| Popis: |
During the last years, different teams have devoted huge efforts to improve retracking algorithms for both conventional and Delay Doppler echoes. For LRM measurements, CLS/CNES has developed and successfully validated a solution called "Adaptive Retracker", implementing a new waveform model and a Nelder Mead optimization method with exact likelihood criterion. Compared to the reference MLE4, the Adaptive brings dramatic improvements over all surfaces. It has been presented several times to the OSTST community [Thibaut et al., OSTST 2021, Ollivier et al. OSTST 2019, Thibaut et al., OSTST 2017] This solution has been implemented as the reference algorithm for nadir echoes in the SWIM ground segment (CFOSAT mission) and in the new reprocessing CNES GDR-F of the Jason-3 mission, with excellent performances as described in [Tourain, 2021] and [Thibaut, 2021], respectively. In the context of different frameworks, historical missions are now being reprocessed with the Adaptive retracker allowing to have homogeneous long-term time-series processed with this up-to-date algorithm. Firstly, in the frame of the CNES GDR-F reprocessing, it is planned to reprocess the whole Jason-2 and Jason-1 time series with the Adaptive retracker, allowing to have more than 20 years of Jason data processed with this algorithm. Poseidon-3B (Jason-3 altimeter) is a replica of Poseidon-3 (Jason-2 altimeter), so the Adaptive can be directly applied to Jason-2. It is not the case for Jason-1’s altimeter Poseidon-2 that has two important characteristics that needs to be considered in a retracker: First, a compression and decompression algorithm has been applied to Jason-1 echoes, changing their shape and therefore the noise statistics, especially on the trailing edge. Second, Jason-1’s platform had pointing issues resulting in numerous periods of strong mispointing. As the Adaptive model is not valid for high mispointing values, and assumes a null mispointing in its current version, the impact of this mispointing must be carefully assessed. Finally, the impact of these two aspects on the neural-network waveform classification must also be analysed as it is used as an input of the Adaptive retracker. Secondly, in the frame of the ESA project FDR4ALT, aiming at reprocessing ERS-1, ERS-2 and ENVISAT Altimeter and Radiometer datasets based on the best state-of-the-art algorithms and corrections, several Thematic Data Products (TDP) will be provided at the end of the project, including Ocean & Coastal Topography dedicated products, and Ocean Waves dedicated products. In the frame of this project, the PTR (Point Target Response) arrays were handled in a new way to allow the Adaptive retracker to be applied on ENVISAT with excellent performances with respect to the current reference algorithm MLE3. After a round robin analysis, the Adaptive has therefore been selected to be the only algorithm providing the Sea Level Anomaly (SLA) for the Ocean & Coastal TDP, and the Significant Wave Heigh (SWH) for the Ocean Waves TDP. The full ENVISAT dataset has now been reprocessed with excellent results obtained for the SLA and SWH. The aim of this talk is to present the methods used to apply the Adaptive retracker on these historical missions in the frame of different projects, and then focus on the excellent performances and results obtained on ENVISAT and Jason-1 compared to the reference algorithms MLE3/MLE4. More specifically, this presentation will address the benefits of this algorithm on historical missions for the SLA and the SWH user communities. |
