doesn't it take altitude into account?" would not dismiss an answer out of-hand.

So, what's your idea?

 

Try this:

en.wikipedia.org/wiki/WGS-84

 

What I said was "Why do you think it doesn't take altitude into account?". I did not say that GPS does not take altitude into account. I asked somebody why they thought it did not. I know that GPS is a 3-D measurement. Please read and do not misquote me.

A large elevation change can result in a small pressure change. Do you know what else can result in a small pressure change? The weather can.

Select Garmin GPS units contain Built-In Barometric Altimeters. Mine did not, but it still reported elevation. I believe BP is good when used in conjunction with GPS. BP alone has issues in calculating elevations on land. Maybe crude was the wrong word, but here is an example of problems.
http://www8.garmin.com/support/faqs/faq.jsp?faq=142

[/SIZE][/FONT]

 

Anyone notice my screen name?

I've been using GPS on both consumer level, and professional engineering level combined for about 12 years. I'll try to answer any specific questions.

I'll tell you this:
Any consumer grade GPS ( those under $2000 ) are very good at X,Y ( Lat, Lon ) but are terrible in Z ( Alt. ). Z is an estimate, and the error is usually 10 times greater than the X,Y.

And the accuracy, or "precision" changes day by day, and hour by hour.

Today you may have good sat. geometry and Lat/Lon to +/- 15 feet.
( Alt would therefore be +/- 150 feet. )

Tomorrow you may have poor sat. geometry, and there's nothing you can do about it. Your Lat/Lon may be +/- 50 feet and your elevation may be +/- 500 feet from actual.

Now, +/- 50 feet is still going to allow your Navigation system to get you on the correct street, and to the correct house. Altitude is not important for that, so being off 500 feet in elevation won't make you miss your turn, but it does make your speed and distance off enough that its error is worse than the speedometer / odometer in your car.

Consumer grade can be ***** on dead accurate. But it is a crap shoot.
Sometimes it is, and sometimes it isn't.

So what do you do? Do 5 or 10 measurements always on separate days, and see if there is an obvious trend to your data. GPS satellites are not geostationary, so you will get different results on different days.

GPS is only a good choice for measuring flat, straight distances.
( Or speed traveling in a straight line )

Speed and distance around an arc is only an estimate.

If you want to know how surveyors get around these problems, I can get into that, but that is a new topic.

 

I picked up a $50 GPS 'mouse' from Ebay and found:

1. at least 15-20 minutes before believing any values - it takes a while before the GPS x, y, z values reach some stability. Fortunately the USB interface from my laptop provides power 24x7, which means it can 'settle' while I load the car.
2. at least 5 satellites - anything less looks like a random set of numbers, particularly in z but x and y wander too. BTW, 5 satellites is nothing to shout about but is closer to 6 and 7 satellites.

I have an interest in Prius energy studies and find that for altitude changes of 6-10 meters, there is good agreement with Prius energy flows. This allows me to map hills and with enough samples, I can get useful correlations.

It looks like an inertial reference combined with a GPS would be the way to go. It looks like there is some lag in GPS reported ephemeris but I'm too busy driving to quantify the effect. I get the distinct impression that at higher speeds, the GPS is reporting positions behind me by some time difference but I can't quantify the effect. Any suggestions on how to measure the GPS lag in position reporting?

One thought I've had is to combined a GPS mouse with data recording and send it to folks who report poor mileage. Then they would return the unit and we could analyze their routes to see what speeds and terrain they are really driving. Ultimately, it could it might include some LEDs to indicate recommended speeds.

Bob Wilson

 

My GPS, a tomtom GO 700, doesn't reveal elevation to me. I understand the concepts of the spheroid and the great circle arcing to find a shortest straight line. Combined with a 2D navigational map (tomtom once wrote an article saying that 3D mapping was a future development), I can see coordinates being extrapolated to points on that spheroid to match the map. Most important to me is the mapping; It doesn't take into account elevation, so even if the GPS is using elevation to help track velocity and distance, at some point the odometer and GPS can't resolve to the map distance, not without some self-correction. Then I read this, concerning GPS vs. odometer in the context of mountain biking. If found a discrepancy due to the more granular terrain. Little dips in a trail are likely not to register on the GPS.

