OS4.4

EDI
Numerical modelling of the ocean: new scientific advances in ocean models to foster exchanges within NEMO community and contribute to future developments

NEMO (Nucleus for European Modelling of the Ocean) is a state-of-the-art modelling framework of the ocean that includes components for the ocean dynamics, the sea-ice and the biogeochemistry, so as a nesting package allowing to set up zooms and a versatile data assimilation interface (see https://www.nemo-ocean.eu/).
NEMO is used by a large community in Europe and world-wide (~200 projects, ~100 publications each year) covering a wide range of applications : oceanographic research, operational oceanography, seasonal forecast and climate projections.
NEMO is in particular used in 6 Earth System Models within CMIP6 and in Copernicus Marine Services (CMEMS) model-based products.

This session will provide a forum to properly address the new scientific advances in numerical modelling of the ocean and their implication for NEMO developments associated with:
• Ocean dynamics at large to coastal scales, up to 1km resolution ;
• Ocean biogeochemistry
• Sea-ice
• New numerical schemes associated to energy conservation constraints
• High performance computing challenges and techniques

The session will cover both research and operationnal activities contributing to new analysis, ideas and developments of ocean numerical models.
Presentations of results based on new NEMO functionalities and new NEMO model configurations are welcome.

Co-organized by CL4/ESSI2, co-sponsored by CMEMS
Convener: Claire Levy | Co-conveners: Mike Bell, Jerome Chanut, Julien Le Sommer, Dorotea Iovino
vPICO presentations
| Mon, 26 Apr, 15:30–17:00 (CEST)

vPICO presentations: Mon, 26 Apr

Chairpersons: Jerome Chanut, Julien Le Sommer, Claire Levy
Sub session "Applications" - Chair: Jérôme Chanut, Co-chair: Dorotea Iovino
15:30–15:32
|
EGU21-8491
Ana Barbosa Aguiar, Jennifer Waters, Martin Price, Gordon Inverarity, Christine Pequignet, Jan Maksymczuk, Kerry Smout-Day, Matthew Martin, Mike Bell, James While, Robert King, Daniel Lea, and John Siddorn

The importance of oceans for atmospheric forecasts as well as climate simulations is being increasingly recognised with the advent of coupled ocean / atmosphere forecast models. Having comparable resolutions in both domains maximises the benefits for a given computational cost. The Met Office has recently upgraded its operational global ocean-only model from an eddy permitting 1/4 degree tripolar grid (ORCA025) to the eddy resolving 1/12 degree ORCA12 configuration while retaining 1/4 degree data assimilation. 

We will present a description of the ocean-only ORCA12 system, FOAM-ORCA12, alongside some initial results. Qualitatively, FOAM-ORCA12 seems to represent better (than FOAM-ORCA025) the details of mesoscale features in SST and surface currents. Overall, traditional statistical results suggest that the new FOAM-ORCA12 system performs similarly or slightly worse than the pre-existing FOAM-ORCA025. However, it is known that comparisons of models running at different resolutions suffer from a double penalty effect, whereby higher-resolution models are penalised more than lower-resolution models for features that are offset in time and space. Neighbourhood verification methods seek to make a fairer comparison using a common spatial scale for both models and it can be seen that, as neighbourhood sizes increase, ORCA12 consistently has lower continuous ranked probability scores (CRPS) than ORCA025. CRPS measures the accuracy of the pseudo-ensemble created by the neighbourhood method and generalises the mean absolute error measure for deterministic forecasts. 

The focus over the next year will be on diagnosing the performance of both the model and assimilation. A planned development that is expected to enhance the system is the update of the background-error covariances used for data assimilation. 

How to cite: Barbosa Aguiar, A., Waters, J., Price, M., Inverarity, G., Pequignet, C., Maksymczuk, J., Smout-Day, K., Martin, M., Bell, M., While, J., King, R., Lea, D., and Siddorn, J.: The Met Office operational global ocean forecast system FOAM-ORCA12, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-8491, https://doi.org/10.5194/egusphere-egu21-8491, 2021.

