Abstract's details

Fast-Adaptive: a new, optimal, unbiased, and computationally efficient retracking solution for the analysis of Conventional Altimetry data

Anna Mangilli (CLS, France)

CoAuthors

Salvatore Dinardo (CLS, France); Fanny Piras (CLS, France); Thomas Moreau (CLS, France); Claire Maraldi (CNES, France); Jean-Alexis Daguze (CLS, France); Pierre Thibaut (CLS, France); François Boy (CNES, France); Picot Nicolas (CNES, France)

Event: 2022 Ocean Surface Topography Science Team Meeting

Session: Instrument Processing: Measurement and Retracking

Presentation type: Type Oral

The improvement of retracking algorithms aiming at the optimal and efficient estimation of the geophysical parameters is a core activity within the altimetry community. The importance of having an optimal, robust and efficient retracking solution is indeed more and more critical with the increase of the amount of data and the high demanding requirements in terms of data quality and resolution. The accuracy of the retracking solution in retrieving the geophysical parameters has a strong impact on the accuracy of the estimation of Essential Climate Variables as the mean sea level, which is a key issue for trend studies and for the robust assessment of climate change at local and global scales.

Currently, the most used retracking solution implemented in the ground segments for conventional altimetry is the MLE4 retracking algorithm, which is computationally fast but known to be sub-optimal and biased, therefore needing corrections that must be applied to avoid systematic bias in the retrieved parameters. Recently, lots of efforts have been done to improve this solution in terms of estimator, modelling and inclusion of instrumental-related effects (numerical PTR), leading to the development of the Adaptive retracker (Tourain et al 2021, Thibaut et al. OSTST 2017 & 2021) which is now successfully integrated in the ground segment of missions like Jason3 and CFOSAT. The Adaptive retracker is based on a Maximum Likelihood Estimator with the exact formulation of the likelihood function for a Gamma distributed multiplicative noise and provides huge improvements in the parameter estimation with respect to the MLE4 solution. Yet, while being optimal and unbiased, the Adaptive retracker is not numerically efficient as the optimisation of the exact likelihood criterion is done with the Nelder-Mead algorithm that has a very high computational cost. This can be a big issue, preventing for instance this innovative algorithm to be included in “near-real-time” and “real-time” official products.

In this talk we present Fast-Adaptive: a new, optimal, unbiased and computationally efficient retracking approach for the analysis of conventional altimetry (Low Resolution Mode, LRM) data. We will present the formalism, describing the estimator, the optimisation method and the model, which is based on the four parameters “Adaptive-like” model with numerical PTR convolution. We will demonstrate that the proposed retracking solution provides with unbiased and optimal parameter estimation, compatible with the Cramer-Rao bounds, while keeping a low computational cost, comparable to the MLE4 retracker. We will present the validation of the Fast-Adaptive retracker on simulations and we will detail the results of the analysis of representative Sentinel-6 and Jason-3 LRM data sets with the new retracking solution and show the comparison with the existing solutions (MLE4 and Adaptive) on the same data.
The Fast-Adaptive retracker is a new promising and powerful tool for the analysis of LRM data of current and future radar altimetry missions.
 

Oral presentation show times:

Room Start Date End Date
Sala Grande Tue, Nov 01 2022,11:25 Tue, Nov 01 2022,11:40
Anna Mangilli
CLS
France
amangilli@groupcls.com