Articles | Volume 20, issue 1
https://doi.org/10.5194/os-20-201-2024
© Author(s) 2024. This work is distributed under
the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Special issue:
https://doi.org/10.5194/os-20-201-2024
© Author(s) 2024. This work is distributed under
the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Research article
| 
20 Feb 2024
Research article |
| 20 Feb 2024
| 20 Feb 2024
Unsupervised classification of the northwestern European seas based on satellite altimetry data
Department of Earth Sciences, University of Gothenburg, Gothenburg, Sweden
now at: National Centre for Climate Research, Danish Meteorological Institute, Copenhagen, Denmark
Dani Jones
British Antarctic Survey, NERC, UKRI, Cambridge, UK
Simon D. A. Thomas
British Antarctic Survey, NERC, UKRI, Cambridge, UK
Department of Applied Mathematics and Theoretical Physics, University of Cambridge, Cambridge, UK
Céline Heuzé
Department of Earth Sciences, University of Gothenburg, Gothenburg, Sweden
Related authors
Céline Heuzé, Linn Carlstedt, Lea Poropat, and Heather Reese
EGUsphere, https://doi.org/10.5194/egusphere-2025-700, https://doi.org/10.5194/egusphere-2025-700, 2025
Short summary
Short summary
Extreme sea level kills and will worsen under climate change. In Northern Europe what drives these extreme events will not change so determining these drivers is of use for planning coastal defenses. Here, using two machine learning methods on hourly tide gauge and weather data at nine locations around the North Sea – Baltic, we determine that the driver of prolonged periods of high sea level is the westerly winds, whereas the drivers of the most extreme peaks depend on the coastline geometry.
Flor Vermassen, Clare Bird, Tirza M. Weitkamp, Kate F. Darling, Hanna Farnelid, Céline Heuzé, Allison Y. Hsiang, Salar Karam, Christian Stranne, Marcus Sundbom, and Helen K. Coxall
Biogeosciences, 22, 2261–2286, https://doi.org/10.5194/bg-22-2261-2025, https://doi.org/10.5194/bg-22-2261-2025, 2025
Short summary
Short summary
We provide the first systematic survey of planktonic foraminifera in the high Arctic Ocean. Our results describe the abundance and species composition under summer sea ice. They indicate that the polar specialist N. pachyderma is the only species present, with subpolar species absent. The data set will be a valuable reference for continued monitoring of the state of planktonic foraminifera communities as they respond to the ongoing sea-ice decline and the “Atlantification” of the Arctic Ocean.
Céline Heuzé, Linn Carlstedt, Lea Poropat, and Heather Reese
EGUsphere, https://doi.org/10.5194/egusphere-2025-700, https://doi.org/10.5194/egusphere-2025-700, 2025
Short summary
Short summary
Extreme sea level kills and will worsen under climate change. In Northern Europe what drives these extreme events will not change so determining these drivers is of use for planning coastal defenses. Here, using two machine learning methods on hourly tide gauge and weather data at nine locations around the North Sea – Baltic, we determine that the driver of prolonged periods of high sea level is the westerly winds, whereas the drivers of the most extreme peaks depend on the coastline geometry.
Salar Karam, Céline Heuzé, Mario Hoppmann, and Laura de Steur
Ocean Sci., 20, 917–930, https://doi.org/10.5194/os-20-917-2024, https://doi.org/10.5194/os-20-917-2024, 2024
Short summary
Short summary
A long-term mooring array in the Fram Strait allows for an evaluation of decadal trends in temperature in this major oceanic gateway into the Arctic. Since the 1980s, the deep waters of the Greenland Sea and the Eurasian Basin of the Arctic have warmed rapidly at a rate of 0.11°C and 0.05°C per decade, respectively, at a depth of 2500 m. We show that the temperatures of the two basins converged around 2017 and that the deep waters of the Greenland Sea are now a heat source for the Arctic Ocean.
