TS-Cast: Deep Learning for Subsurface Ocean Reconstruction from Satellite Observations in the Northwestern Pacific

Chae, Jeong-Yeob; Donohue, Kathleen A.; Park, Jae-Hun

doi:10.5194/egusphere-2025-5546

Preprints

https://doi.org/10.5194/egusphere-2025-5546

Preprints

19 Nov 2025

| 19 Nov 2025

TS-Cast: Deep Learning for Subsurface Ocean Reconstruction from Satellite Observations in the Northwestern Pacific

Jeong-Yeob Chae, Kathleen A. Donohue, and Jae-Hun Park

Abstract. Since the 1990s, satellite observations have been providing reliable estimates of ocean surface states, including absolute dynamic topography (ADT), sea surface temperature (SST), and sea surface salinity (SSS) at sufficient space and time scales to characterize ocean dynamics. Together with the extensive hydrographic dataset from Argo and ship-based hydrographic profiles, these measurements provide a comprehensive view of oceanic conditions. While ADT represents integrated information for subsurface water properties, it is challenging to relate SST, SSS, and ADT with subsurface water profiles due to their complex spatial and temporal variations. To address this issue, we introduce a novel deep neural network, the thermohaline profile estimating network termed TS-Cast. Sourcing from monthly climatological profiles, TS-Cast is designed to adjust these profiles to align with satellite-measured SST, SSS, and ADT data, by training with approximately 150,000 Argo and ship-based thermohaline profiles in the northwestern Pacific. TS-Cast’s capability is demonstrated by comparisons with independent time-series data from moorings that measured temperature and salinity or vertical acoustic travel time. The network significantly improves upon the climatological baseline, achieving an overall Root Mean Square Error (RMSE) of < 1 °C for temperature and < 0.1 psu for salinity in the upper 500-m depths at the Kuroshio Extension region. This performance surpasses that of data-assimilated numerical models and is comparable to that of a data-assimilated statistical model, validating TS-Cast as a powerful tool for ocean monitoring. Critically, this framework reveals not only TS-Cast's high fidelity but also demonstrates that the limitations of the input satellite data fundamentally constrain its predictive skill.

Received: 08 Nov 2025 – Discussion started: 19 Nov 2025

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Jeong-Yeob Chae, Kathleen A. Donohue, and Jae-Hun Park

Status: final response (author comments only)

RC1:
'Comment on egusphere-2025-5546', Anonymous Referee #1, 27 Dec 2025
This paper describes a deep neural network to reconstruct subsurface thermohaline profiles from a combination of in-situ measurements and satellite data, focused on the Northwestern Pacific.
The technique by itself presents very relevant and innovative solutions to extract information from time series of satellite-based sub-images (centered on the target profile location) and combine them to correct available climatological data. However, the manuscript presents some misleading statements and some questionable claims that need to be corrected before publication. Additionally, several technical clarifications are also necessary, as detailed in the following.
Major remarks:
In different sections, the authors cite “geostrophy” as a limiting factor in designing 3D reconstructions based on satellite altimetry. This is a conceptual imprecision, as geostrophy is an approximation used to derive horizontal currents from horizontal pressure gradients and has no influence on the sea level observations (which reflect all processes driven by full dynamics). Moreover, while it is commonly stated by the community that the sea surface elevation represents an integral measure of subsurface processes, this is actually true only in purely baroclinic flows and it is not true in general. Surface elevation is useful to infer the vertical distribution of hydrographic properties because the ocean is predominantly baroclinic and surface intensified (so that pressure differences are detectable at the surface). While this is discussed in section 4, it is not correctly introduced in the Abstract. Consequently, the second and third points listed in section 4 are redundant and should be merged removing any reference to non-geostrophy and eventually referring to non-baroclinic (the two terms should not be used interchangeably).

the FiLM layer is not detailed to the level that would allow replicating the study. In fact, the neural network that modulates the scaling and shifting factors (conditioning network) is described nowhere (number of layers/neurons, etc.). It might also be worth adding a dedicated panel in figure 3.

I understood the target of the TS-cast regression model is represented by the observed T/S profiles (2 channels, 128 layers). What is the “Error Variance” 3x128 tensor in figure 3? How is it computed and what are the three channels? Are the predicted/target variables normalized before entering the model? If yes, how (full profile standardization, depth-by-depth, min/max., etc.)? How does this eventually impact the loss functions defined in section 2.4?

