A simulation with no obvious conclusion is being conducted on servers in a data center in Brussels. The simulation is the ocean, or at least as near to a useful approximation of it as any computing system has yet tried. Satellite altimetry data from Copernicus and bathymetric surveys gathered over decades by research vessels from twelve European nations are being combined with temperature gradients collected by sensors on the North Sea seabed. The end product is a living, updateable model of the marine environment that can be queried, altered, and run forward in time to find out what happens next. Researchers are referring to this as a “digital twin.”
The foundation of Europe’s Digital Twin of the Ocean is the data architecture of EMODnet, the European Marine Observation and Data Network, which has been working for years to standardize, harmonize, and make publicly available the marine datasets created by more than a hundred organizations in more than thirty countries. It was not a model that EMODnet put together. Seafloor geology, species distribution, water chemistry, sediment composition, and decades of physical oceanographic measurements were all included in this library, which was one of the biggest collections of marine observation data ever assembled. On top of that library, the digital twin creates a dynamic simulation.
Compared to the technological infrastructure, the practical applications are simpler to comprehend. Instead of waiting for stock assessments for the following season, a fisheries regulator can run queries against the model to determine how a warming Bay of Biscay is changing mackerel migration patterns. Before any concrete is poured, a coastal engineer designing a new breakwater in the Dutch Wadden Sea may test how various designs interact with anticipated sea level rise scenarios. Current and wave forecasts produced by the same data architecture are accessible to a shipping firm that is optimizing routes via the Baltic during winter. The same underlying simulation, which is constantly updated from sensor and satellite feeds, is used by each of these users.
It is actually challenging to communicate the scope of data integration needed to make this work without resorting to figures that rapidly become meaningless. The seafloor is covered by EMODnet’s bathymetric layer alone at resolutions that, if attempted by a single institution, would have required decades of devoted survey labor.
Real-time satellite readings of sea surface temperature, chlorophyll content, anomalies in sea level, and ocean currents are added by the Copernicus Marine Service with worldwide coverage. Both the supercomputing power to handle petabytes of incoming data and the data standards work to guaranty that a temperature reading from a Norwegian Argo float can be meaningfully combined with one from a Spanish research buoy are necessary to merge these into a cohesive, navigable model that stays current rather than becoming a historical snapshot.
The most significant results of the digital twin are anticipated in climate modeling. The political framework that has given this project substantial funding is the European Commission’s Ocean Mission, which is specifically focused on restoring the ocean and producing the scientific data required to guide marine policy. Policymakers have more useful information than trend extrapolation when they have access to a model that can predict how the Mediterranean will react to ongoing atmospheric warming or how changes in Arctic Ocean circulation will affect North Atlantic storm patterns over the next thirty years. It doesn’t simplify climate regulation politics. However, it makes it harder to argue that the scientific foundation is inadequate.
One of the choices that makes the EMODnet method apart from other approaches to managing similar datasets is that open access was incorporated into the project from the start. In the past, publicly funded marine data was kept in institutional archives and available to anyone who understood what to ask and could negotiate the complexities of data sharing agreements. Scientists, journalists, planners, and eventually any interested person can access and download the contents of the digital twin because it is intended to be publicly queryable. The topic of whether this transparency continues as the model gains greater commercial value has not yet been thoroughly investigated.