Céline Heuzé, Oliver Huhn, Maren Walter, Natalia Sukhikh, Salar Karam, Wiebke Körtke, Myriel Vredenborg, Klaus Bulsiewicz, Jürgen Sültenfuß, Ying-Chih Fang, Christian Mertens, Benjamin Rabe, Sandra Tippenhauer, Jacob Allerholt, Hailun He, David Kuhlmey, Ivan Kuznetsov, and Maria Mallet
Earth Syst. Sci. Data, 15, 5517–5534, https://doi.org/10.5194/essd-15-5517-2023, https://doi.org/10.5194/essd-15-5517-2023, 2023
Short summary
Short summary
Gases dissolved in the ocean water not used by the ecosystem (or "passive tracers") are invaluable to track water over long distances and investigate the processes that modify its properties. Unfortunately, especially so in the ice-covered Arctic Ocean, such gas measurements are sparse. We here present a data set of several passive tracers (anthropogenic gases, noble gases and their isotopes) collected over the full ocean depth, weekly, during the 1-year drift in the Arctic during MOSAiC.
Dani C. Jones, Maike Sonnewald, Shenjie Zhou, Ute Hausmann, Andrew J. S. Meijers, Isabella Rosso, Lars Boehme, Michael P. Meredith, and Alberto C. Naveira Garabato
Ocean Sci., 19, 857–885, https://doi.org/10.5194/os-19-857-2023, https://doi.org/10.5194/os-19-857-2023, 2023
Short summary
Short summary
Machine learning is transforming oceanography. For example, unsupervised classification approaches help researchers identify underappreciated structures in ocean data, helping to generate new hypotheses. In this work, we use a type of unsupervised classification to identify structures in the temperature and salinity structure of the Weddell Gyre, which is an important region for global ocean circulation and for climate. We use our method to generate new ideas about mixing in the Weddell Gyre.
Fouzia Fahrin, Daniel C. Jones, Yan Wu, James Keeble, and Alexander T. Archibald
Atmos. Chem. Phys., 23, 3609–3627, https://doi.org/10.5194/acp-23-3609-2023, https://doi.org/10.5194/acp-23-3609-2023, 2023
Short summary
Short summary
We use a machine learning technique called Gaussian mixture modeling (GMM) to classify vertical ozone profiles into groups based on how the ozone concentration changes with pressure. Even though the GMM algorithm was not provided with spatial information, the classes are geographically coherent. We also detect signatures of tropical broadening in UKESM1 future climate scenarios. GMM may be useful for understanding ozone structures in modeled and observed datasets.
Rachael N. C. Sanders, Daniel C. Jones, Simon A. Josey, Bablu Sinha, and Gael Forget
Ocean Sci., 18, 953–978, https://doi.org/10.5194/os-18-953-2022, https://doi.org/10.5194/os-18-953-2022, 2022
Short summary
Short summary
In 2015, record low temperatures were observed in the North Atlantic. Using an ocean model, we show that surface heat loss in December 2013 caused 75 % of the initial cooling before this "cold blob" was trapped below the surface. The following summer, the cold blob re-emerged due to a strong temperature difference between the surface ocean and below, driving vertical diffusion of heat. Lower than average surface warming then led to the coldest temperature anomalies in August 2015.
Simon D. A. Thomas, Daniel C. Jones, Anita Faul, Erik Mackie, and Etienne Pauthenet
Ocean Sci., 17, 1545–1562, https://doi.org/10.5194/os-17-1545-2021, https://doi.org/10.5194/os-17-1545-2021, 2021
Short summary
Short summary
We propose a probabilistic method and a new inter-class comparison metric for highlighting fronts in the Southern Ocean. We compare it with an image processing method that provides a more localised view of fronts that effectively highlights sharp jets. These two complementary approaches offer two views of Southern Ocean structure: the probabilistic method highlights boundaries between coherent thermohaline structures across the entire Southern Ocean, whereas edge detection highlights local jets.
Martin Mohrmann, Céline Heuzé, and Sebastiaan Swart
The Cryosphere, 15, 4281–4313, https://doi.org/10.5194/tc-15-4281-2021, https://doi.org/10.5194/tc-15-4281-2021, 2021
Short summary
Short summary
Polynyas are large open-water areas within the sea ice. We developed a method to estimate their area, distribution and frequency for the Southern Ocean in climate models and observations. All models have polynyas along the coast but few do so in the open ocean, in contrast to observations. We examine potential atmospheric and oceanic drivers of open-water polynyas and discuss recently implemented schemes that may have improved some models' polynya representation.
