Sub-meter GPS Receivers can be highly accurate and if properly used can give repeatability in the submeter range. The Trimble AgGPS 132 is the most popular high end GPS used by StarPal HGIS mapping customers. This report is applicable to all sub-meter GPS receivers; we will also identify areas specific to Trimble.  This report will give you tips that GPS Manufacturers won't tell you. You don't need to be an engineer to understand this report. Learn how to get sub-meter results, as well as how not to. We will explain all of the common terms needed to accomplish successful sub-meter mapping.

95th Percentile / 2DRMS-95%
This specification is often used for (legal or cadastral) survey mapping applications.  It is the most appropriate specification for mapping with HGIS. This specification is called by various names: 2DRMS-95%, 2-sigma, 2 standard deviations, HRMS-95%, or the 95th Percentile.   Other GPS specifications are not as precise, in particular RMS, 2DRMS, and 2d-sigma.  These other specifications are not as useful when creating maps with HGIS.  Please ignore GPS terms that have NO specifications such as "Sub-meter GPS".

Satellites
You need to know is that there are Global Positioning Satellites up in space.  You don't need to know where they are.  If you are mapping a difficult canyon where you can only see a small portion of the sky, you will want to know where the satellites are.  But if you are just using HGIS to map fields daily, you don't care where the satellites are.  The satellites rise, travel across the sky, and set like stars and planets. They are "someplace" up there.

View Of the Sky
The accuracy of your GPS depends upon how many satellites it sees. Most of the time, there are at least 6 satellites visible if your GPS can see the whole sky.  At certain times, there could be as many as 12 satellites. Before you shut down HGIS after mapping, you can look at the GPS Statistics Screen for the numbers for "Minimum Sats" and "Average Sats"  These numbers will tell you how many satellites the GPS saw while you were in the field.   For short runs, use the Minimum Sats value.  For runs of 4 hours or longer use the "Average Sats" value. From these numbers, the following table will give you an idea of how much of the sky your GPS could see.  If the GPS is mounted where a person (or something else) blocks part of the sky, your satellite count will be lower.   Hills, trees, buildings all reduce the view of the sky.   A Premium View of the sky doesn't happen every day because conditions change from day to day.

HOW TO DETERMINE YOUR EFFECTIVE VIEW OF THE SKY

A heavily overcast sky can reduce your view of the sky by one level. If your GPS normally has a Full View of the Sky, but the sky is heavy overcast; you will be reduced to Fair.  A person sitting next to the GPS can reduce the view of the sky by as much as two levels. If you mount the GPS too low on a vehicle; for example directly behind a person on an ATV, or below a grain bin, you can reduce the GPS view of the sky be two levels.  With clouds and someone/something blocking part of the GPS view of the sky can reduce the view from Premium all the way to Poor.

Correct GPS Mounting

Basic GPS and View of the Sky
Before we talk about "Differential", let us first talk about a basic GPS Receiver.  The following graphs show you the accuracy you can expect during periods when the GPS cannot receive a differential signal (In this example, the Trimble AgGPS 132 set to an inactive frequency).  With a Premium View (7 - 10 satellites), we can get accuracy in the 1.4 to 3 meter range. However, with a Poor View, we only get accuracy in the 3 - 5.2 meter range.  The main point: the better its view of the sky, the better your GPS will work.

This was a relatively short 72 hour test so the bump is not meaningful. Each added satellite will improve your overall mapping accuracy.

Differential Correction
A sub-meter GPS can receive data from earth or sky based transmitters to correct its location. This is called "Differential Correction".  The plot below is of the Trimble AgGPS 132 with four different sources of differential correction and a Premium View of the sky. The yellow is uncorrected, and the blue is corrected. At two meter accuracy, with the AgGPS 132, it doesn't matter very much what we do.  As we look for better accuracy, see that the blue (corrected) circles get smaller, and also begin to separate from each other.

Below are the sources available at our test location to the AgGPS 132, with the accuracy of each source.  Your receiver may accept one or more of these sources of differential correction.

Using a different source will put you in a different location.  If you recreate this plot, your plot will be laid out differently.  The reason the average satellites was below 8.5 was due to cloudy conditions.   The averages shift from day to day, and circles grow and shrink with the weather (and other factors).  Multiple tests were run on some sources (dotted lines). Each circle was collected for 24-72 hours.  Note that to be within one meter, you must always use the same source of differential correction. 

