NGS SURVEY MARKER ACCURACY
Text by Dave Doyle and Ed McKay (NGS) 4 May 2000
Organization: National Geodetic Survey

The standards are as follows:

A-Order: 1 part in 10,000,000
B-Order: 1 part in 1,000,000
1st-Order: 1 part in 100,000
2nd-Order Class I: 1 part in 50,000
2nd-Order Class II: 1 part in 20,000
3rd-Order Class I: 1 part in 10,000
3rd-Order Class II: 1 part in 5,000

Translation of proportional accuracy to spatila accuracy gets confusing.  These standards are the legacy of nearly 200
years worth of conventional (angle and distance -- line of sight) surveying methods and relate the accuracy as a function of distance.  For example 1st-order, 1 part in 100,000 (1:100,000) means that the accuracy between any 2 points of will not be worse than 1 unit in 100,000 units -- e.g. 1 meter in 100,000 meters, 1 foot in 100,000 feet, 1 inch in 100,000 inches etc.

Unfortunately there is no way to convert a proportional accuracy directly to a spatial accuracy - 1 foot, 1 meter, 1 cm etc. without performing a comprehensive national readjustment of the entire National network.   This is being planned for about 2003 or 2004.  Until then we will have to live with approximations.   These values can be "estimated" as a function of each stations relationship to the local network as and would be consistent with differential GPS applications.

A-Order:   3 cm (0.10 foot)
B-Order:   5 cm (0.16 foot)
1st-Order:  10 cm (0.33 foot)
2nd-Order Class I:  20 cm (0.66 foot)
2nd-Order Class II: 30 cm (0.98 foot)
3rd-Order Class I: 50 cm (1.64 feet)
3rd-Order Class II: 1 meter (3.28 feet)

It sounds like your users are trying to make an "absolute" assessment using point-positioning techniques.   If that is the case, then you must also consider the relationship of the North American Datum 1983 (NAD 83) coordinate system relative to it's relationship to earth-mass-center which is approximately 2 meters (3-dimensionally).  If this is the scenario you're dealing with then virtually any station in the reference frame -- regardless of it's positional accuracy with respect to the NAD83 -- should be approximately 2 meters.

------------------------------------------------------------------------

NOTE:  The NGS has a full explanation of the above 2m difference between an NAD-83 position and WGS-84 at:
ngs-accuracy.txt

Summarizing that document, improvements in GPS and other space-based positioning technologies (e.g. Very Long Baseline Interferometry, Satellite Laser Ranging etc.) have improved our knowledge of earth center to approximately 4-5 cm.  This information is being implemented in the GPS coordinate system quicker than NAD-83.

NGS notes that, "NAD 83 has had several improvements of the international precision using GPS.  These improvements are referred to as the High Accuracy Reference Networks (HARN), and Continuously Operating Reference Stations (CORS).
Consequently the internal or relative relationships of these points in NAD 83 are at the 1-4 cm level.  However, we have not changed the origin or orientation of  NAD 83 as has been done with WGS 84 (G873).  Therefore, the "absolute"
difference between points in NAD 83 and WGS 84 are approximately 2 meters."

So, measuring GPS accuracy on NGS markers will apparently be limited to 2m unless one applies the mathematical transformation between the two datums.

I notice that by switching between NAD-83 and WGS-84, the waypoints stored in Garmin GPSs do not change their UTM position, which  presumably has a 1m resolution, nor the recorded altitudes which have a one-foot resolution.