Amy Solomon, Céline Heuzé, Benjamin Rabe, Sheldon Bacon, Laurent Bertino, Patrick Heimbach, Jun Inoue, Doroteaciro Iovino, Ruth Mottram, Xiangdong Zhang, Yevgeny Aksenov, Ronan McAdam, An Nguyen, Roshin P. Raj, and Han Tang
Ocean Sci., 17, 1081–1102, https://doi.org/10.5194/os-17-1081-2021, https://doi.org/10.5194/os-17-1081-2021, 2021
Short summary
Short summary
Freshwater in the Arctic Ocean plays a critical role in the global climate system by impacting ocean circulations, stratification, mixing, and emergent regimes. In this review paper we assess how Arctic Ocean freshwater changed in the 2010s relative to the 2000s. Estimates from observations and reanalyses show a qualitative stabilization in the 2010s due to a compensation between a freshening of the Beaufort Gyre and a reduction in freshwater in the Amerasian and Eurasian basins.
Céline Heuzé, Lu Zhou, Martin Mohrmann, and Adriano Lemos
The Cryosphere, 15, 3401–3421, https://doi.org/10.5194/tc-15-3401-2021, https://doi.org/10.5194/tc-15-3401-2021, 2021
Short summary
Short summary
For navigation or science planning, knowing when sea ice will open in advance is a prerequisite. Yet, to date, routine spaceborne microwave observations of sea ice are unable to do so. We present the first method based on spaceborne infrared that can forecast an opening several days ahead. We develop it specifically for the Weddell Polynya, a large hole in the Antarctic winter ice cover that unexpectedly re-opened for the first time in 40 years in 2016, and determine why the polynya opened.
Rachel Furner, Peter Haynes, Dave Munday, Brooks Paige, Daniel C. Jones, and Emily Shuckburgh
Geosci. Model Dev. Discuss., https://doi.org/10.5194/gmd-2021-132, https://doi.org/10.5194/gmd-2021-132, 2021
Revised manuscript not accepted
Short summary
Short summary
Traditional weather & climate models are built from physics-based equations, while data-driven models are built from patterns found in datasets using Machine Learning or statistics. There is growing interest in using data-driven models for weather & climate prediction, but confidence in their use depends on understanding the patterns they're finding. We look at this with a simple regression model of ocean temperature and see the patterns found by the regression model are similar to the physics.
Céline Heuzé
Ocean Sci., 17, 59–90, https://doi.org/10.5194/os-17-59-2021, https://doi.org/10.5194/os-17-59-2021, 2021
Short summary
Short summary
Dense waters sinking by Antarctica and in the North Atlantic control global ocean currents and carbon storage. We need to know how these change with climate change, and thus we need accurate climate models. Here we show that dense water sinking in the latest models is better than in the previous ones, but there is still too much water sinking. This impacts how well models represent the deep ocean density and the deep currents globally. We also suggest ways to improve the models.
David Ian Duncan, Patrick Eriksson, Simon Pfreundschuh, Christian Klepp, and Daniel C. Jones
Atmos. Chem. Phys., 19, 6969–6984, https://doi.org/10.5194/acp-19-6969-2019, https://doi.org/10.5194/acp-19-6969-2019, 2019
Short summary
Short summary
Raindrop size distributions have not been systematically studied over the oceans but are significant for remotely sensing, assimilating, and modeling rain. Here we investigate raindrop populations with new global in situ data, compare them against satellite estimates, and explore a new technique to classify the shapes of these distributions. The results indicate the inadequacy of a commonly assumed shape in some regions and the sizable impact of shape variability on satellite measurements.
Lovisa Waldrop Bergman and Céline Heuzé
Ocean Sci. Discuss., https://doi.org/10.5194/os-2018-122, https://doi.org/10.5194/os-2018-122, 2018
Preprint withdrawn
Short summary
Short summary
How to force a model where no suitable observation exists? We here determine using MITgcm the relative influence of the choice of wind, initial hydrography, and sea ice cover on the resulting ocean circulation in Nares Strait, northwest Greenland. The input with the largest effect is the density gradient in the upper layer. We argue that it should be prioritised over high resolution wind for cost-effective simulations of the Arctic straits, crucial for modelling the Arctic freshwater export.