15:32–15:34
|
EGU21-2663
|
Highlight
Rachid Benshila, Julien Jouanno, Léo Berline, Antonin Soulié, Marie-Hélène Marie-Hélène, Guillaume Morvan, Frédéric Diaz, Julio Sheinbaum, Cristele Chevalier, Thierry Thibaut, Thomas Changeux, Frédéric Menard, Sarah Berthet, Olivier Aumont, Christian Ethé, Pierre Nabat, and Marc Mallet

The Tropical Atlantic is facing a massive proliferation of Sargassum since 2011, with severe environmental and socioeconomic impacts. The development of Sargassum modelling is essential to clarify the link between Sargassum distribution and environmental conditions, and to lay the groundwork for a seasonal forecast on the scale of the Tropical Atlantic basin. We present here a modelling framework based on the NEMO ocean model which integrates transport by currents and waves, stranding at the coast, and physiology of Sargassum with varying internal nutrients quota. The model is initialized from basin scale satellite observations and performance was assessed over the Sargassum year 2017. Model parameters are calibrated through the analysis of large ensembles of simulations, and the sensitivity to forcing fields like riverine nutrients inputs, atmospheric deposition, and waves is investigated. Overall, results demonstrate the ability of the model to reproduce the seasonal cycle and large-scale distribution of Sargassum biomass.

How to cite: Benshila, R., Jouanno, J., Berline, L., Soulié, A., Marie-Hélène, M.-H., Morvan, G., Diaz, F., Sheinbaum, J., Chevalier, C., Thibaut, T., Changeux, T., Menard, F., Berthet, S., Aumont, O., Ethé, C., Nabat, P., and Mallet, M.: A NEMO-based model of Sargassum distribution in the Tropical Atlantic , EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-2663, https://doi.org/10.5194/egusphere-egu21-2663, 2021.

15:34–15:36
|
EGU21-11156
|
ECS
Polina Verezemskaya, Bernard Barnier, Jean-Marc Molines, Sergey Gulev, and Alexander Gavrikov

A regional model of Subpolar Gyre in the North Atlantic is implemented. The NNATL12 model development aimed at a realistic representation of Subpolar Northern Atlantic's complex dynamics during the satellite era (starting from 1993 to nowadays) by using a high-resolution regional model that relies on the most up-to-date atmospheric and lateral forcing datasets and modeling techniques. Configuring this model, we focused on the representation of key processes in the Northern Atlantic, such as Irminger Rings, the boundary currents, deep convection, and convective eddies, dense waters cascading through the narrow straits between the Arctic and the Atlantic basins. NNATL12 model is based on NEMO4. The model domain covers the area between 47-70˚N and 84˚W-10˚E with a grid of 1/12˚ in horizontal and 75 vertical levels. In this region, the model is partially eddy-resolving. Three lateral open boundaries and initial conditions are set from the new GLORYS12 reanalysis (Lellouche et al., 2018). The surface forcing is provided by the new RAS NAAD dynamical hindcast based on the WRF model with a spatial resolution of 14 km (Gavrikov et al. 2020). The model adopted the most recent developments in the forced ocean modeling, such as upper boundary forcing schemes (Renault et al., 2020, Brodeau et al., 2016) and local-sigma vertical coordinate in the area of the overflows (Colombo et al., 2020). The model solution is sensitive to new parameterizations and vertical coordinate, which is demonstrated in various tests. The model provides a reliable estimate of the Subpolar North Atlantic circulation system at the surface and medium depth compared to observations. The model represents the ocean stratification at depths above 2000 m showing higher temperatures in the bottom of the Irminger Sea. At daily timescales, it is capable of representing the volume transport comparable to observed values. Irminger Rings TS-structure and dynamics are simulated consistent with the glider data. Comparing to the reanalysis model overestimates the March mixed layer depths and overextends the region of convection north. At the same time, the short-scale and decadal variability of MLD are reproduced by the model. Significant improvements of the deep stratification are obtained with the implementation of the local-sigma vertical coordinate. The model provides vertical profiles of temperature and salinity similar to the observed ones. However the Denmark Strait overflow waters are still too warm, but this is for a large part due to too warm waters at the sill. The high-frequency variability in the Denmark Strait is also in good accordance with the observations.

How to cite: Verezemskaya, P., Barnier, B., Molines, J.-M., Gulev, S., and Gavrikov, A.: The regional model of Subpolar Gyre based on NEMO 4, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-11156, https://doi.org/10.5194/egusphere-egu21-11156, 2021.

