All altimeter data over the ice sheets must be post-processed to produce accurate surface elevation measurements. This post-processing is called "retracking" and is required because the leading edge of the ice-sheet return "waveform" deviates from the on-board altimeter tracking gate, causing an error in the telemetered range measurement. A return "waveform" is the received power sampled at the satellite and results from the interaction of the altimeter's transmitted pulse with the scattering surface or volume directly beneath the altimeter. Retracking altimetry data is done by computing the departure of the waveform's leading edge from the altimeter tracking gate and correcting the satellite range measurement (and surface elevation) accordingly. The following figure illustrates this concept.
There are many different methods described in the literature that are available for retracking ice-sheet altimeter data. All the various retracking algorithms have their own unique advantages and disadvantages, and it is generally accepted that no single algorithm can meet the diverse needs of all ice-sheet altimetry data users. Thus, as a result of a workshop held at NASA/GSFC in November, 1995, several retracking corrections are available on the ice altimetry. The GSFC retracking is based on a fit of the waveform to a 5 or 9 parameter function. The GSFC retracking should be used when using the altimetry data for topography as it gives a more accurate picture of the surface profile. Several different threshold retracking levels are also included on the level 2 data. Threshold retracking is based on setting the retracking point at a percentage of the waveforms maximum power. This retracking algorithm is most useful when repeatability of measurements is necessary, such as time-change studies of ice sheet surface elevations.
The differences between the different threshold retracking values and the GSFC functional fit are a function of which threshold value is chosen and the type of surface. We have made an attempt to quantisize this by calculating the mean and standard deviations of the differences for each 500m elevation band over Greenland. The GSFC V4 - threshold 20% differences show that the 20% threshold is the closest approximation to the functional fit, but there are significant biases ranging from 19 cm over the higher plateau regions, to 89 cm in the ablation zone. The GSFC V4 - threshold 10% differences are much larger varying from 62 cm over the plateau regions to 2.9 m over the ablation zone always biased so that the elevations using the 10% threshold retracker will be higher than those calculated using the functional fit. The GSFC V4 - threshold 50% differences show that the 50% threshold elevation is is always lower than that calculated with the functional fit with biases ranging from 90 cm over the sea ice to 3.5m in the higher elevations of the ablation zone.