Céline Heuzé
Ocean Sci., 13, 609–622, https://doi.org/10.5194/os-13-609-2017, https://doi.org/10.5194/os-13-609-2017, 2017
Short summary
Short summary
Climate models are the best tool available to estimate the ocean’s response to climate change, notably sea level rise. To trust the models, we need to compare them to the real ocean in key areas. Here we do so in the North Atlantic, where deep waters form, and show that inaccurate location, extent and frequency of the formation impact the representation of the global ocean circulation and how much heat enters the Arctic. We also study the causes of the errors in order to improve future models.
Mathew A. Stiller-Reeve, Céline Heuzé, William T. Ball, Rachel H. White, Gabriele Messori, Karin van der Wiel, Iselin Medhaug, Annemarie H. Eckes, Amee O'Callaghan, Mike J. Newland, Sian R. Williams, Matthew Kasoar, Hella Elisa Wittmeier, and Valerie Kumer
Hydrol. Earth Syst. Sci., 20, 2965–2973, https://doi.org/10.5194/hess-20-2965-2016, https://doi.org/10.5194/hess-20-2965-2016, 2016
Short summary
Short summary
Scientific writing must improve and the key to long-term improvement of scientific writing lies with the early-career scientist (ECS). We introduce the ClimateSnack project, which aims to motivate ECSs to start writing groups around the world to improve their skills together. Writing groups offer many benefits but can be a challenge to keep going. Several ClimateSnack writing groups formed, and this paper examines why some of the groups flourished and others dissolved.
C. Heuzé, J. K. Ridley, D. Calvert, D. P. Stevens, and K. J. Heywood
Geosci. Model Dev., 8, 3119–3130, https://doi.org/10.5194/gmd-8-3119-2015, https://doi.org/10.5194/gmd-8-3119-2015, 2015
Short summary
Short summary
Most ocean models, including NEMO, have unrealistic Southern Ocean deep convection. That is, through extensive areas of the Southern Ocean, they exhibit convection from the surface of the ocean to the sea floor. We find this convection to be an issue as it impacts the whole ocean circulation, notably strengthening the Antarctic Circumpolar Current. Using sensitivity experiments, we show that counter-intuitively the vertical mixing needs to be enhanced to reduce this spurious convection.
Related subject area
Approach: Remote Sensing | Properties and processes: Sea level, tides, tsunamis and surges
Benefits of a second tandem flight phase between two successive satellite altimetry missions for assessing instrumental stability
Understanding uncertainties in the satellite altimeter measurement of coastal sea level: insights from a round-robin analysis
Regional sea level budget over 2004–2022
M2 Monthly and annual mode 1 and mode 2 internal tide atlases from altimetry data and MIOST: focus on the Indo-Philippine Archipelago and the region off the Amazon shelf
Statistical analysis of dynamic behavior of continental shelf wave motions in the northern South China Sea
Spatial and temporal variability in mode-1 and mode-2 internal solitary waves from MODIS-Terra sun glint off the Amazon shelf
Michaël Ablain, Noémie Lalau, Benoit Meyssignac, Robin Fraudeau, Anne Barnoud, Gérald Dibarboure, Alejandro Egido, and Craig Donlon
Ocean Sci., 21, 343–358, https://doi.org/10.5194/os-21-343-2025, https://doi.org/10.5194/os-21-343-2025, 2025
Short summary
Short summary
This study proposes a novel cross-validation method to assess the instrumental stability in sea level trends. The method involves implementing a second tandem flight phase between two successive altimeter missions a few years after the first phase. The trend in systematic instrumental differences made during the two tandem phases can be estimated below ± 0.1 mm yr-1 (16–84 % confidence level) on a global scale for time intervals between the tandem phases of 4 years or more.
