CoAuthors

Pierre Thibaut (CLS, France); Fanny Piras (CLS, France); Sophie Le Gac (CNES, France); François Boy (CNES, France); Nicolas Picot (CNES, France)

For many years, the extraction of the geophysical parameters from LRM altimeter waveforms has been ensured by MLE-4 and/or MLE-3 algorithms. Both retrackers use a Maximum Likelihood Estimation method degraded to a least squares estimator to fit a Brown model to the radar echoes. MLE algorithms were originally designed to account for speckle noise statistics in their convergence criterion, but the use of a Newton-Raphson algorithm to optimize this criterion actually degrades the estimator into a simple least squares.

With the development of the Numerical Retracker and the introduction of the real impulse response in the Brown model, the resulting numerical model is closer to the returned radar echo and allows improving the estimation process. We propose in this talk to introduce a strong improvement of the optimization method by performing a true Maximum Likelihood Estimator fully exploiting the speckle noise statistics and thus improving the estimation performances. With this new algorithm, high rate data (20Hz for Jason and 40Hz for Saral) can be fully exploited with an estimation noise reduction of about 10% for the altimeter range and 60% for the wave height. As for the numerical retracker already presented in past OSTST, this new solution does not need look-up correction tables anymore. Implications on SAR altimeter waveform estimation will also be presented.