For the following plots, we let our NovAtel PropakII dual frequency reciever log position data at 30 second intervals from aproximately 4pm Eastern Time on Monday May 1st to 8am on Tuesday May 2nd. We used a dual frequency antenna on top of Rhodes hall, Cornell University.
This plot shows the position solution in latitude longitude altitude coordinates (WGS84) - altitude is MSL.
This plot shows the norm of the error vector found by subtracting the Novatel realtime solution from the 'known' position (by known, we mean measured with DGPS Magellan recievers from a GPS control point). It is easy to tell when SA was deactivated, as the error in the solution drops from an average of 32 meters to about 5 meters.
This next plot is a 3D position plot in Latitude, Longitude and Altitude. For this, we plotted only the data taken before local midnight (4am UT) when SA was still activated.
The following is a 3D plot on the same scale as the previous, but using data after 0406 UT on May 2nd.
The following is a 3D plot on the same scale as the previous, but using data from the GPSBuilder2. We ran it from Rhodes Hall antenna 2 on the stealth/scintmon computer for aproximately 2 hours on May 3 2000 from about 1pm to 3pm. As you can see the solution is nearly as good as the one using the Novatel reciever (remember the NovAtel is on RHD antenna 1 and the Builder is on RHD2). The solution is much better than the solution obtained with the Novatel before SA was switched off, so this GPSBuilder2 plot has no SA.
This is the GPSBuilder2 absolute or 'norm' error from the Magellan DGPS solution on the same vertical scale as the Novatel solution above. We can see that the average error is about 11 meters for about 2 hours of data.