Florence Birol, François Bignalet-Cazalet, Mathilde Cancet, Jean-Alexis Daguze, Wassim Fkaier, Ergane Fouchet, Fabien Léger, Claire Maraldi, Fernando Niño, Marie-Isabelle Pujol, and Ngan Tran
Ocean Sci., 21, 133–150, https://doi.org/10.5194/os-21-133-2025, https://doi.org/10.5194/os-21-133-2025, 2025
Short summary
Short summary
We take advantage of the availability of several algorithms for most of the terms/corrections used to calculate altimetry sea level data to quantify and analyze the sources of uncertainty associated with the approach to the coast. The results highlight their hierarchy. Tidal corrections and mean sea surface height contribute to coastal sea level data uncertainties. Improving the retracking algorithm is today the main factor to bring accurate altimetry sea level data closer to the shore.
Marie Bouih, Anne Barnoud, Chunxue Yang, Andrea Storto, Alejandro Blazquez, William Llovel, Robin Fraudeau, and Anny Cazenave
EGUsphere, https://doi.org/10.5194/egusphere-2024-3945, https://doi.org/10.5194/egusphere-2024-3945, 2025
Short summary
Short summary
Present-day sea level rise is not uniform regionally. For better understanding of regional sea level variations, a classical approach is to compare the observed sea level trend patterns with those of the sum of the contributions. If the regional sea level budget is not closed, this allows to detect errors in the observing systems. Our study based on this approach shows the the budget is not closed in the North Atlantic Ocean and identifies as main suspect, errors in Argo-based salinity data.
Michel Tchilibou, Simon Barbot, Loren Carrere, Ariane Koch-Larrouy, Gérald Dibarboure, and Clément Ubelmann
EGUsphere, https://doi.org/10.5194/egusphere-2024-3947, https://doi.org/10.5194/egusphere-2024-3947, 2025
Short summary
Short summary
This study presents the annual and monthly MIOST (MIOST24) internal tide atlases for the Indo-Philippine archipelago and the region off the Amazon shelf. Derived from 25 years of altimetry data and an updated wavelength database, the atlases reveal significant monthly variability of internal tides in both regions. The new atlas improves the correction of internal tides in altimetry data and outperforms MIOST 2022 and HRET existing atlases, thus supporting the development of a global atlas.
Junyi Li, Tao He, Quanan Zheng, Ying Xu, and Lingling Xie
Ocean Sci., 19, 1545–1559, https://doi.org/10.5194/os-19-1545-2023, https://doi.org/10.5194/os-19-1545-2023, 2023
Short summary
Short summary
This study aims to analyze the statistical behavior of the continental shelf wave motions, including continental shelf waves (CSWs) and arrested topographic waves (ATWs), in the northern South China Sea. The cross-shelf structure of along-track SLAs indicates that Mode 1 of CSWs is the predominant component trapped in the area shallower than about 200 m. The cross-shelf structures of CSWs and ATWs illustrate that the methods are suitable for observing the dynamic behavior of the CSWs.
Carina Regina de Macedo, Ariane Koch-Larrouy, José Carlos Bastos da Silva, Jorge Manuel Magalhães, Carlos Alessandre Domingos Lentini, Trung Kien Tran, Marcelo Caetano Barreto Rosa, and Vincent Vantrepotte
Ocean Sci., 19, 1357–1374, https://doi.org/10.5194/os-19-1357-2023, https://doi.org/10.5194/os-19-1357-2023, 2023
Short summary
Short summary
We focus on the internal solitary waves (ISWs) off the Amazon shelf, their velocity, and their variability in seasonal and tidal cycles. The analysis is based on a large remote-sensing data set. The region is newly described as a hot spot for ISWs with mode-2 internal tide wavelength. The wave activity is higher during spring tides. The mode-1 waves located in the region influenced by the North Equatorial Counter Current showed a velocity/wavelength 14.3 % higher during the boreal summer/fall.
