Please don't hesitate to call us with questions about our mapping, GPS, Survey and Weatherproof Paper products!
We can help you choose the right product for your application!   +1 801 412-0011 x12
 

GPS Accuracy Considerations

Before we address the accuracy of Magellan GPS receivers, lets get some definitions and basics out of the way.
If your in a hurry to compare GPS receivers: [ Click Here ] to skip right to the Magellan Professional GPS Accuracy page.

Let's be honest, GPS Salespeople are kind of like the stereotypical used car salesperson. And we kind of pull a fast one when we talk about GPS accuracy:

CEP

When we quote GPS accuracy for GIS real time measurements, we use CEP (Circular Error Probability). CEP is defined as the radius of a circle centered on the true value that contains 50% of the actual GPS measurements. So a receiver with 1 meter CEP accuracy will be within one meter of the true measurement 50% of the time. The other 50% of the time the measurement will be in error by more than one meter.

Please note that it is probable that CEP measurements of the same point on the ground will differ by twice the probability. For example if a receiver has CEP of 1 meter, then it is probable that measurements of the same point will differ by 2 meters.

When you look at the accuracy specifications for a consumer GPS receiver, or the MobileMapper CX in real-time, WAAS corrected mode you will get an accuracy statement like “Real-Time Accuracy <1 Meter CEP”.

This means that the MMCX with WAAS corrections, will be within 1 meter of the true ITRF 2000 (reference frame) coordinate, 50% (half) of the time. Assuming (of course) that the receiver has a clear view of the sky, there are 5 or more satellites visible and the PDOP is reasonable.

Remember that we are quoting horizontal accuracy; the vertical accuracy will be 2-3 times worse.

DRMS (also called RMS, 1Sigma)

2DRMS

Twice the DRMS accuracy, with a 95% probability.

For example, the ProMark 3 receiver in static operation has a horizontal accuracy rating of 0.005 meters 2DRMS 95%. That means we expect the receiver to deliver 5 millimeter accurate measurements 95% of the time (when used within the specification parameters.)

R95

The radius of a circle centered at the true position, containing the position estimate with probability of 95%.

Converting

To a first order, you can approximate one error estimate from another.

            1 meter CEP = 1.2 meters RMS = 2.1 meters R95 = 2.4 meters 2DRMS

(assuming σx / σy = 1 and that VDOP / HDOP = 1.9 and PDOP / HDOP = 2.1)

A great example of accuracy slight of hand can be found when we compare the MobileMapper Pro to the MobileMapper CX in real-time WAAS corrected operation. The MobileMapper Pro accuracy is specified as 3.0 meters 2DRMS. The MMCX accuracy is specified as 1 meter CEP.

It sounds like the CX is a lot more accurate than the MMPro, right?

Well not a lot more: the MMCX is 2.4 meters 2DRMS, nearly the same as the MobileMapper Pro.

Baseline and PPM

Survey receivers include static and RTK specifications. The datasheet for the PM3 Survey Static mode looks like this:

So what is up with the “1 ppm” (1 part-per-million) and "2 pm" on the Horizontal and Vertical specs?

Baseline is the distance from the rover to the reference station. If you set your reference base up on the southeast corner of a section and occupy the northwest corner with your rover the baseline is 1.4 miles or 2,250 meters.

Assuming that our static occupation is long enough to fix (typically 5-15 minutes) and meets the conditions of footnote 3 (5 or more SV’s, SBAS, reasonable multi-path) we would expect our measurement to have an rms (that is root-mean-square 1-sigma) accuracy of:

            0.005 meters + 1E-6 * 2,250 = 0.00725 meters (7.2 millimeters)

1 part-per-million of baseline is pretty standard for most GPS receivers so just remember:

            1 KM ->           0.001 meters (1 millimeter)
            10 KM ->         0.010 meters (1 centimeter)
            100KM ->        0.100 meters (10 centimeters)

Typically GPS receivers are always specified with vertical specifications that are twice as bad as the horizontal. Depending upon satellite geometry (where the satellite vehicles are in the sky) the vertical error may actually be less than the horizontal. But in general just remember that the Vertical accuracy is going to be half as good as the horizontal.

Multiple Baseline Solutions

It bears mentioning that single frequency (L1) GPS measurements start to fall apart for baselines longer than 15 KM (about 9 miles). This rule holds true for ‘Single Baseline Solutions’, which are solutions where there is one reference base and the rover.

It is possible to post process L1 static data against multiple reference stations (like CORS stations) in a network solution (sometimes called a multiple baseline solution). A multiple baseline solution benefits from software adjustment based on known vector lengths between the CORS stations. In a well designed network, with long occupations, one might expect to get 3 cm solutions, even with 400 mile vectors between the reference stations.

Magellan's GNSS Solutions program has just introduced (in version 3.10) a new VRS function that constructs a 'Virtual Reference Station' from three nearby physical reference stations. This technique allows static and Stop & Go occupations to fix very quickly because the baseline is very short.

Technique

Not only can a good GPS salesperson put a spin on Accuracy, but we can pull some real slight-of-hand during a demo. I can take two receivers that are exactly the same and collect data with both of them, at the exact same time and get results that are 30 times worse on one vs. the other!