Radio Beacons
Radio Beacons (also called Coast Guard Beacons) are available over most of the United States, and nearly all of the world’s coastline.  These may also be referred to by the specification RTCM-104.  Coverage maps for the US are available at http://www.navcen.uscg.gov/dgps/coverage/Default.htm   If you have "Dual Coverage" (two beacons); set the more accurate beacon as primary.  Beacons are not guaranteed to be sub-meter, but usually are repeatable within one meter horizontal relative to their individual long term average. When compared with other beacons, they will be off as demonstrated above.  The 307khz beacon has been specifically adjusted to NAD-83, but you cannot assume this with your beacon because beacons are not guaranteed to do this. If you cannot maintain solid differential correction signals using beacons, then you have to choose another type of differential correction.

OmniSTAR
OmniSTAR is a private paid subscription service for differential correction.  It has basic VBS and premium services (XP and HP) available. The AgGPS 132 (plot above) is of the basic VBS service. Use one of these services if you need sub-meter repeatability but cannot reliably receive Radio Beacons or WAAS. Or if you need to use maps that were collected with OmniSTAR  OmniSTAR is a wide area single source with repeatable accuracy, but may not be available everywhere. And, OmniSTAR comes with an annual subscription cost.

WAAS / EGNOS
These satellites hover in stationary positions over the equator.  WAAS was designed as an aid to aircraft in the United States.  EGNOS is in use in Europe.  Their primary purpose is for navigation in the air. In the US, the Eastern WAAS signal is strongest in Florida and weakens progressively at you move North and West. The Eastern WAAS satellite is about 20 degrees above the horizon in Colorado. There is also a Western WAAS satellite for use in California. The WAAS signal can be blocked by trees, hills, heavily overcast sky, buildings, people, as well as obstructions on the vehicle to which the GPS is mounted. On flat ground at our test site, a 36 foot tree 100 feet away can block the WAAS signal (a shorter tree if it is on a rise). If you plan to use WAAS, find out how high your satellite is above the horizon.  Then check out your location for low obstructions or hills in the direction of the satellite. At our test location, WAAS is drifting 0.025 meters per year. This may not mean too much next year, but in 40 years it will be one meter off. Even still, WAAS will stay inside its US Government specification of 7 meters for 250 years.
AgGPS 132 WAAS Conclusion: The AgGPS 132 is more accurate using No Differential than most GPS receivers that use WAAS.  Mounting the antenna 12-24 inches above any sheet metal present will improve WAAS reception.

No Differential
The yellow No Differential plot is included for comparison. This is a premium view of the sky. This circle of uncertainty will grow rapidly with decreasing views of the sky, as shown in the first plot above.

Differential Correction and View of the Sky
This is an example of a GPS with WAAS differential correction (Trimble AgGPS 132).  The GPS only needs one source of correction data but it needs it solid all the time. [Technically, you can have up to 30 second gaps before accuracy starts to degrade.]  If you lose the differential correction signal, your repeatability will drop to the "No Differential" plot above.  The view of the sky hasn't changed, but you can see the improvement of differential correction. Most GPS receivers can't correct to submeter if the number of satellites falls below 6.
If you have a full view of the sky or better, you will be less than 1 meter (the red line).

This is was a 244 hour test with WAAS.  A 1000 hour test would probably be a little worse, but still submeter..

How Bad Can It Get?  - Multipath Interference
Multipath is something GPS manufacturers may mention briefly. However, they don't publish their own test results, because all GPS receivers look bad. Multipath interference can be caused by any conductive surface - water, ice, sheet metal nearby or on the vehicle. I haven't seen multipath interference from a person, but it is probably possible.  We created a multipath test site to see how bad it can really get. The results at our site will be different from any other multipath test site. However, such tests can be useful for the following purposes:
A) We can learn what multipath interference looks like. We can recognize multipath when it shows up in our maps (as it certainly will).
B) We can compare how two different GPS receivers perform in the same multipath environment (both GPS receivers must be run for 24 hours at the identical spot, but not at the same time).
C) We can see how bad performance can really get. Some GPS receivers we have tested have jumped by hundreds of meters during multipath interference.