Cited articles
Ablain, M., Cazenave, A., Larnicol, G., Balmaseda, M., Cipollini, P., Faugère, Y., Fernandes, M. J., Henry, O., Johannessen, J. A., Knudsen, P., Andersen, O., Legeais, J., Meyssignac, B., Picot, N., Roca, M., Rudenko, S., Scharffenberg, M. G., Stammer, D., Timms, G., and Benveniste, J.: Improved sea level record over the satellite altimetry era (1993–2010) from the Climate Change Initiative project, Ocean Sci., 11, 67–82, https://doi.org/10.5194/os-11-67-2015, 2015. a
Barbosa, S., Gouveia, S., and Alonso, A.: Wavelet-based clustering of sea level records, Math. Geosci., 48, 149–162, https://doi.org/10.1007/s11004-015-9623-9, 2016. a
Björnsson, H. and Venegas, S.: A Manual for EOF and SVD Analysises of Climatic Data, CCGCR Rep. 97-1, McGill University, Montréal, Canada, 52 pp., 1997. a
Bulczak, A. I., Bacon, S., Naveira Garabato, A. C., Ridout, A., Sonnewald, M. J. P., and Laxon, S. W.: Seasonal variability of sea surface height in the coastal waters and deep basins of the Nordic Seas, Geophys. Res. Lett., 42, 113–120, https://doi.org/10.1002/2014GL061796, 2015. a
Camargo, C. M. L., Riva, R. E. M., Hermans, T. H. J., Schütt, E. M., Marcos, M., Hernandez-Carrasco, I., and Slangen, A. B. A.: Regionalizing the sea-level budget with machine learning techniques, Ocean Sci., 19, 17–41, https://doi.org/10.5194/os-19-17-2023, 2023. a
Cao, J., Kwong, S., Wang, R., Li, X., Li, K., and Kong, X.: Class-specific soft voting based multiple extreme learning machines ensemble, Neurocomputing, 149, 275–284, https://doi.org/10.1016/j.neucom.2014.02.072, 2015. a
Chafik, L., Nilsson, J., Rossby, T., and Kondetharayil Soman, A.: The Faroe-Shetland Channel Jet: Structure, Variability, and Driving Mechanisms, J. Geophys. Res.-Ocean., 128, e2022JC019083, https://doi.org/10.1029/2022JC019083, 2023. a
Dangendorf, S., Calafat, F. M., Arns, A., Wahl, T., Haigh, I. D., and Jensen, J.: Mean sea level variability in the North Sea: Processes and implications, J. Geophys. Res.-Ocean., 119, 6820–6841, https://doi.org/10.1002/2014JC009901, 2014. a, b, c, d
Dangendorf, S., Frederikse, T., Chafik, L., Klinck, J., Ezer, T., and Hamlington, B.: Data-driven reconstruction reveals large-scale ocean circulation control on coastal sea level, Nat. Clim. Change, 11, 514–520, https://doi.org/10.1038/s41558-021-01046-1, 2021. a
Dempster, A. P., Laird, N. M., and Rubin, D. B.: Maximum Likelihood from Incomplete Data via the EM Algorithm, J. Roy. Stat. Soc. Ser. B, 39, 1–38, 1977. a
Ducet, N., Le Traon, P. Y., and Reverdin, G.: Global high-resolution mapping of ocean circulation from TOPEX/Poseidon and ERS-1 and -2, J. Geophys. Res.-Ocean., 105, 19477–19498, https://doi.org/10.1029/2000JC900063, 2000. a
Fox-Kemper, B., Hewitt, H., Xiao, C., Aðalgeirsdóttir, G., Drijfhout, S., Edwards, T., Golledge, N., Hemer, M., Kopp, R., Krinner, G., Mix, A., Notz, D., Nowicki, S., Nurhati, I., Ruiz, L., Sallée, J.-B., Slangen, A., and Yu, Y.: Ocean, Cryosphere and Sea Level Change, Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, 1211–1362, https://doi.org/10.1017/9781009157896.011, 2021. a
Frederikse, T. and Gerkema, T.: Multi-decadal variability in seasonal mean sea level along the North Sea coast, Ocean Sci., 14, 1491–1501, https://doi.org/10.5194/os-14-1491-2018, 2018. a, b
GEBCO Bathymetric Compilation Group: The GEBCO_2022 Grid – a continuous terrain model of the global oceans and land, NERC EDS British Oceanographic Data Centre NOC [data set], https://doi.org/10.5285/e0f0bb80-ab44-2739-e053-6c86abc0289c, 2022. a, b