Which is probably why the bulletin for the Caddy said to test on a flat, straight road. But your discussion on the other forum made it sound as if you were proving odometer discrepancy over longer distances with mixed terrain.

 

Ahem. My Peugeot 307 SW SE 2.0l (about to be replaced with a Prius) tells me I average 43.5 mpg while my running average (miles driven / fuel purchased since new) tells me 49 mpg...

 

My first attempt to check out the odometer was done in western Colorado driving from Grand Junction to Denver on I-70 and using interstate mile markers. The wild errors I saw with the mile markers over hilly terrain prompted me to buy a handheld GPS. My next attempt to check out the odometer was done on I-70 traveling east and mostly through Kansas. I recorded data points for segments just under 100 miles. All of these segments were strictly on one interstate (no turns, stops, exits, etc.). This terrain was relatively flat, but it did consist of a gradual elevation drop.

Anyway, I was recording trip odometer numbers and the GPS odometer numbers at various points between 0 and 100 miles. I then plotted these numbers and used the slope to determine the offset. I did a plot for each set of data and averaged the slopes. Any comments? Is this a good way to calibrate an odometer?

It would certainly be a good experiment for me to do to use the GPS to determine distance over the 20 mile marker segments in western Colorado where I observed great variations by my odometer, but I haven't made it out that way again.

Gpsman, I do have some questions for you. How do GPS units calculate distance traveled? Do they calculate the distance between 2-D points or 3-D points? With that elevation error, does it calculate changes in elevation over a shorter time frame with more accuracy than what it is able to do for absolute elevations?

 

All GPS units calculate distance like this:

At time = 0 you are at point 'A'
At time = 1 you are at point 'B'
At time = 2 you are at point 'C'
At time = 3 you are at point 'D'

Now.... connect the dots with straight lines.

If you are traveling straight, no biggie.
If you are traveling in a circle, the plot will look like a stop sign, not a true circle.

I will say most units use a time interval of 1 second between points.
Some could be 2 seconds, 5 seconds, 10 seconds, or 60 seconds.
It's up to the manufacturer and particular software.

Time interval is no biggie if you are traveling slow. Walking for example.
Your change in distance per second will be slight.

Driving, your change in distance is 88 feet per second at 60 MPH. 110 feet per second at 75 MPH.

If there is a bump, bend, dip, or anything else in the 110 feet, it will get missed. Now what if your unit plots every two seconds? Any feature in the road the past 220 feet will get missed.

When looking at road mileposts, they are just guidelines, or estimates.
Each post was not placed by a survey crew. Most likely, it was placed by a guy in a truck, using the truck's odometer, and reads to the nearest 0.1 mile. The exception is, most major highways will have a "speedometer check section". It will be a straight, flat section with 5 posts in a row that ARE very accurately placed. I think the distance between mile marker 1 and mile marker 100 should be very close to 100 miles exactly. However, the distances between individual posts is only +/- 0.1 mile.

And isn't the distance different if you travel in the right lane or left lane?
Of course it is if the road has curves. On an oval track, you always want to race on the "inside" right?

GPS is great. It is exceptionally good at measuring things when you are standing still. It is exceptionally good at measuring property lines, for example. I've done GPS measurements of property lines with +/- 1 inch precision over a 1400 foot distance.

It is also great to tell you where to turn, and to keep you from getting lost in the woods, when +/- 50 feet is "good enough". +/- 50 feet per measurement is about as good as it gets when you are moving down the highway. It is not good enough to give you the kind of detail you guys all want for your cars.

Back to the original question. Is it 2-D or 3-D?
I would say most car systems, or mapping systems are 2-D.
Most hand-held are 3-D. But it is up to the manufacture, and software.
Even in 3-D, it is still "connect the dots".
And if you go under bridges, tunnels, canyons, urban "canyons" with tall buildings, you miss "dots" all the time. You don't know it, but you do.

GPS was designed to navigate ships at sea, when they are out of range of all land-based radio beacons. It has evolved into much more than that.
While you can get very precise measurements with lots of processing power, that is not the design intent.