TYPICAL MULTIPATH INTERFERENCE PLOT (Trimble AgGPS 132)

This is a tough 24 hour multipath test.  The red circles identify examples of multipath interference.  The GPS will suddenly jump 10 meters (or more).  Each time during this test, the GPS recovered and jumped back within 80 seconds.  In this plot, there are other examples of multiple overlapping multipath interference, but these three examples are clearly separated by wide jumps.  These three cases are typical of what you will see in the field and in your maps.  You will be drawing a boundary, and all the sudden the GPS will say you are on the other side of the ditch.  Some number of seconds later, the GPS will jump back to the correct location. It could be because of water in the ditch, a metal fender or roof of your vehicle, a nearby car or building, etc.  The factor that most affects multipath interference is the satellite count. Depending upon your environment and view of the sky, this may happen anywhere from once a week to several times a day.  The occurrence of multipath decreases as the number of satellites increases.  Improve your GPS Receiver's view of the sky and your occurrence of multipath interference will decrease.   [Technical users: HDOP cannot predict inaccuracy due to multipath interference.]

Trimble AgGPS 132: All GPS Receivers are susceptible to multipath interference. For multipath interference rejection at our test site, we rated the AgGPS 132 as significantly better than the Thales MobileMapper CE.

What Repeatability Can I Actually Get?  What does this mean to me?
Let's say you locate an old well (or field boundary) and mark it on your map. Then you send someone else out to dig (or fertilize) using that map.  Because we are comparing two GPS readings, we have two sets of GPS errors.  If both users are using OmniSTAR (2DRMS-95% = 0.59 meter), then the second user has a 95% likelihood of digging within 0.89 meters of the original location. [The total error of two GPS readings is (1.4 - 1.7) times the error of a single GPS reading because the scatter is an ellipse. We used a value of x1.5]   The error is worse if other discussed conditions exist or different models of GPS Receivers are used. Note that using the worst combination of differential sources resulted in an error almost as bad as no differential at all. 

THESE ARE FIELD REPEATABILITY TESTS
THESE RESULTS ARE NOT THE SAME AS MANUFACTURER SPECIFICATIONS.

Note 1: Does not meet GPS Manufacturer's requirements for submeter. 

A full view of the sky and using the same differential source are the most important factors. If you do these two things, you will usually be within 1.5 meters. If you test your differential sources, then you can obtain repeatability of about one meter. At our test site, our average number of satellites was 8.0 - 8.5 (Premium).  Some locations or days, you may only average 7 satellites; those periods even with the better differential sources you will only achieve 2 meter repeatability.  And if part of the sky is blocked, it could be even worse.

Sunspot Cycle
The Sunspot cycle also affects GPS accuracy.  The year 2006 is a relative lull for sunspots and accuracy will be very good.  Each year after 2006 will be slightly worse.  Sunspots will peak in the year 2011 with GPS accuracy being at its worst.  Submeter accuracy will fluctuate with sunspot activity.  After 2011, accuracy will get better again. We don't have any tests to prove what GPS accuracy will be in 2011. We think perhaps normal sunspot activity in 2011 could be similar to reducing view of the sky by one level; while the worst sunspot activity could be similar to reducing the view of the sky by two levels. There will be one (or more) new generations of GPS receivers by 2011, so they may compensate for this loss of accuracy to some extent.

Running your own tests
HGIS provides a GPS Statistics Screen to help you run your own stationary tests to compare your local differential sources. You can use this test to compare different GPS Receivers. To be valid, any tests must be run for multiples of 24 hours (24, 48, or 72 hours). After the test period, the GPS Statistics screen will indicate the 2DRMS-95% results for that GPS, Differential Source, and test conditions. Use the Plot Circle button after the test period to plot the 2DRMS-95% circle. The Average Number of Satellites ("Sats = min / avg / max) will give you the view of the sky.  Differential % will tell you if your Differential Source is solid.  Average Number Of Satellites and Differential % may also be used after field operations just to estimate your view of the sky in that field and tell if you were receiving solid differential corrections.

Trimble AgGPS 132 Results
The AgGPS 132 is recommended, but you have to use it properly for best results. The internal Beacon Receiver is among the best we have tested. With regard to WAAS, the AgGPS 132 is not as good at tracking WAAS as other WAAS GPS receivers we have tested.   At a WAAS satellite elevation of less than 20 degrees, the WAAS receiver is marginal.  However, the AgGPS 132 is the best GPS receiver we have tested with regards to multipath rejection. Follow the general instructions and guidelines here for all sub-meter GPS receivers and you will get accuracy with the AgGPS 132 of about one meter.

