the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 20 Jun 2024
Marine data assimilation in the UK: the past, the present and the vision for the future
Abstract. In the last two decades UK research institutes have led a wide range of developments in marine data assimilation (MDA), covering areas from the MDA applications in physics and biogeochemistry, to MDA theory. We review the progress over this period and formulate our MDA vision for both the short-term and the longer-term future. We focus on identifying the MDA stakeholder community and current/future areas of impact, as well as the current trends and the future opportunities. This includes rapid growth of machine learning (ML) / artificial intelligence (AI) and digital twin applications. We articulate the MDA needs for future types of observational data (whether planned missions, or hypothetical) and what should be the response of the MDA community to the increase in computational power and new computer architectures (e.g. exascale computing). Although the specifics depend on the MDA area, we advocate for balanced redistribution of the new computational capability among increased model resolution, model complexity, more sophisticated DA algorithms and uncertainty representation (e.g. ensembles). We also advocate for integrated approaches, such as strongly coupled DA (ocean/atmosphere, physics/biogeochemistry, ocean/sea ice) and the use of ML/AI components (e.g. for multivariate increment balancing, bias-correction, model emulation, observation re-gridding, or fusion).
- Preprint
(1262 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-1737', Anonymous Referee #1, 03 Sep 2024
general comments
The article “Marine data assimilation in the UK: the past, the present and the vision for the future” by Skakala et al gives an overview of the progress made over the last two decades in the field of marine data assimilation in the UK, including the work carried out at the UK-based European Centre for Medium-Range Weather Forecasts (ECMWF).
The article provides an interesting overview of past and current efforts and planned developments, including the coordination of activities within the UK and with the international MDA community. These efforts and this vision of the future take into account the revolution in AI acceleration and GPU computing, as well as future satellite missions. Another key factor influencing the future of ocean forecasting and marine data assimilation is the maintenance and extension of the in situ observing system, including the development of smart platforms, which is well defended in the article. The article also explains how hybrid variational and ensemble methods, taking advantage of different types of AI acceleration, will be developed, building on NEMOVAR, JEDI and also PDAF for biogeochemistry. A very important point is made that the community must continue to attract new scientists to the field.
The article is well written and easy to read, there is some repetition between sections, but this is difficult to avoid as there are overlaps between MDA fields.
Some sections contain fewer references than others, which could be improved.
Some paragraphs summarize very complex concepts, although useful references are generally provided. Those paragraphs can be difficult to follow and could benefit from a few extra sentences to better explain the scientific problem.
Mostly coastal or high resolution applications are considered as good examples for digital twins of the ocean, but it could be stressed (in the summary) that OSEs, OSSEs and reanalyses could also be strong assets for such future services.
Specific comments
L 110; L115; L135; L150, L630-660; L670 : references could be added to ensure consistency with the rest of the article.
L340, L880 : EO has not been introduced, and it seems a little odd to mention the Earth Observation community only there in the summary. It could also be explicitely mentioned in the paragraphe on observing systems design L893
What we see on Fig 2 is not really explained.
What is error 242? (see title of the figure), does blue mean an error reduction?
L698: Kd has not been introduced
L244 : It may be easier to follow by adding one sentence, maybe on the time scales of the impact of the atmospheric forcing.
L426-427 : a couple of sentences could explain explain what we see on Fig4.
L455: "the combined impact of physics and BGC DA is also dependent on the assimilation methodology": maybe one or two examples would help to link with future tests and potential improvements.
L 473: as there are slightly more details about this in 3.2.2 the authors may refer to this section
technical corrections
L1040 : inconsistency between Eyre et al (2021) in the text, and (2022) in the references
L 1188 : add the year in reference Nerger et al (2023)
Citation: https://doi.org/10.5194/egusphere-2024-1737-RC1 - AC1: 'Reply on RC1', Jozef Skakala, 27 Sep 2024
-
RC2: 'Comment on egusphere-2024-1737', Anonymous Referee #2, 04 Nov 2024
Review of "Marine data assimilation in the UK: the past, the present and the vision for the future" by Skákala
and co-authors.My background is in ocean data assimilation, ocean modeling and the development of an operational ocean forecasting system so I am familiar with the tools discussed by the authoring team, but maybe not so much familiar with the regional seas around the UK.
The past and present are quite well covered by the manuscript but the most interesting - the future - is yet mysterious in the manuscript.
