Review for “15 Years of Satellite, In‑Situ and Reanalyzed Salinity Data: Insights into Sea Surface Salinity Trends, Variability, and Water Cycle Links”

By authors Ana Claudia Parracho, Patricia Zunino, Michela Sammartino, Jacqueline Boutin, Eric Greiner, Bruno Buongiorno Nardelli, Nicolas Kolodziejczyk

This study analyzes 15 years of satellite, in‑situ, and reanalysis sea surface salinity (SSS) data to reassess global salinity trends and their links to the intensification of the global water cycle. By comparing six independent SSS products, the authors identify consistent regional patterns of salinification and freshening, revealing both long‑term signals and shorter‑term variability influenced by ENSO. The work highlights the “salty gets saltier and fresh gets fresher” paradigm, while also pinpointing regions departing from classical patterns. By contrasting satellite‑era trends with longer records, the study supports disambiguating anthropogenic signals from natural variability.

The introduction is clear and well structured, providing a solid and up‑to‑date overview of long‑term surface salinity studies. It also provides an informative summary of the different types of salinity measurements available to the scientific community.

We recognize the authors’ effort in the Data section to provide a summary table describing the products used in this study, including data access and associated documentation. However, it remains difficult for the reader to follow which products are being referenced throughout the text. For example, Table 1 introduces product reference numbers that are not used consistently in the manuscript, while the authors rely on acronyms in the text that do not appear in the table. Product #1 is referred to in several different ways: ‘SMOS SSS (L3G product)’ (line 107), ‘SMOS L3G’ (line 131), ‘L3G’ (Figures1 and 2) and ‘SMOS‑only’ in other sections. We recommend replacing product reference numbers with clear acronyms and using them consistently throughout the manuscript (text body, tables and captions).

In addition, there is also confusion regarding the IAP product: it is introduced in the manuscript and later used for long‑term comparisons, yet it does not appear in the Data section or Table 1. Including it alongside the other datasets, with a brief description, would greatly improve consistency. Adding a column to Table 1 indicating the period of availability for each product would also help readers understand the specificity of this dataset.

To further support readers, we recommend adding a column to Table 1 indicating each product’s dependence on ancillary data (e.g., other SSS sources, in situ measurements, SST, SLA). This would clarify which datasets are truly independent and which share common inputs. The authors suggest that independence is determined primarily by the absence of shared SSS inputs; however, it would be helpful to briefly justify why this criterion does not extend to other variables. For example, why are two products that assimilate the same SST field still considered independent? Providing a short explanation would prevent confusion and enhance the transparency of the dataset intercomparison.

More broadly, improving the Data Section by organizing and adding relevant information, harmonizing acronyms and product names between the table and the text would significantly enhance clarity and traceability.

At present, the Results section is difficult to read not only because of the numerous non‑standard acronyms, but also because it is highly descriptive and progresses in a strictly linear manner without a clear narrative thread. As a result, readers may struggle to discern the main scientific messages from secondary details. Reorganizing the section around a clearer storyline—highlighting the key results first and then supporting them with detailed descriptions—would greatly improve accessibility and impact.

It is also important to enhance the readability of the Results section so that readers do not need to continually refer back to maps and tables to follow the text. We recommend avoiding acronyms that are not widely used in the scientific community (e.g., SASG, WAUS) and instead using explicit region names such as “South Atlantic” or “Southeast Indian Ocean.” To further support clarity and reproducibility, the authors should add a table listing the exact coordinates of the boxes shown in Figure 3a.

We would expect the authors to take the analysis one step further. For example, they describe the L3G and Glorys datasets as “noisier” (line 192) but do not explain the origin of this noise—whether it is related to instrument resolution, sensitivity to sea‑state, data processing differences, or other error sources. Similarly, the manuscript notes inconsistencies in 2024, 2023, and 2011 (lines 192–195) without offering hypotheses that might account for these deviations. Presenting such observations without attempting to interpret or contextualize them limits the readers’ ability to understand the differences between products. Providing even brief explanations or possible causes would significantly strengthen the results analysis and improve the interpretability of the comparisons.

