March 14, 2003

Q: What are the accuracies of the CORS sites? Are they 1st order, 2nd order, ....?  Where can I find the position and velocity sigmas?

A: (1) CORS do not have standard classes of accuracy such as first-order. It is better to think in terms of positional tolerances.

"Old" accuracy standards such as "first-order" are related to distance. The Federal Geographic Data Committee (FGDC) has recognized that the accuracy of new survey technology (i.e., GPS) is inconsistent with classical accuracy methodology, (i.e., based on distance). Furthermore, the FGDC recognized that accuracy classification of survey data under the old system is inconsistent with what GIS users want (i.e., local accuracy and network accuracy). The FGDC therefore established new standards for geodetic control networks. You may view the old accuracy standards at:

http://www.ngs.noaa.gov/FGCS/tech_pub/GeomGeod.pdf

See in particular Table 1 on web page 12 (scanned document page 6), formula 1 on web page 14 (scanned document page 8), Appendix B on web page 41 (scanned document page 35), and Appendix C on web page 42 (scanned document page 36) for more information. Formula 1 is reprinted below, illustrating the dependence on distance.

s = sqrt ( e**2 + (0.1pd)**2 )

where, s = maximum allowable error in centimeters at 95 percent confidence level d = distance in kilometers between any two stations p = the minimum geometric relative position accuracy standard in parts-per-million (ppm) at the 95 percent confidence level e = base error in centimeters (this includes station-dependent setup error)

You may view the new standards at "Geospatial Positioning Accuracy Standards, Part 1: Reporting Methodology":

http://www.fgdc.gov/standards/standards_publications/

and "Geospatial Positioning Accuracy Standards, Part 2: Standards for Geodetic Networks":

http://www.fgdc.gov/standards/standards_publications/

(2) Presently NGS does not publish position and velocity sigmas, because GPS formal errors are notoriously optimistic. Instead NGS provides time-series 60-day plots of coordinate repeatability. See

ftp://www.ngs.noaa.gov/cors/Plots/plots.html

Recorded at the top of each plot are the N,E,U discrepancies in centimeters between the published ITRF00 coordinate (represented as the dashed red line) and the average of the 60 daily computed coordinates (represented as blue symbols with realistic errors bars), which are corrected to epoch 1997.0 by the station's published velocity. Recorded too is the RMS repeatablity of the 60 coordinates.

For example at station GOL2 in Southern California, the discrepancy between the published coordinate and the computed coordinate for the last 60 days is -1.7mm north, -13.3mm east, 3.4mm up. This is a measure of accuracy--where the point is on the Earth's surface. The repeatabilty is +/- 1.5mm north, +/- 3.5mm east, +/- 6.0mm up. This is a measure or precision--how well the coordinate repeats.

Station GNAA (Glennallen) in Alaska shows poor accuracy, as a result of the magnitude 7.9 earthquake of November 3, 2002. We are in the process of updating coordinates and velocities for all stations in Alaska, as a result of this earthquake.

Station ASPA (American Samoa) in the far Pacific shows poor repeatability, due to the long baselines from Hawaii and due to the rapidly changing amounts of atmospheric water vapor at either end of this island-to-island baseline.

The 60-day plots thus display the accuracy and precision of a given CORS published coordinate and velocity, in forms consistent with new FGDC methodology. (Note that because CORS define the datum, network and local accuracy are the same.)