Even though there is no denying that the UK has contributed - and is still contributing - pioneering efforts in marine data assimilation, I cannot see the purpose of publishing a national perspective on marine data assimilation strategy in an international scientific journal. A possible benefit for readers would have been if the UK proposed a way forward that was applicable to all other nations, as if the UK were a laboratory where an efficient collaboration between universities, the national marine institutes and the weather forecasts centres followed a common strategy and a detailed implementation plan. However I failed to see signs of coordination in the manuscript that are beyond the status quo today.Six UK institutes are actively using marine data assimilation in the UK, according to Table 1. Two institutes are operational weather forecasting centres (UK Metoffice and ECMWF) and four institutes use DA for research. This is an impressive number for a single nation and the question naturally arises of how these institutes will cooperate and build up synergies despite their respective institutional constraints.
A cornerstone of collaboration efforts is the use of common tools. This is not compulsory for scientists who can exchange ideas, but leads to significant economies of scale when the software needs a multidisciplinary team for maintenance, for example the European NEMO ocean modeling engine that has been successfully adopted by most UK institutes. So I would have expected the paper to announce a software development plan that would accommodate all the wishes from all institutes. There is indeed an intention to reduce the number of softwares from four to two, one optimised for operations and one simpler for research, but that information comes very late in the paper (Section 4.2) so it is too late to sort out which ideas will fit in which softwares and how. In other words, there is no research-to-operations plan in the manuscript, and the reader is left to imagine the sequence of steps that will incorporate the numerous ideas listed in the manuscript into a common tool.About the main contents of the paper, the authors have chosen to provide either long lists of bullet points or short sections exposing individual ideas to be followed up. These bullet points and ideas read as shopping lists without logical links: each item apparently provided by one institution or another. One example is section 3.3.2 (coupled DA) that ends on an enumeration of four steps that mix scientific and technical considerations, ending with (iv) that does not seem connected with the previous three. There are also repetitions between sections (for example lines 204 to 209 are repeated from section 2.2). As such, the manuscript makes a tedious read and does not provide a positive example of consolidated cross-institutional collaboration.
Loose ends are yet another weakness of the manuscript. One example among many others can be found on line 360: "More complex balance relationships could use as their starting point the mass conservation scheme of Hemmings et al (2008)". Why this one in particular? "or focus on ML/statistical modelling, which is
being pursued in a current studentship between PML and the University of Reading". This is blue sky to the reader: It does not say why ML/statistical modelling is relevant here, how much can be expected from a studentship or if there are other options on the table than these two.
The wishlists relating to data assimilation methodology can be found in international community papers as the issues raised are not more specific to UK marine research than anywhere else. Therefore, contributions to international community papers such as those issued by the OceanPredict community every four years have a larger impact on the community and these almost always receive inputs from the UK institutes involve here.The explanations are often missing. Here is another example: "Simple post-processing balances will be applied in the next ECMWF systems such that sea ice increments will induce near surface ocean temperature increments, but not the other way around." Why is that? "Similar plans are being
considered at the Met Office." This is another very vague indication, is the reader expected to take action of that?The Figures are taken out of their context from already published material and are exposed without explanations. For example Fig. 4 is a map of correlations between ocean and atmospheric variables, intended to convey the point that coupled assimilation is a good thing. However, without a minimum of explanations, it is not possible to tell whether these values of correlation are realistic or not. The same point can be made about all the other figures. Figure 1 - a map of the institutes in the UK - does not provide any additional information above the list of authors affiliations.
The only interestingly novel statements can be found about the implementation of advanced Fortran softwares NEMO and NEMOVAR on GPUs, which would be very useful ways to reduce the energy consumption of marine data assimilation systems. Unfortunately these strong claims are not substantiated, would it only be by some indications of computational efficiency, so I would take them with a pinch of salt.
The Digital Twin of the Ocean (DTO), an emerging concept in the domain, is first cited out of the blue in line 490, without context. This is another missed occasion to provide a singular reflexion from the perspective of prominent UK institutes.To conclude, the manuscript provides a literature review of profuse marine data assimilation research performed in the UK but does not synthesizes the inventory into a common vision for the future. The manuscript seems like a potluck of research ideas from a broad range of experts, explained in very terse sentences. I could follow the arguments on the topics where I knew the background, but the manuscript is not understandable without a profound knowledge of the state of the art and of the literature cited, which is unsurprisingly UK-biased in this paper. It seems to me that reshaping the paper into a more logical white paper require too much efforts and discussions for revisions and I would prefer if the authors would spend their time and energy into international community papers where the same ideas would reach more readers, as they have usually excelled at in the past.
Citation: https://doi.org/10.5194/egusphere-2024-1737-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 374 | 84 | 73 | 531 | 19 | 18 |
- HTML: 374
- PDF: 84
- XML: 73
- Total: 531
- BibTeX: 19
- EndNote: 18
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1