Numerical ocean models are essential tools for scientific research, for operational monitoring and forecasting applications, and as a component of the Earth system models used for climate projections. Their development takes place over a long period of time and mobilises teams of high-level experts combining varied and complementary skills (oceanography, physics, chemistry, biology, applied mathematics, computer science, high-performance computing). The definition of the areas of application for these models and their development priorities are based on well-identified scientific communities, which take a close interest in the good (and not so good) results of these models. There are several communities of this type in France today, including those responsible for developing the CROCO and NEMO numerical codes. The aim of this note is to explain how the work carried out by these two communities is linked and how it complements each other. A complementary strategy and approachThe CROCO and NEMO digital codes are based on the work of two communities that have developed coherent and complementary visions over the last decade. The CROCO and NEMO codes target different applications, in terms of components, scales and processes represented: -The NEMO code is made up of three main components covering ocean dynamics (OCE), sea ice (SI3) and marine biogeochemistry (PISCES). The targeted scales range from global to kilometre scale (effective resolution). NEMO is used to understand ocean variability at these scales above the kilometre, to prepare ocean observations (particularly from space), for climate projections and as a component of Copernicus operational systems. -The CROCO code consists of a dynamic component, interfaced with complementary, multidisciplinary modules (PISCES, MUSTANG, BIOéBUS, ECO3M, etc.). The scales targeted range from the ocean basin to sub-kilometre scales, including non-hydrostatic regimes and LES modelling at decametre scales or laboratory experiments (i.e. close to the DNS). CROCO is used to understand ocean variability at these scales, to prepare ocean observation and as a component of operational demonstrators or future operational systems. As we can see, the perimeters of the applications and the scales targeted by each of the codes are different, so the approaches of these two communities are de facto complementary. Complementary, not only because the two approaches meet different needs, but also because they complement each other: many CROCO applications use boundary conditions derived from NEMO simulations on a larger scale; in return, CROCO is a tool of choice for carrying out process studies with a view to formulating sub-mesh parameterisations for NEMO (internal waves, convection, etc.). In addition to the de facto complementarities, linked to the differences in the respective scientific and application positions of the two codes, there are also close collaborations between the two communities. In fact, as there is a partial overlap between the application areas of the two codes at regional scales, the two communities share many concerns, particularly on issues relating to numerical schemes for resolving kilometre scales. Numerous scientific and technical exchanges currently exist on these issues. Finally, the two communities share a common vision of what could eventually be a coherent interface between the two models. This should make it possible to build applications that exchange relevant information at the right scales and draw on the strengths of each of the models to correctly represent a wide range of scales and processes. The refinement of this vision into an effective demonstrator is the subject of the last proposal in this document.Development actions in progress The articulation of the approaches of the two communities is already reflected in the sharing or joint development of a number of modules and components of the NEMO and CROCO codes. Some components are already shared between the two codes. These include -the interface to the XIOS input/output management module, developed by IPSL,-the PISCES marine biogeochemistry model, the development of which is coordinated by the PISCO group (which includes, among others, NEMO and CROCO developers),-the surface module (air-sea interface management),-the interface for coupling to atmospheric models, implemented via the CERFACS OASIS coupler. -Several developments and research activities have also been carried out in recent years with a view to incorporating them into each of the codes. These include the development of: -the ABL atmospheric boundary layer module (carried out as part of the CMEMS ALBATROS and LEFE SIMBAD projects and the H2020 IMMERSE project),-parametrisation of vertical mixing in Generic Length Scale surface layers,-numerical time and space discretisation schemes that best combine the imperatives of accuracy, stability, conservation, compactness and computer cost.Finally, a series of actions to be undertaken jointly has been established for the medium term. These include -exchanges about the HPC strategies of the two codes (in terms of parallelization and GPU porting)-the development of boundary condition processing approaches using penalty methods; -exchanges about the respective strategies in terms of data assimilation, with a view to pooling the approach (e.g. via the use of NEMOVAR)Note that several of the joint actions listed above have involved the development or integration of interfaces to third-party codes, which are necessary for certain applications. This shared approach to interfaces is consistent with the needs and skills of each of the communities. It is naturally intended to continue and be strengthened for numerous interfaces between the two ocean models: marine biogeochemistry (PISCES, etc.), atmospheric boundary layer (ABL), phase-averaged wave models (WW3, etc.), atmospheric models (WRF, etc.). The Inria AIRSEA team plays a major role in these ongoing developments: this Grenoble-based research team works with each of the CROCO and NEMO communities on a large number of topics that are essential to the development of ocean models. With its expertise in numerical methods, coupling algorithms, multigrid approaches (AGRIF) and high performance computing (HPC), this team is a natural interface between the CROCO and NEMO communities.Towards greater visibility of our approaches As the previous paragraph shows, many approaches and interactions are already underway to avoid unnecessary duplication and to make the most of the skills available in the CROCO and NEMO development teams. The aim of this note is to make these actions more visible. To do this, we propose -The setting up of a "red telephone" type exchange group which will bring together, as often as necessary, pilots and experts from the two development groups as well as "resource" people from the community. Encourage the cross-participation of developers from each of the codes in the development processes within the Working Groups: those that already exist on the NEMO side (including, by way of non-exhaustive examples, the Kernel, Air-sea interactions, HPC, TOP, Verification & Validation Working Groups) and those in the process of being set up, financed by the CROCO GdR. The definition and implementation of a demonstrator for the joint use of NEMO and CROCO on an application relevant to the development of the two models and their interfaces.ConclusionsThe CROCO and NEMO codes each have a development strategy in line with their objectives and their community. The aim of this note is to make the strategy and the co-construction processes more visible and more explicit: they explain the common strategy and enhance the value of all the efforts made. The proposals set out above should make this co-construction process easier to understand for the developers and beyond, for the scientific communities and for the supervisory authorities.