And you won’t be able to tell how I did it. (Unless you read the next few paragraphs.)

How to make a receiver inaccurate:

1. Block the receiver’s view of the sky:

a. Good = hold receiver away from your chest, with your back to the North at arms length. The GPS antenna can see every satellite directly.

b. Bad = hold receiver under my belly, with my back to the South. The GPS antenna is blocked by my body for most satellites. Actually, all you may need to do is casually pass your hand over the top of the antenna once every five minutes and the receiver should fail to ever fix ambiguities.

2. Let the receiver fix:

a. Good = Keep the GPS antenna continuously oriented so that the antenna ALWAYS has a clear view of four or more SV’s.

b. Bad = Once every five minutes, make sure the antenna looses sight of all SV’s to force dump the GPS’s lock. All you need to do is point the GPS towards your stomach and lean over it, then take a step and pull it out.

3. Hold the receiver still for 3 epochs before logging a feature:

a. Good = wait five seconds after the receiver stops moving before logging a new point. Log at least 15 seconds over the point. Continue to hold the receiver at the point for another second after ending the shot.

b. Bad = move to a new position and immediately log a point. Or even worse, don’t stop while the point is taken. Just keep moving.

4. Use the correct reference frame:

a. Good = when using WAAS corrections, you are getting ITRF2000. Convert your NAD83 coordinates to ITRF and compare ITRF positions against ITRF positions.

b. Bad = compare WAAS corrections against NAD83 published coordinates. There is an excellent 7-point FAQ on Optimizing Single Frequency accuracy at: ftp://ftp.magellangps.com/Mobile%20Mapping/MM%20Pro/Application%20Notes/Optimizing%20accuracy%20with%20MM%20Pro.pdf

Don’t’ Ignore Reference Frame When using WAAS corrected GPS to achieve sub-meter real-time, the corrections are framed to ITRF00, not NAD83. While ITRF00 is a better datum in all respects, NAD83 is preferred by most all applications in the United States. There typically is about 1 – 2 meters difference between these frames. (See http://www.ngs.noaa.gov/TOOLS/Htdp/Htdp.shtml for the translation program)  Reference Frame mismatch will appear as a consistent offset to users and is often misconstrued as receiver inaccuracy.

If you don't understand Reference Frames, try an internet search on "reference frame geodetic" for plenty of great articles.

Some Common Issues that Ruin GPS Measurements and Projects

1. GPS receivers don't work under trees. GPS antenna needs to be above your head/body. GPS is difficult in urban canyons. If you want to work in difficult conditions, get PM500's with GLONASS. More satellites = more easy = faster = better.

2. Holding receiver with antenna pointed towards the ground (or some direction other than up.) Consider using external antenna.

3. Holding the receiver directly next to a pipe or pole (or under a tree) when measuring position. Consider measuring with offset.

4. Not allowing enough time for position to fix or init while running static or Stop and Go shots.

5. Wrong Reference Frame. Mismatched epoch dates. Wrong datum (NAD83 vs. NAD27). (See above.)

6. Feet or Survey Feet. What is the difference? (1 foot = 0.3048 meters; 1 survey foot = 0.3048006096 meters)

7. Ellipsoid elevation or Orthometric elevation. NAVD88 or NGVD29?

8. Grid or ground distances?

9. Wrong equipment for the job. You can't measure sewer inverts to 1 cm and measure rim shots to 1 meter if you are running a flow study. You need a 1 cm rim shot too.

10. Ignoring productivity: If you are going to shoot 30 shots per day, a static PM3 pair or PM3.RTK or PM500 will all do a great job. But if you want to store 1,000's of points per day or stake hundreds of shots per day, then dual-frequency RTK with GLONASS will make you much more productive.

11. Mission Planning: what's mission planning? There are times during every day when it is probable that you will have a low satellite count. Static shots will take twice as long, it will take 4 minutes to fix a PM3.RTK pair instead of 10 seconds. If you don't want to mission plan, purchase PM500's with GLONASS so you always have plenty of birds in the sky. If you are willing to take a break for 30 minutes and wait for some more satellites to rise, then consider PM3 receivers.

12. Figuring out what you are doing wrong after you have spent an entire summer systematically collecting poor quality data.

Don't fall in to the trap of saving $50 when your purchasing $7,000 worth of equipment and purchase from a dealer who does not know the difference between:

• ITRF2000 and NAD83 CORS 96
• Feet and Survey Feet
• Ellipsoid and Orthometric Elevation
• Grid and Ground distances

Let your dealer know what your intend to do, and listen to their advice. A local dealer is usually going to be a much better resource because they can provide on-site demos and equipment comparisons. But if you don't have a good local dealer, try to find a great national dealer who has experience with your application. Don't hesitate to rent equipment to find out if a particular model will meet your needs. Don't hesitate to take the dealer's training courses.

Before you make a huge equipment investment, make sure that the equipment will meet your expectations before you

Often you will save money by paying an experienced professional to help with your project design and planning! You won't typically get that kind of help looking for the lowest price on the web!