Land altimetry using satellite data from the GEOSAT sea altimeter

Several techniques are proposed or assessed from the GEOSAT sea altimeter, and changes are recommended in future RADAR altimeters. The first technique tried was to cross-correlate each waveform with the preceding one. Then the position giving the maximum correlation was taken as the correct placement of the new waveform in its data window. The resulting altitude profile was slightly more variable than that of the on-board tracker over flat rock (Salar de Uyuni), so apparently successive waveforms are too dissimilar to correlate. When cross-correlation failed, compensation was made for averaging. Raw waveforms are averaged in groups of 100 and it is these average waveforms which are available for ground processing. But while the 100 are being received, the window position is moved in time, at different constant rates for the first 50 and last 50. Assuming an unchanging waveform and a single window rate for the 100, first and last waveforms were obtained. Averaging their half-height arrival times gave an altitude profile similar to the median and centroid methods. The variability of this altitude profile suggests that even raw waveforms in a group of 100 may be too dissimilar to correlate. In other techniques, the pulse was judged to have arrived when one of these criteria was met: the wave amplitude meets a certain absolute threshold; wave amplitude meets a certain relative threshold; certain fraction of the area of the waveform has passed; and centroid of the waveform has passed. All methods gave altitude profiles at least as variable as the on-board tracker, and all were biased at least 0.5 meter to low altitudes, except the threshold detector. The threshold detector can be filtered spatially to resemble the on-board tracker, and perhaps it could be implemented without feedback. For operating over land, these changes are recommended in future altimeters: The window should not move while raw waveforms are being averaged; some of the raw waveforms should be telemetered to earth for study, placing the waveform in the window should allow for shape variety; and some form of threshold detector might be used to find the signal and place it.