the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 08 May 2026
Intercomparison of Three SWOT-Derived Level-4 Products: From Mapping Accuracy to Multi-Scale Dynamical Representation
Abstract. Oceanic submesoscale dynamics play a critical role in energy cascades and vertical tracer transport. The Surface Water and Ocean Topography (SWOT) mission, through its high-resolution wide-swath sea surface height (SSH) observations, provides an unprecedented capability for resolving these processes. While this enhanced spatial coverage represents a major advance over conventional nadir altimetry, it also introduces new challenges for constructing dynamically consistent gridded Level-4 products. To address these challenges, a range of data fusion and reconstruction approaches have been developed to incorporate SWOT observations into next-generation SSH mapping systems. This study presents a comparative evaluation of three SWOT-derived Level-4 products (MIOST, 4DvarQG, and 4DvarNet) over the North Atlantic (25° N–50° N, 80° W–10° W). The assessment combines Eulerian metrics of mapping accuracy, Lagrangian trajectory mprediction skill based on surface drifter observations, and diagnostics of dynamical structures using Rossby number (Ro) and finite-size Lyapunov exponent (FSLE) fields, with SWOT Level-3 data as a reference. The results reveal a pronounced scale dependence in product performance. In mesoscale-dominated regimes such as the Gulf Stream, 4DvarQG achieves the highest velocity reconstruction accuracy and improves short-term (0–4 days) Lagrangian prediction skill, reflecting the benefits of quasi-geostrophic dynamical constraints. In contrast, 4DvarNet shows greater sensitivity to smaller-scale variability, characterized by sharper SSH gradients, elevated Ro, and more filamentary strain structures, indicating an enhanced representation of fine-scale features. However, the physical realism of these intensified small-scale signals requires further validation against higher-resolution or less-filtered observations. MIOST demonstrates stable and consistent performance across a wide range of spatial scales for global ocean mapping. These results highlight inherent trade-offs between dynamical consistency and small-scale variability representation among current SWOT-based Level-4 products. Future developments may therefore benefit from hybrid approaches that integrate data-driven flexibility with explicit physical constraints.
- Preprint
(2191 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-2236', Anonymous Referee #1, 30 Jun 2026
-
RC2: 'Comment on egusphere-2026-2236', Anonymous Referee #2, 14 Jul 2026
Review of Intercomparison of Three SWOT-Derived Level-4 Products: From Mapping Accuracy to Multi-Scale Dynamical Representation
The authors provide a useful inter-comparsion of three publically available SWOT-derived L4 SSH products: MIOST, 4DvarQG, and 4DvarNET. The authors validate the three products against SWOT L3 SSH data, surface drifter data, and the DUACS 1/8^{o} L4 product using two Eulerian error metrics, RMSE and R, and two Lagrangian metrics, a pixel-wise Rossby number (Ro) and finite-size Lyapunov exponent (FSLE) fields. The authors also include direct comparisons of the L4 products via the relative gain/loss ratios of the derived rossby numbers between each of them.The authors appropriately restrict their evaluation region to the North Atlantic and provide three direct comparisons between the products and SWOT L3 measurements. The authors conclude with a discussion of each product’s strengths and weaknesses and gives recommendations for the conditions when a specific product may be advantageous.
I think that this manuscript is quite good and is close to publication-worthy. The results provided span the scope of the study and then some, and I found the writing and presentation quite clean and thorough. I appreciate the interesting discussion of the results sprinkled throughout the manuscript. Most of my requests boil down to showing extended results, since I think the sweep done here is very useful operationally. I do think it could be useful to include even a limited discussion of seasonality: this could entail recreating Figures 3/4 for a couple of different seasons and adding the plots to a supplement or, if seasonal changes in the statistics are negligible, saying that somewhere in the text. I also think it could be worth it to tweak the raw SWOT L3 evaluation: are the gradients that the authors use for Ro and FSLE calculated directly on the SWOT L3 SSHA and hence contaminated by waves? It may be useful to apply a filter here to account for that in the SWOT - L4 product comparisons. One could fall back to the spline-filtered product in the v3.0 release of the L3 product, rather than go down a filtering rabbit hole.
minor comments:
- (l. 15) typo: mprediction -> prediction. I did not find any other obvious typos in the main text but maybe good to double check.
- (l. 49-59): there are a couple of other studies the authors can cite wrt the ongoing efforts to understand the SWOT noise / effective resolution and extract velocities from the SWOT data. These include: Archer et al., 2025 (https://www.nature.com/articles/s41586-025-08722-8), Carli et. al., 2025 (https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2024JC021216), Wang et al., 2025 (https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2025GL114936), and Tranchant et al., 2025 (https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2025EA004248). These more targeted efforts might be a useful contrast to the L4 products.
- (l. 95-109): Can the author provide some context about how the various schemes treat small-scale SSH variability? Are unbalanced motions implicitly filtered out of MIOST and 4DvarQG? A short discussion here could help provide context for the later results.
- (l. 115-120): Small quibble with the SWOT L3 version - could the author specify whether they are using the “Smoothed” or “Unsmoothed” version of the v2.0.1 SWOT SSH data?
- (l. 115-120): Release v3.0 of the L3 data includes SSH gradients calculated on spline-filtered fields (aka the filtering method used in Tranchant [2025], see https://www.aviso.altimetry.fr/fileadmin/documents/data/tools/handbook_duacs_SWOT_L3.pdf). Presumably the authors used v2.0.1 for consistency with the various L4 products, but I’m curious if the using the v3.0 filtered fields improves the agreement in Section 4. Could be worth it to take a look at this (for example Figures 9-12) and make a note of any difference or lack thereof in the text.