Hermans, T. H. J., Le Bars, D., Katsman, C. A., Camargo, C. M. L., Gerkema, T., Calafat, F. M., Tinker, J., and Slangen, A. B. A.: Drivers of Interannual Sea Level Variability on the Northwestern European Shelf, J. Geophys. Res.-Ocean., 125, e2020JC016325, https://doi.org/10.1029/2020JC016325, 2020. a
Iglesias, I., Lorenzo, M. N., Lázaro, C., Fernandes, M. J., and Bastos, L.: Sea level anomaly in the North Atlantic and seas around Europe: Long-term variability and response to North Atlantic teleconnection patterns, Sci. Total Environ., 609, 861–874, https://doi.org/10.1016/j.scitotenv.2017.07.220, 2017. a
Jones, D. C., Holt, H. J., Meijers, A. J. S., and Shuckburgh, E.: Unsupervised Clustering of Southern Ocean Argo Float Temperature Profiles, J. Geophys. Res.-Ocean., 124, 390–402, https://doi.org/10.1029/2018JC014629, 2019. a
Kaiser, B. E., Saenz, J. A., Sonnewald, M., and Livescu, D.: Automated identification of dominant physical processes, Eng. Appl. Artif. Intel., 116, 105496, https://doi.org/10.1016/j.engappai.2022.105496, 2022. a
Kohonen, T.: The self-organizing map, Proc. IEEE, 78, 1464–1480, https://doi.org/10.1109/5.58325, 1990. a
Kohonen, T. and Mäkisara, K.: The self-organizing feature maps, Phys. Scripta, 39, 168, https://doi.org/10.1088/0031-8949/39/1/027, 1989. a
Le Traon, P. Y. and Ogor, F.: ERS-1/2 orbit improvement using TOPEX/POSEIDON: The 2 cm challenge, J. Geophys. Res.-Ocean., 103, 8045–8057, https://doi.org/10.1029/97JC01917, 1998. a
Le Traon, P. Y., Faugère, Y., Hernandez, F., Dorandeu, J., Mertz, F., and Ablain, M.: Can We Merge GEOSAT Follow-On with TOPEX/Poseidon and ERS-2 for an Improved Description of the Ocean Circulation?, J. Atmos. Ocean. Technol., 20, 889–895, https://doi.org/10.1175/1520-0426(2003)020<0889:CWMGFW>2.0.CO;2, 2003. a
Lehmann, A., Krauss, W., and Hinrichsen, H.-H.: Effects of remote and local atmospheric forcing on circulation and upwelling in the Baltic Sea, Tellus A, 54, 299–316, https://doi.org/10.3402/tellusa.v54i3.12138, 2002. a
Liu, Y. and Weisberg, R. H.: Patterns of ocean current variability on the West Florida Shelf using the self-organizing map, J. Geophys. Res.-Ocean., 110, C06003, https://doi.org/10.1029/2004JC002786, 2005. a, b, c
Lloyd, S.: Least square quantization in PCM, IEEE Trans. Inform. Theory, 28, 129–137, https://doi.org/10.1109/TIT.1982.1056489, 1982. a
Mangini, F., Chafik, L., Madonna, E., Li, C., Bertino, L., and Nilsen, J. E. Ø.: The relationship between the eddy-driven jet stream and northern European sea level variability, Tellus A, 73, 1886419, https://doi.org/10.1080/16000870.2021.1886419, 2021. a, b
Maze, G., Mercier, H., Fablet, R., Tandeo, P., Lopez Radcenco, M., Lenca, P., Feucher, C., and Le Goff, C.: Coherent heat patterns revealed by unsupervised classification of Argo temperature profiles in the North Atlantic Ocean, Prog. Oceanogr., 151, 275–292, https://doi.org/10.1016/j.pocean.2016.12.008, 2017. a, b, c, d
Mihanović, H., Cosoli, S., Vilibić, I., Ivanković, D., Dadić, V., and Gaćić, M.: Surface current patterns in the northern Adriatic extracted from high-frequency radar data using self-organizing map analysis, J. Geophys. Res.-Ocean., 116, C08033, https://doi.org/10.1029/2011JC007104, 2011. a
Papadopoulos, A. and Tsimplis, M. N.: Coherent Coastal Sea-Level Variability at Interdecadal and Interannual Scales from Tide Gauges, J. Coast. Res., 2006, 625–639, https://doi.org/10.2112/04-0156.1, 2006. a
