Doppler Orbit Error

Operational Doppler orbit error based on May '98 SDRs

The figure below, based on four days of SDR data, illustrates the magnitude of orbit error inherent in the operational orbits. We've plotted up differences between sea surface height and geoid height as a function of latitude. In our initial analysis the 1-second averaged heights were mistakenly time-tagged at the beginning, vs. the middle, of the data frame. This lead to a 441 msec timing error which resulted in a 7 meter twice-per-rev signal. After properly assigning the time to the mid-frame position, the dominant orbit errors are once-per-rev. A fit of the height differences to a bias and once plus twice-per-rev sinusoid leads to the following fit:

dh = h0 + h1*sin(w1*r+p1) + h2*sin(w2*r+p2)

w1 = 1-cyc/rev; w2 = 2-cyc/rev; r = rev (0-360)
p1,p2 phased relative to 0 degrees at 72 N

h0 = -22.4 m

h1 = 5.2 m
p1 = -70 deg.

h2 = 0.9 m
p2 = 41 deg.

The twice-per-rev signal that remains is equivalent to a 42 msec timing bias. Since the onboard GPS system isn't fully operational, there is no precise time stamping on the altimetry data and a timing bias of 40+ msec is actually quite good. There is still a constant offset of ~22 meters (equivalent to an ~130 ns waveform indexing error). Most of the remaining orbit error is due to the slowly varying once-per-rev signal, on the order of five meters.

rev.ps