Conclusion
For greatest accuracy, when installing a GPS obtain the fullest view of the sky and minimize nearby sheet metal.  If using WAAS, raise the GPS 12-24 inches above any sheet metal present, and higher than any objects on the vehicle. Other antennas should be mounted at least 1 meter away from the GPS. If you are sending your maps to a contractor to do some work for you, specify the GPS model and differential correction source to be used.

We hope that using the information in this report will make a difference in your results.

Happy Sub-Meter Mapping!
for more information please visit:
http://www.starpal.com


TECHNICAL APPENDIX

2DRMS-95%
GPS has systematic non-random errors, therefore position errors do not fit the standard "Bell Curve". And the curve changes depending upon your location.  Because of this, we use the mathematical value of 2-sigma which is a 95.44% confidence level.  Surveyors sometimes use a 95.00% confidence level, which gives nearly the same result as 95.44%. 

GPS Manufacturers often use a similar sounding specification - 2D-RMS or 2dRMS which has a confidence level ranging from 63% or 92% - 98%.  WARNING: Some GPS Manufacturers use the definition: "2-Dimensional Distance Root Mean Squared (2dRMS), as defined in STANAG 4278, is the radius of a circle that contains 63% of all measurements", which yields a 63% confidence level.  Other manufacturers say  "2dRMS is defined as two times the distance RMS error", which at our test location is around a 92% confidence level.  Some GPS Manufacturers claim that their test locations yield as high as a 98% confidence level.  Be cautious of manufacturer's 2DRMS specs.  Always look for the "-95%" at the end.  

2DRMS-95% is defined as follows: If certain specified conditions are met (such as Number Of Satellites, low multipath environment, etc.), then 95.44% of the time you will be within this horizontal distance of some precise location.  All necessary conditions are required to be specified.  The second problem is defining what the precise location is; because in our tests the precise location varied by up to 1.5 meters (not submeter). We can generally assume that the precise location is the long term average using that very same GPS, the very same setup(including differential source), and very same atmospheric conditions, and averaged 1 month or longer.  Later, when you return to that precise location, measure how close you are 95.44% of the time.

DATUMS
The GPS reports position relative to the Datum of the Differential Source. This is the position recorded by HGIS.  For example with WAAS, this is ITRF-2000 (also called ITRF00). If you always use the same differential source, then you will have highest accuracy (to that Datum).  Sources with different Datums will drift relative to each other over time.  In the US, beacons are relative to the NAD-83 Datum, but not always at the sub-meter level. So you cannot use data from one beacon with data from another beacon at the sub-meter level.

HDOP
HDOP and the related GDOP, PDOP, EDOP, NDOP, TDOP, VDOP are unit-less numbers (we call them mystery numbers) that have very low correlation to your actual accuracy "right now".  GPS receivers that estimate your error are also using these mystery numbers. First, they do not consider your differential source, which you see from this report has a large effect.   We prefer the easier to visualize "Number Of Satellites" -- you know exactly what this means.  There is some merit, but more confusion when using both concepts together: DOPs plus Number of Satellites. If you use GPS planning software to determine the best time and location to collect GPS data and know that you have a full view of the sky; then DOPs are useful. Unfortunately, planning software doesn't take into account "View of the Sky", which is the overriding influence on actual DOPs.  If you do not have a full view of the sky because of location or installation, then planning software is not as useful.  Few HGIS users have the luxury to decide when or where to collect GPS data, so the view of the sky is much more easily to calculate, visualize, and correct.

Satellite Health
Normally, there are enough satellites visible for submeter accuracy.  However, sometimes one or more satellites that should be visible, may not be available for periods of maybe 30 minutes to one hour in a given day (Usually just one satellite at a time).  These satellites are marked "Unhealthy" and the GPS will ignore them.  This can also reduce your accuracy and average number of satellites during those periods.