15:36–15:38
|
EGU21-8728
Tuomas Kärnä, Ida Ringgaard, Vasily Korabel, Adam Nord, Patrik Ljungemyr, Saeed Falahat, Lars Axell, Anja Lindenthal, Simon Jandt-Scheelke, Ilja Maljutenko, and Svetlana Verjovkina

We present Nemo-Nordic 2.0, the latest version of the operational marine forecasting model for the Baltic Sea used and developed in the Baltic Monitoring Forecasting Centre (BAL MFC) under the Copernicus Marine Environment Monitoring Service (CMEMS). The most notable differences between Nemo-Nordic 2.0 and its predecessor Nemo-Nordic 1.0 are the switch from NEMO 3.6 to NEMO 4.0 and an increase in horizontal resolution from 2 to 1 nautical mile. In addition, the model's bathymetry and bottom friction formulation have been updated. The model configuration was specially tuned to represent Major Baltic Inflow events. Focusing on a 2-year validation period from October 1, 2014, covering one Major Baltic Inflow event, Nemo-Nordic 2.0 simulates Sea Surface Height (SSH) well: centralized Root-Mean-Square Deviation (CRMSD) is within 10 cm for most stations outside the Inner Danish Waters. CRMSD is higher at some stations where small-scale topographical features cannot be correctly resolved. SSH variability tends to be overestimated in the Baltic Sea and underestimated in the Inner Danish Waters. Nemo-Nordic 2.0 represents Sea Surface Temperature (SST) and Salinity (SSS) well, although there is a negative bias around -0.5°C in SST. The 2014 Major Baltic Inflow event is well reproduced. The simulated salt pulse agrees well with observations in the Arkona basin and progresses into the Gotland basin in 3 to 4 months.

How to cite: Kärnä, T., Ringgaard, I., Korabel, V., Nord, A., Ljungemyr, P., Falahat, S., Axell, L., Lindenthal, A., Jandt-Scheelke, S., Maljutenko, I., and Verjovkina, S.: Nemo-Nordic 2.0: Updated Baltic Sea model based on NEMO 4.0, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-8728, https://doi.org/10.5194/egusphere-egu21-8728, 2021.

15:38–15:40
|
EGU21-12802
|
ECS
Marion Bezaud, Julie Deshayes, Stéphane Pous, and Julien Jouanno

Recently, the ocean dynamics of the Caribbean region has seen growing interest due the societal consequences of Sargassum beaching and storm surges, among other occasional extreme phenomena. Understanding the hydrodynamics in this area (mean currents and water mass properties, and mechanisms of variability) becomes urgent, to support operational developments forecasting the occurrence of such extreme phenomena, and also before one can foresee the local impacts of climate change. Building from an existing regional configuration at 1/12º (~10km), we implemented version 4.0.5 of NEMO to study the ocean dynamics of the Caribbean archipelago. This preliminary configuration is used to support sensitivity studies to atmospheric conditions, over the past 20 years. It also hosts AGRIF zooms to refine grid resolution up to 1km in the vicinity of the French islands, to enable a better understanding of the local dynamics.

How to cite: Bezaud, M., Deshayes, J., Pous, S., and Jouanno, J.: NEMO in Caribbean archipelago, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-12802, https://doi.org/10.5194/egusphere-egu21-12802, 2021.

15:40–15:42
|
EGU21-10122
|
ECS
Guillaume Boutin, Einar Ólason, Pierre Rampal, Camille Lique, Claude Talandier, and Laurent Brodeau

Sea ice is a key component of the earth’s climate system as it modulates air-sea interactions in polar regions. These interactions strongly depend on openings in the sea ice cover, which are associated with fine-scale sea ice deformations. Visco-plastic sea ice rheologies used in most numerical models struggle at representing these fine-scale sea ice dynamics without going to very costly horizontal resolutions (~1km). A solution is to use damage propagation sea ice models, which were shown to reproduce well sea ice deformations with little dependency on the mesh resolution. 

Here we present results from the first ocean--sea-ice coupled model using a rheology with damage propagation. The ocean component is the NEMO-OPA model. The sea ice component is neXtSIM, introducing the newly developed Brittle Bingham-Maxwell rheology. Results show that sea ice dynamics are very well represented from large scales (sea ice drift) to small-scales (sea ice deformation). Sea ice properties relevant for climate, i.e volume and area, also show a remarkable match with satellite observations. This coupled framework opens new opportunities to quantify the impact of small-scale sea ice dynamics on ice-ocean interactions.

