A European observatory built for the Arctic seasons
Europe took a fully equipped research icebreaker into the central Arctic and made the drifting ice a laboratory. Led by Germany’s Alfred Wegener Institute, the 2019–2020 MOSAiC expedition connected measurements of air, snow, sea ice, seawater and living systems across the seasons. Its lasting achievement is shared evidence: observations gathered in conditions that are difficult to reach, preserved so other researchers can question and build on them. Europeans can be proud of the public institutions, crews and scientific cooperation that made this possible.
MOSAiC means Multidisciplinary drifting Observatory for the Study of Arctic Climate. AWI’s expedition account records the completed mission and its return to Bremerhaven on 12 October 2020. The ship was Polarstern, a German research platform operated by AWI. This was a European-led international undertaking, rather than an EU agency operating its own mission. The partner record spans 20 countries, including European institutions and collaborators in the United States, Russia, China, Japan, Canada and South Korea.
The platform, the partners and the connected system
The distinction strengthens the achievement: European capability created a platform on which the world could work. The project’s team account separates scientific coordination from the crew who kept the ship functioning. Germany supplied leadership and the central platform; researchers and support teams supplied instruments, expertise and logistics across borders. The cruise-report archive also records the Russian Akademik Fedorov’s role in establishing the distributed observing network. Those are historical contributions to this particular expedition, not a claim about present partnerships.
The expedition design joined a central ice camp with observing sites extending roughly 50 kilometres around Polarstern. Instruments drifted with the ice, allowing researchers to study processes together rather than assemble a story from unrelated summer visits. The scientific question was how energy, moisture and material move between atmosphere, snow, ice and ocean. A cloud changes radiation at the surface; snow changes insulation and reflected light; seawater exchanges heat with the ice above it. Observing the connections is the point of the system.
The seasonal cycle matters because a summer photograph cannot reveal how the same environment behaves through darkness, freezing, spring light and melt. MOSAiC’s mission statement identifies the central-Arctic winter and spring observation gap that motivated the project. Its historical warming figures and future-tense language should not be read as a new 2026 climate assessment. The continuing value lies in measured processes that researchers can compare with the mechanisms inside models.
Keep the expedition and data clocks separate
Read the expedition’s clocks carefully. The official figures give 389 days for the whole expedition, more than ten months with Polarstern frozen into the ice, 300 research days on the first floe and another 30 on a second. These measures describe different things. They do not establish one uninterrupted year of every instrument operating on one floe. Cruise reports and individual dataset coverage are the evidence for a particular observation period. That precision makes the remarkable achievement easier to assess.
Fieldwork was followed by a second construction job: turning measurements into documented products. The MOSAiC data guide dates worldwide free availability from 1 January 2023 and distinguishes fully published repository datasets from primary material still requiring processing or analysis. PANGAEA is the main repository, alongside national archives. A dataset’s DOI, processing description, units, uncertainty and licence make the record usable beyond the expedition team. An open file and a ready-to-interpret result are different stages.
The expedition, the product and the later test
| Evidence stage | Dated source | What it establishes |
|---|---|---|
| Completed field mission | September 2019–October 2020; AWI return date 12 October 2020 | An observatory worked through an Arctic seasonal cycle; individual instruments have their own coverage |
| A processed, citable product | 2022 PANGAEA Level 3 radiosonde series | Specified processing, uncertainty, file formats and reuse licence |
| Worldwide free data access | MOSAiC policy: from 1 January 2023 | Access to evidence; processing status still differs between products |
| Continuing scientific use | CAPS v1.1 study published 9 November 2026 | A named model evaluated against observations, with improvements and remaining biases |
Original synthesis of linked primary sources, checked 15 November 2026. Data release and paper publication dates do not redate the expedition.
Read one real dataset before trusting a chart
One concrete example is Maturilli and colleagues’ 2022 Level 3 radiosonde publication. Weather-balloon measurements were processed through the GRUAN reference-network algorithm, with uncertainty values for measured parameters. The record warns that GPS-derived heights near the ship can contain artefacts and recommends pressure-based height calculations in the atmospheric boundary layer. It also says the tab-delimited files contain only a subset of parameters; the complete processed data are in netCDF files. A reader who notices those details can avoid a misleading comparison.
For a student comparing two published temperature profiles, a useful evidence note therefore records the exact dataset and month, instrument, processing level, height coordinate, units and uncertainty. Before joining another product, check its time window and measurement location. The observatory moved, so differences between dates are not automatically changes at a fixed point. This is an editorial reading protocol, not a new analysis of MOSAiC measurements. It shows why Europe’s investment in documentation matters as much as a dramatic image of an icebreaker.
November 2026: the observations keep testing models
The scientific legacy is still active. On 9 November 2026, Yang and colleagues published a CAPS v1.1 model evaluation using MOSAiC atmosphere, ocean, snow and ice observations. The team includes researchers at Sun Yat-sen University in China and AWI in Germany. The study reports improved consistency of heat exchange between model components and remaining errors. It also distinguishes improvements associated with constrained atmospheric conditions from those attributable to the revised coupling itself. This is one documented model test, not proof that all Arctic forecasts are now accurate.
That example explains the route from ship to public knowledge. Observations describe what instruments recorded; a processed product makes those measurements comparable and citable; a model represents processes and is tested against evidence. A result that exposes remaining errors is a useful outcome: it identifies where the next scientific effort belongs. The expedition did not need to answer every climate question at sea to create a durable research capability. Its archive lets later teams ask better questions with an inspectable starting point.
What kind of claim is the reader looking at?
| Material | Question to ask | Evidence to retain |
|---|---|---|
| Instrument observations | What was measured, where and when? | Instrument, position/time coverage, units and quality information |
| Processed data product | Which processing and uncertainty apply? | Exact DOI/version, coordinate choice, format and licence |
| Model evaluation | What experiment improved, against which observations? | Model version, experiment design, comparison metrics and remaining bias |
| Broader climate conclusion | Does this source support that geographic and time scale? | Supporting studies and synthesis; do not infer an Arctic-wide trend from one moving profile |
Original Federal Europa evidence-reading aid. No new calculation, dataset analysis or forecast is claimed.
A European public capability with a global legacy
MOSAiC belongs alongside Euro-Argo’s sustained ocean profiles, EUMETSAT’s weather observations and Copernicus’s operational information services. Their jobs differ: an intensive drifting campaign resolves local processes, repeated observing networks sustain broader records, and services combine observations with models for practical use. No single dataset replaces that whole chain. European strength comes from being able to build and connect its parts.
The next test is stewardship: preserve the records, maintain access, document corrected products and let the evidence constrain confident claims. A democratic Europe benefits when publicly supported knowledge remains open to scrutiny across institutions and generations. MOSAiC shows that Europe can lead a demanding common endeavour and make its results useful far beyond its own borders. That is a substantial European achievement—and a reason to keep investing in the people and institutions behind the Proud Europe story.