- (l. 125-128): Have the authors experimented with correcting for the Ekman component as in Tranchant (2025)? See https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2025EA004248.
- Figures 3-4: (very minor suggestion) it may be useful to include the region boundaries on one or more of the subplots in each figure, if there is a way to do that cleanly..
- Section 3.1: it would be helpful to see a plot or plots of the performance changes binned across the different regions that the author is describing verbally in the text. Scatter plots or histograms showing the spread in RMSE or gain/loss ratios over all of the pixels in each region would be interesting. It may be ok to exclude Region 1 here, or push these plots to an appendix or supplement.
- (l. 252-289, Figure 5, Figure 6): This is a really interesting section, I’m grateful for the authors contribution here. I wonder if it would be possible to include some discussion of the spread in prediction skill for the various methods across regions and/or seasons (maybe push to a supplement). It would be interesting to see the standard deviation in Gain/Loss ratios for each method overlaid on Figure 6 as shaded envelopes, for example.
- (l. 324-329): I think the author should expand on this discussion of the role of the potential role of the training dataset in 4DvarNET’s accuracy either here or in the conclusions. If I recall correctly NATL60 does not include large-scale tidal signals, and seems to underrepresent SSHA signal from fast-moving intertia-gravity waves. Could this deficiency play a role in it’s over-representation of Ro? By eye it seems like the highest gain regions for 4DvarNET are in the more wave-dominated region outside of the Gulf Stream.
Citation: https://doi.org/10.5194/egusphere-2026-2236-RC2 -
EC1: 'Comment on egusphere-2026-2236', Karen J. Heywood, 20 Jul 2026
Thank you for submitting your paper to Ocean Science. I am grateful to both reviewers for their careful and insightful comments. Both reviewers identify areas where the paper can be strengthened. Please consider these suggestions carefully in preparing your responses and in revising the paper.
The next stage is your online responses in this discussion forum. It is acceptable to post what you will do during revisions, even if you have not yet done that. You have about a month for posting those. After posting these responses, you have about another month for submission of the revised text and the final responses. These final responses can be the same as you posted online, or may need to be updated. If you require additional time at any stage, please do not hesitate to ask.
Karen J. Heywood, co-editor-in-chief
Citation: https://doi.org/10.5194/egusphere-2026-2236-EC1
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 197 | 44 | 14 | 255 | 28 | 23 |
- HTML: 197
- PDF: 44
- XML: 14
- Total: 255
- BibTeX: 28
- EndNote: 23
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This manuscript intercompares three different level-4 SWOT-derived SSH products within their common domain in the North Atlantic, a region which includes the western-boundary current and lower kinetic energy sub-regions. The "level-4" products are those which blend information from multiple satellite altimeters, forming a sea level estimate on a gap-free spatially uniform grid, which typically requires some form of interpolation to estimate SSH in the gaps between observations, and it also involves smoothing the observations to reduce the impact of measurement noise. Level-4 products are widely used in applications for identifying eddies or other sea level features, and for estimating geostrophic currents. Therefore, I think the topic of this manuscript will be of interest to many readers.
The authors use a nice methodology to intercompare the products. They examine Eulerian currents (velocity at fixed locations) as well as Lagrangian flow structures (following particle trajectories) through comparison with drifter observations. They also examined dynamically-relevant statistics of the products, such as the Rossby number, and qualitative aspects of the flow field around a small set of eddies.
Overall, I found the paper does a nice job of illustrating the different aspects of the level-4 products. The presentation style and quality of writing is excellent. I think the article could be published as is, but in a few places I disagree with their characterization of results from the 4DvarNET product. While a lot of good work has gone in to producing the 4DVarNET product, I think the results here demonstrate rather conclusively that the approach based on universal approximators with neural nets fails to produce meaningfully useful results. The fact that such approaches can produce qualitatively realistic results---which are quantitatively less accurate than much simpler linear estimators---should be held up as a warning or a cautionary note, rather than being lauded as "particularly promising."
Detailed comments:
l41-l42: Provide references for "recent studies" or rephrase.
Aha: The next 3 sentences provide these references.
l68-l71: Good summary of introduction.
l100-l109: Aha! Two of the products are not available on a global grid!
Well stated in l110-l112.
l119: Not sure what an "event-based perspective" could refer to.
l214-l229: Nice summary of Eulerian velocity comparisons.
l273: States that DUACS was used as a benchmark for normalizing
the Lagrangian comparisons. Good idea. Fig 6 is nice.
l406: The word "captures" implies that these strain structures are
realistic. I think a better word is "contains" since the realism of the
structures is unknown.
l421: 4DvarNET "better preserves fine-scale spatial organization"
is totally speculative. A more justified or balanced statement
could be, "4DvarNET generates spurious small-scale flow features which look
realistic, but which are not supported by any validation data. Hence, dynamical
inferences from this product should be avoided."
l438: How do you know that 4DvarNET exhibits "enhanced sensitivity to
fine-scale variability"? Could this not simply be due to spurious
structures in the training data or mis-tuning of the assumed noise level?
l495-l496: omit "as it better preserves ..."
l502: Why does inclusion of the neural networks "appear particularly
promising"? Your results appear to show the opposite is true.
Please proofread the references list more carefully. I note some problems
in Le Guillou 2025 with a quick spot-check, for example.