How to cite: Boutin, G., Ólason, E., Rampal, P., Lique, C., Talandier, C., and Brodeau, L.: A new coupled model to shed light on sea-ice--ocean interactions, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-10122, https://doi.org/10.5194/egusphere-egu21-10122, 2021.

15:42–15:44
|
EGU21-8461
Renaud Person, Martin Vancoppenolle, Olivier Aumont, and Manon Malsang

Glacial iron (Fe) sources associated with continental ice (ice shelves and icebergs) and sea ice have recently been suggested as important to Southern Ocean (SO) biogeochemistry, where Fe limits primary production. Icebergs and ice shelves act as fully external sources of Fe while sea ice, which has a great Fe storage capacity, efficiently conveys Fe from the coasts to offshore locations. Large Fe concentrations in sea ice are typically explained by a sedimentary origin, however recent observations suggest an additional contribution from continental ice to the sea ice Fe inventory. Here, to further explore this hypothesis, we analyze factorial simulations performed with an ocean sea-ice biogeochemical model (NEMO-LIM3-PISCES version 3.6) in which interactive Fe sources from continental and marine glacial sources are activated, separately and in concert. Our simulations indicate that (i) about 15% of the iron content of sea ice comes from icebergs and ice shelves, (ii) sea ice motion conveys this extra Fe to regions where it limits productivity, which results in (iii) a modest increase in primary and export production, reaching ~1% of the SO total, or ~10% of the contribution of the SO cryosphere.

How to cite: Person, R., Vancoppenolle, M., Aumont, O., and Malsang, M.: Iron fertilization of the Southern Ocean: Synergy between sea ice, icebergs and ice shelves, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-8461, https://doi.org/10.5194/egusphere-egu21-8461, 2021.

15:44–15:46
|
EGU21-944
|
ECS
Juliana M. Marson and Paul G. Myers

Icebergs represent around half of the yearly mass discharge from the Greenland Ice Sheet. They are not only important freshwater sources, but also pose a threat to navigation and other offshore activities. Since monitoring individual icebergs in large numbers is unfeasible, numerical models are great tools to evaluate their role in freshwater distribution and their general trajectory patterns. While large-scale iceberg modelling is in its infancy, we show recent model improvements done in the Nucleus for European Modelling of the Ocean (NEMO) iceberg module. Among those, we highlight a newly implemented iceberg-sea ice dynamic, where icebergs are locked in concentrated and strong sea ice packs, so they will move with sea ice instead of across it. Additionally, recent code modifications allow the user to choose if the iceberg melt plume is inserted in the ocean’s first model layer or distributed along the iceberg draft. Results will show if these code upgrades change the way freshwater is distributed in the ocean and if they better represent iceberg trajectories and their surge seasonality off the Labrador shelf.

How to cite: M. Marson, J. and Myers, P. G.: Iceberg modelling with NEMO, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-944, https://doi.org/10.5194/egusphere-egu21-944, 2021.

15:46–15:48
|
EGU21-10776
Laura Stefanizzi, Stefania Ciliberti, Mehmet Ilicak, and Giovanni Coppini

Setting new model configurations based on NEMO requires the definition of initial/boundary condition and the validation of numerical solutions. In the framework of IMMERSE H2020 project, CMCC is developing new tools and technological capacities for handling in easy and reliable way external products, such CMEMS or coastal ocean data, for research-to-operations applications. Generic Interfaces for NEMO (InterNEMO) allow for 3 main scopes: 1) to access and discover the CMEMS catalogue, including both model and observational data; 2) to manipulate accessed datasets, including coastal ocean data, to extract relevant physical information to use for setting initial/boundary conditions for a new NEMO-based configurations; 3) to prepare NEMO set of upstream files and to validate NEMO solution by using CMEMS observational datasets. InterNEMO implements also technologies to connect a NEMO user to Wekeo DIAS (https://www.wekeo.eu/) for the interoperable accessing and processing of CMEMS data. In this contribution, we present the InterNEMO architecture developed in Python via Jupyter Notebooks, to support the user/researcher to easily discover, design and configure modeling components required by the new NEMO-based configuration. InterNEMO is tested for the Black Sea hydrodynamical model configuration, developed by CMCC in the framework of the Black Sea Monitoring and Forecasting Centre (BS-MFC) for CMEMS a) to show how to access CMEMS observations through Wekeo DIAS and use them to validate numerical solutions and b) to define open boundary conditions from an unstructured grid model configuration based on Shyfem, developed for the Marmara Sea.

