The popularity of GPS-based guidance systems for agriculture is increasing. However, no dynamic test standards
are available to evaluate these receivers. In April 2002, GPS receivers were dynamically tested on a 0.8 km length of railroad
track using a small rail cart. The track was surveyed with an RTK GPS receiver and referenced to a NGS benchmark. The
GPS receivers were evaluated over a 24 h period. The tests were conducted in both directions and at two different speeds.
Cross-track and pass-to-pass errors were determined for a John Deere StarFire receiver with dual-frequency correction and
a Trimble AgGPS 132 in autonomous mode. Cross-track errors were sequentially correlated, creating problems for using typical
statistical analysis methods. A Voronoi step interpolation method was used to generate evenly spaced data for Fourier
analysis. Fourier analysis indicated that the cross-track errors contained significant periodic structure below 6 cycles per
day. This implies that cross-track accuracy tests should be conducted for minimum durations of 12 h. Fourier analysis indicated
that pass-to-pass errors were more random than cross-track errors, with no clear concentrations of frequency content,
implying that pass-to-pass accuracy tests can yield meaningful results in less time than required for cross-track accuracy
testing.