GPS Testing Issues
US Beacon operators say the beacons are operating as designed.  You may get sub-meter with some and others you won't.  The beacons are specified to be within 10 meters of the NAD-83 Datum.  The GPS Manufacturers view is that the government should update beacons to work with sub-meter GPS receivers.  The governments view is that submeter is not in the specification.  Our view is that GPS Manufacturers should publish what beacons they have actually tested and approve.  Otherwise, poor accuracy will be the downfall of beacons once users find out beacon's true accuracy.  For more info on this, see http://www.navcen.uscg.gov/gps/WAAS-DGPS.htm

You can see the 307khz beacon lines up with OmniSTAR within 0.1 meter (at NAD-83). The 310khz beacon does not. Both beacons are the same distance from our local test site. They both have the same signal strength, the same baud rate, and the same signal/noise ratio. In other words, they appear identical to the user.  Trimble used the 307khz beacon at their local test site (about 50 miles from our test site) with OmniSTAR for published test results; but not the 310khz beacon.   We believe that this 307khz beacon was adjusted to NAD-83 at the sub-meter level for improved results with submeter GPS Receivers, even though sub-meter accuracy is not specified for this beacon.  The 310khz is an older beacon site that has not been updated to NAD-83 at the sub-meter level (perhaps it has older hardware). But remember, sub-meter accuracy is not specified for any US beacon.

In trying to reproduce GPS Manufacturers tests, we find we must add additional unspecified conditions, such as a "non-overcast sky", "use this particular beacon", or "only test on certain days".  Note in the above plot that WAAS shifted 0.3 meters during a 24 hour test that was coincident with a cold front coming through.  GPS specifications are like the Mileage Rating on your car; "Your mileage may vary".  Independent GPS testing is important to know how to get the best results possible. That is why we do it.

WAAS is also functioning as designed and was also not designed for sub-meter.  Don't assume that any differential source is sub-meter. If you plan to use beacons, test all of the beacons you can solidly receive using a 24 hour test for each beacon using the HGIS GPS Statistics screen and the Plot Circle function. The above plots were created using this method.  Also check the Percent Differential statistic to see how solid the differential correction source is. You can then determine which differential sources will work best for you.

WAAS Static Test [Trimble AgGPS 132 specific test] - A level AgGPS 132 antenna in an open field away from any vehicle or obstruction registered an (s/n) reading of 5 for the WAAS satellite.  Mounted sixteen inches above a metal vehicle roof, it read about the same as open field.  The GPS needs a minimum reading of 2 to use WAAS, and it did track WAAS consistently in our static test. Using the Magnet Mount on the metal roof, the reading was 3-4.  A small tilt (15 degrees) of the vehicle reduced the reading to 2-3.  We are near the limit with the Magnet Mount particularly at a slope.  If you point the top of the AgGPS 132 antenna directly at the WAAS satellite, you get a reading of 12+ which shows a strong WAAS signal. [A signal strength of 12 has four times the power of a signal strength of 3.]  The WAAS signal is strong but the sensitivity of the AgGPS 132 antenna at this low angle reduces the signal strength when the antenna is magnetically mounted on a metal roof.  If you want to use WAAS, Trimble recommends mounting the AgGPS 132 Antenna 12-24 inches above any sheet metal present. From our test you can see that this will improve WAAS reception.   Locations where the WAAS satellite is higher above the horizon (such as southeast US) will receive a better signal, even with the Magnet Mount. If the satellite elevation angle is lower than 20 degrees at your location, then you should mount the GPS 12-24 inches above any sheet metal present, and should have flat ground with no hills or obstructions.

AgGPS 106 WAAS Dynamic Test - The Trimble AgGPS 106 is specified at 1-3 meters, not quite sub-meter.  We ran this test on the AgGPS 106. The AgGPS 132 is more sensitive than the AgGPS 106 and should perform better. We ran the AgGPS 106 in a typical field 30 minutes mounted 16" above a sheet metal roof. The AgGPS 106 tracked WAAS 80% of the time with maximum differential age of 42 seconds. The AgGPS 106 would be suitable for mapping where accuracy is not quite as important and the gaps in WAAS don’t cause as large percentage error.

MultiPath Test
Multipath jumps on the AgGPS 132 tend to be in the range of 5 - 25 meters, much better than other GPS receivers we have tested. This is primarily because of the Antenna Element design of the AgGPS 132.  When we tested it with an external CSI Patch Antenna, the AgGPS 132 was significantly more susceptible to multipath interference.

Does the AgGPS 132 meet the 2DRMS-95% GPS specification for sub-meter?
The answer is Yes, IF...
1) The GPS has a Full or Premium View of the sky, and
2) You always have Differential Correction, and
3) You always use the same Differential Correction Source, and
4) You have a low Multipath environment.

TESTING DIFFERENTIAL SOURCES UNDER PREMIUM CONDITIONS (Average Number of Satellites >= 8)

HGIS^® is a Registered Trademark of StarPal Inc.