Validation of Retracked Sentinel-3 Altimetry Observations along the Norwegian Coast

Sentinel -3 is the first satellite altimetry mission that operates exclusively in SAR (Synthetic Aperture Radar) mode providing an alongtrack spatial resolution of 300 m. The number of reliable observations close to the coast may be hampered due to complicated waveforms that are difficult to interpret or from inaccurate geophysical corrections. In order to make use of altimetry data with a better spatial resolution and as close to the coast as possible, several specialized altimetry products have been developed. One such product is based on the Adaptive Leading Edge Subwaveform (ALES) retracker, which aims to provide more accurate sea level observations in coastal areas. In this study, which serves as a preparatory step for the determination of an updated mean sea surface model for Norway, we have tested the performance of Sentinel -3 observations along the Norwegian coast retracked using both the ALES-retracker as well as the standard open -ocean SAR retracker (SAMOSA2). Both datasets were compared with observations from the Norwegian network of permanent and temporary tide gauges as well as with the numerical ocean model NorKyst-800 to test their temporal and spatial variability, respectively. From the validation in terms of temporal variability we found that the SAMOSA2 dataset provides more observations closer to the coast than data retracked with ALES because lower -quality observations (most of which are located inside fjords) are excluded from ALES during the retracking process. Subsequently, this leads to increased accuracy for the ALES data. In comparison to observations from tide gauges, ALES shows standard deviations of differences of 10.8 cm (permanent tide gauges) and 14.2 cm (temporary tide gauges) while the original Sentinel -3 data show standard deviations of differences of 18.8 cm (permanent) and 17.8 cm (temporary). Regarding spatial variability, we found that the performance of these two datasets depends on a chosen area for investigation. Compared to a group of tide gauges towards the open sea, ALES outperforms the original dataset, which is possibly linked to the many brighttarget radar echoes in this area. For the tide gauge group in the fjords and at permanent tide gauges, however, the original dataset outperforms the ALES dataset, likely due to the sparsity of observations in the latter dataset. Centered root -mean -square differences (RMSDs) to the tide gauges range between 4.0 cm and 7.0 cm for the ALES dataset, and between 3.6 cm and 7.0 cm for the original dataset, depending on the tide gauge group considered. Compared to NorKyst-800 over its whole domain, both datasets give an RMSD of 7.2 cm, while in a coastal subarea, the ALES dataset and the original dataset give RMSDs of 6.4 cm and 5.8 cm, respectively. A future updated coastal MSS for Norway could rely on a combination of the two datasets in an optimal interpolation scheme taking the external quality measures obtained in the present study into account. (c) 2024 COSPAR. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).