In conclusion, this manuscript addresses an important topic and offers valuable insights into recent SSS trends by leveraging multiple datasets across the satellite era. The study is scientifically sound and has strong potential, but significant improvements in clarity, structure, and dataset documentation are needed to strengthen its readability and impact. More importantly, providing additional interpretation of dataset discrepancies and results would greatly enhance the manuscript. With these revisions, the work would offer a clearer and more compelling contribution to understanding global salinity changes and their drivers from satellite observations.

General comments

This paper investigate SSS trends in different kinds of up-to-date products based on either in situ data, satellite data, or both, combined with a model or not. While not groundbreaking as it builds on other studies of SSS trends and do not deeply investigate processes at work, it provides a very interesting comparison of short-term trends for which satellite data allow unprecedented monitoring, and long-term SSS trends that only historical in situ observations can capture. Moreover it gives a good overview of the different kind of products available through Copernicus, and therefore fits perfectly within the State of the Planet report.

The multi-product analysis is done rigorously, the writing and figures are of very good quality. My main recommandations are to always present the different products in the same order in the figures and tables throughout the paper to improve the reading flow, and to check that all references are in the end list as I found that some of them missing.

I recommend to accept the paper, while I provide below a number of minor comments (mostly typos, rephrasing) that should help to polish it before publication.

Minor comments

L39 by saltier regions getting saltier L54 is probably of anthropogenic origin rather than due to natural variability

L63 least-damped

L76 there is now at any time around 4000 Argo floats in the global ocean, so the cumulative number of floats deployed since the early 2000s is certainly higher. Please rephrase

L100 remain

L104 in remote sensing of SSS

Table 1 Information on the spatial resolution of these products is lacking, is it similar from one product to the other and have they been interpolated on a similar gird ? Also tit would be good to give the available period for these products, even if only the period common to all is later used in the analysis. I suggest to give the short name of each product in a separate column for more visibility, and used « SMOS L3G » rather than « L3G » (both are used, depending on the figures, better to be consistent). Also it seems some QUID and PUM references are missing in the reference list (e.g. Tarot, 2024 ; Verbrugge, 2023), please verify and complete.

Table 1, Product 2 : Kolodziejczyk

L115 Theil, 1950; Sen, 1968 : missing in the references list

L131 « SMOS L3G (product 1) and Glorys (product 5) » It would be more intuitive to present the maps in Figure 1 in the same order as the products are presented in Table 1. Also I suggest to present in that order : in situ-only product, satellite-only product, hybrid products.

Figure 1 « SSS trend (pss/decade) 2011-2024  » is common in all subplot titles while the name of the product is only included as a subscript. I suggest to rather put forward the name of the products in these titles.

L135 Which areas of higher variability that are distinct from river plumes do you refer to ? Red areas in Figure 2d are all associated with river plumes, it seems. Could you clarify ?

L154 SSS trends obtained from the 6 products.

L167 When maps of the mean salinity trends estimated from the 6 products (Figure 2a) and SSS climatology (Figure 2c) are compared

L172 are found in the Indian Ocean east of India (where a fresh region shows a freshening trend)

L183 The link between the Pacific ENSO/PDO modes and the Amazon flow is not so direct, can you add a reference ? Which different periods and different techniques to estimate trends are you referring to ? Please clarify. Figure 3b The order of products used in the legend is again different from the order in Table 1 or Figure 1. Please keep consistent.

Table 2 I do not see the underlining of non-significant trends mentioned in L200. Again products seem to be presented in random order.

L233 « mooring etc » : it would be worth knowing if this product includes underway SSS measurements from thermosalinographs. Also details on this product should rather be moved to the data section.

L241 There is an inconsistency between the periods given in the legend and in the subplot titles, please correct.

L246 Figure 5d).

L253 « trends in salinity are not linear » : it is right but by definition you here discuss differences in linear trends, and it is difficult to see trends otherwise on the short period Figure 3b. Thus I would recommand to rephrase with something like « In addition, when looking at such a short

period as 15 years, interannual and interdecadal variability have a strong influence on the estimation of a linear trend (as seen in time series of Figure 3b) ».

L256 Vinogradova and Ponte (2017) is missing in the references list

L258 compared to

L263 saltening ?

L270 up-to-date products derived from satellite data on one hand, and from in situ data on the other hand

L274 Glorys (a model reanalysis which assimilates only in situ salinity). L282 showed that the intensity

L283 salinity sensors