Following from previous point: in Section 2.4 the authors write: “TS-Cast network directly estimates temperature (ˆT) and salinity (ˆS) profiles, but also their associated depth-dependent uncertainty, represented by the logarithmic variance (log σ2)”. Are the authors speaking of the true variance or error variance? How is this practically computed? I suggest the terms “uncertainty” and “variance” are used much more carefully throughout the text as they are not synonyms. Again, please provide formulae to explain your operational definitions. In the present form, this section is very confusing.

By the way, if the authors are relying on variance and not error variance, the Loss function used forces the model to learn more from the less dynamically active depth levels, which seems quite a sub-optimal choice, requiring further justification. In fact, this seems in contrast to the objective of the model, which would likely be to “resolve” the most relevant dynamics.

The physics-informed loss function accounting for the equation of state has been first introduced in Sammartino et al., https://doi.org/10.1016/j.envsoft.2025.106660, and that paper should thus be cited in this section.

How did the authors compute coherence?

The authors' critique of what they name 'closed-loop' validation may be overly broad (both in the introduction and in the concluding remarks). Validation against independent data from the same platform (e.g., held-out Argo floats) is a standard and robust practice. While time-series validation at mooring sites provides valuable complementary information on temporal dynamics, it has its own spatial limitations. I suggest reframing this as a complementary strength of the present work rather than a limitation of prior approaches.

Minor points:
Line 1: “estimates of ocean surface states” -->“estimates of ocean surface state”
Line 3: “at sufficient space and time scale” --> “at sufficient space and time resolution”
Line 4: “While ADT represents” --> “While ADT contains”
Line 13 and following: “data-assimilated” --> “data-assimilating”
Line 18: remove “fundamental variables,”
Line 28: remove “primarily”
Line 39: change to “like multilinear regressions combined with optimal interpolation”
Line 44: change to “designed to capture sequential dependencies” (at least the first paper is not using LSTM over time, but over depth)
Line 46: add reference to Sammartino et al. (2025) - see above
Line 61 and following: considering the model domain, what is the advantage of transforming Lat-lon coordinates to 3 different variables?
L262: “explains”-->”might explain”
Line 284: “non-geostrophic”-->”non-baroclinic”
Citation: https://doi.org/10.5194/egusphere-2025-5546-RC1
RC2: 'Comment on egusphere-2025-5546', Anonymous Referee #2, 12 Jan 2026

The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2025-5546/egusphere-2025-5546-RC2-supplement.pdf

Citation: https://doi.org/10.5194/egusphere-2025-5546-RC2
EC1: 'Comment on egusphere-2025-5546', Meric Srokosz, 19 Jan 2026

Based on the reviewers' comments the manuscript requires a major revision.

Citation: https://doi.org/10.5194/egusphere-2025-5546-EC1

Jeong-Yeob Chae, Kathleen A. Donohue, and Jae-Hun Park

Data sets

TS-Cast: Deep Learning for Subsurface Ocean Reconstruction from Satellite Observations in the Northwestern Pacific Jeong-Yeob Chae et al. https://doi.org/10.5281/zenodo.17504047

Jeong-Yeob Chae, Kathleen A. Donohue, and Jae-Hun Park

Viewed

Total article views: 410 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
290	89	31	410	22	19

HTML: 290
PDF: 89
XML: 31
Total: 410
BibTeX: 22
EndNote: 19

Views and downloads (calculated since 19 Nov 2025)

Month	HTML	PDF	XML	Total
Nov 2025	134	17	11	162
Dec 2025	51	27	10	88
Jan 2026	96	41	10	147
Feb 2026	9	4	0	13

Cumulative views and downloads (calculated since 19 Nov 2025)

Month	HTML	PDF	XML	Total
Nov 2025	134	17	11	162
Dec 2025	51	27	10	88
Jan 2026	96	41	10	147
Feb 2026	9	4	0	13

Viewed (geographical distribution)

Total article views: 399 (including HTML, PDF, and XML) Thereof 399 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 07 Feb 2026

Short summary

We introduce TS-Cast, a novel deep neural network that reconstructs subsurface thermohaline structures from satellite observations (ADT, SST, and SSS). Validated against independent time-series data, TS-Cast achieves RMSE < 1 °C and < 0.1 psu in the upper 500 m of the Kuroshio Extension, comparable or surpassing data-assimilated numerical models. Critically, we demonstrate that the physical limitations of the input satellite data fundamentally constrain the model's predictive skill.


Total:	0
HTML:	0
PDF:	0
XML:	0