Passaro, M., Cipollini, P., and Benveniste, J.: Annual sea level variability of the coastal ocean: The Baltic Sea-North Sea transition zone, J. Geophys. Res.-Ocean., 120, 3061–3078, https://doi.org/10.1002/2014JC010510, 2015. a
Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., Blondel, M., Prettenhofer, P., Weiss, R., Dubourg, V., Vanderplas, J., Passos, A., Cournapeau, D., Brucher, M., Perrot, M., and Duchesnay, E.: Scikit-Learn: Machine Learning in Python, J. Mach. Learn. Res., 12, 2825–2830, 2011. a
Poropat, L. and Thomas, S. D. A.: leapor/GMMensemble: GMM ensemble (v1.0.0), Zenodo [code], https://doi.org/10.5281/zenodo.10356064, 2023. a
Pujol, M.-I. and Mertz, F.: Global Ocean Gridded L 4 Sea Surface Heights And Derived Variables Reprocessed 1993 Ongoing, E.U. Copernicus Marine Service Information (CMEMS), Marine Data Store (MDS) [data set], https://doi.org/10.48670/moi-00148, 2020. a, b, c
Pujol, M.-I., Faugère, Y., Taburet, G., Dupuy, S., Pelloquin, C., Ablain, M., and Picot, N.: DUACS DT2014: the new multi-mission altimeter data set reprocessed over 20 years, Ocean Sci., 12, 1067–1090, https://doi.org/10.5194/os-12-1067-2016, 2016. a, b
Rosso, I., Mazloff, M. R., Talley, L. D., Purkey, S. G., Freeman, N. M., and Maze, G.: Water Mass and Biogeochemical Variability in the Kerguelen Sector of the Southern Ocean: A Machine Learning Approach for a Mixing Hot Spot, J. Geophys. Res.-Ocean., 125, e2019JC015877, https://doi.org/10.1029/2019JC015877, 2020. a
Rousseeuw, P. J.: Silhouettes: A graphical aid to the interpretation and validation of cluster analysis, J. Comput. Appl. Math., 20, 53–65, https://doi.org/10.1016/0377-0427(87)90125-7, 1987. a
Scotto, M. G., Alonso, A. M., and Barbosa, S. M.: Clustering Time Series of Sea Levels: Extreme Value Approach, Journal of Waterway, Port, Coast. Ocean Eng., 136, 215–225, https://doi.org/10.1061/(ASCE)WW.1943-5460.0000045, 2010. a
Sterlini, P., de Vries, H., and Katsman, C.: Sea surface height variability in the North East Atlantic from satellite altimetry, Clim. Dynam., 47, 1285–1302, https://doi.org/10.1007/s00382-015-2901-x, 2016. a
Taburet, G., Sanchez-Roman, A., Ballarotta, M., Pujol, M.-I., Legeais, J.-F., Fournier, F., Faugere, Y., and Dibarboure, G.: DUACS DT2018: 25 years of reprocessed sea level altimetry products, Ocean Sci., 15, 1207–1224, https://doi.org/10.5194/os-15-1207-2019, 2019. a, b, c
Thompson, P. R. and Merrifield, M. A.: A unique asymmetry in the pattern of recent sea level change, Geophys. Res. Lett., 41, 7675–7683, https://doi.org/10.1002/2014GL061263, 2014. a
Winther, N. G. and Johannessen, J. A.: North Sea circulation: Atlantic inflow and its destination, J. Geophys. Res.-Ocean., 111, C12018, https://doi.org/10.1029/2005JC003310, 2006. a
Short summary
In this study we use a machine learning method called a Gaussian mixture model to divide part of the ocean (northwestern European seas and part of the Atlantic Ocean) into regions based on satellite observations of sea level. This helps us study each of these regions separately and learn more about what causes sea level changes there. We find that the ocean is first divided based on bathymetry and then based on other features such as water masses and typical atmospheric conditions.
In this study we use a machine learning method called a Gaussian mixture model to divide part of...
