You must have a javascript-enabled browser and javacript and stylesheets must be enabled to use some of the functions on this site.


Analysis of Radar Altimeter Waveform Retracking Algorithms for Geodynamics Studies

Hyongki Lee(1) , C.K. Shum(1) , and Yuchan Yi(1)

(1) Ohio State University, Laboratory for Space Geodesy and Remote Sensing, 43210 Columbus, United States


This study tests the potential use of NASA’s radar altimeter TOPEX/POSEIDON to potentially detect solid Earth geodynamic deformation signals such as the Glacial Isostatic Adjustment (GIA). Our study region covers a relatively flat land area where the maximum solid Earth crustal uplift of ~1 cm/yr due primarily to the GIA occurs near the Hudson Bay. We choose to conduct collinear analysis along a single TOPEX track using data from TOPEX’s 10-day repeat cycles over the study region. A mean ground track has been generated using TerrainBase global DEM. Even though TOPEX can maintain lock over moderately smooth land surface, the return waveforms over different surfaces/conditions, including ice, land and vegetations are complicated and therefore the accuracy of topographic recovery will typically be at the meter level. To obtain more precise radar altimeter measurements over land, we first categorized each waveform according to its shape and number of ramps and interpreted the corresponding distinct surface features, and then applied several waveform retracking algorithms to assess their respective appropriateness. These algorithms include the NASA V4, the surface/volume scattering model, the ESA/UCL COG, the threshold method, the cross-correlation technique and ESA’s ICE2 algorithms. We have generated a mean height profile using each of the available retracking algorithms, and then calculated deviations to determine which methods agree with each other. We found most of the returned waveforms are more or less specular and contain a pre-leading edge bump which is likely due to the topography. There is no way a model-based retracker such as S/V model can account for this problem since the algorithm assumes a flat surface. Since NASA V4 fits a waveform to a function whose variables are estimated to achieve the best fit, it performed well on these topographically noisy waveforms and waveforms with fast decaying trailing edge. Unlike the 10% threshold algorithm which has been used to calculate ice-sheet height change at cross-over points, a 20% threshold level has been proven to be adequate since it reduces the effect of this pre-leading edge noise. Another model-independent cross-correlation technique calculates retracking differences from the non-zero-lag of the correlation function between a pair of waveforms. Optimal cross-correlation coefficient and spread value have been determined to reject dissimilar waveforms. The advantage of this algorithm is that it can work for any pair of similar waveforms. After retracking each radar altimeter measurements, we chose a relatively flat area to perform spatial averaging and 3-sigma editing to reduce the effect of data outage. Finally, we will present preliminary results using the retracked TOPEX waveform data for potentially measuring the deformation near the Hudson Bay due primarily to the Glacial Isostatic Adjustment.



                 Last modified: 07.10.03