How to cite: Stefanizzi, L., Ciliberti, S., Ilicak, M., and Coppini, G.: Tools and technologies for NEMO models: the case of the Generic Interfaces developed in the framework of IMMERSE, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-10776, https://doi.org/10.5194/egusphere-egu21-10776, 2021.

Sub session "NEMO developments" - Chair: Julien Le Sommer, co-chair: Mike Bell
15:48–15:50
|
EGU21-12427
Stella Valentina Paronuzzi Ticco, Oriol Tintó Prims, Mario Acosta Cobos, and Miguel Castrillo Melguizo

At the beginning of 2021 a mixed precision version of the NEMO code was included into the official NEMO repository. The implementation followed the approach presented in Tintó et al. 2019. The proposed optimization despite being not at all trivial, is not new, and quite popular nowadays. In fact, for historical reasons many computational models over-engineer the numerical precision, which leads to an under-optimal exploitation of computational infrastructures. By solving this miss-adjustment a conspicuous payback in terms of efficiency and throughput can be gained: we are not only taking a step toward a more environmentally friendly science, sometimes we are actually pushing the horizon of experiment feasibility a little further. For being able to smoothly include the changes needed in the official release an automatic workflow has been implemented: we attempt to minimize the number of changes required and, at the same time, maximize the number of variables that can be computed using single precision. Here we present a general sketch of the tool and workflow used.
Starting from the original code, we automatically produce a new version of the same, where the user can specify the precision of each real variable therein declared. With this new executable, a numerical precision analysis can be performed: a search algorithm specially designed for this task will drive a workflow manager toward the creation of a list of variables that is safe to switch to single precision. The algorithm compares the result of each intermediate step of the workflow with reliable results from a double precision version of the same code, detecting which variables need to retain a higher accuracy.
The result of this analysis is eventually used to perform the modification needed into the code in order to produce the desired working mixed precision version, while also keeping the number of necessary changes low. Finally, the previous double precision and the new mixed precision versions will be compared, including a computational comparison and a scientific validation to prove that the new version can be used for operational configurations, without losing accuracy and increasing the computational performance dramatically.

How to cite: Paronuzzi Ticco, S. V., Tintó Prims, O., Acosta Cobos, M., and Castrillo Melguizo, M.: An automatic implementation of the mixed precision in NEMO 4.2, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-12427, https://doi.org/10.5194/egusphere-egu21-12427, 2021.

15:50–15:52
|
EGU21-4762
Italo Epicoco, Silvia Mocavero, Francesca Mele, Alessandro D'Anca, and Giovanni Aloisio

One of the main bottlenecks for NEMO scalability is the time spent performing communications. Two complementary strategies are here proposed to reduce the communication frequency and the communication time: the MPI3 neighbourhood collective communications instead of multiple point to point exchanges and the increasing of the halo region size.

NEMO performs Lateral Boundaries Conditions update by using four point to point MPI communications at north, south, east and west for each MPI domain. The model completes east-west exchange before performing north-south communications. The order of the exchanges allows us to preserve both 5-point and 9-point stencils. MPI3 neighbourhood collectives provide a way to have sub-communicators used to perform collective communications. Two different sub-communicators can be defined in order to support the two different stencils. A single MPI message is needed to be built for all neighbours instead of 4 different messages before calling the collective communication, while the received message is used to update the halo region, following the order of the neighbours in the sub-communicator.

The new communication strategy has been tested on two computational kernels (i.e. one for 5-point stencil and one for 9-point stencil), selected among the main relevant routines from the computational point of view. Preliminary tests, performed on a domain size of 3000x2000x31 grid points on the Zeus Intel Xeon Gold 6154 machine, available at CMCC, show a gain in communication time for the 5-point stencil use case up to 31% on 2016 cores. The improvement is reduced when communications with processes on the diagonal are activated. However, a modest gain is still achieved, depending on the number of cores.

On the other side, the analysis of some NEMO routines shows how the exchange of more than one row/column of halo would allow to move communications outside the routine, preserving data dependencies. A wider halo size reduces the frequency of message exchanges whilst increases the message size at each exchange. It allows us to adopt some optimisation strategies (i.e. loop fusion, tiling, etc.) to improve the data locality. Nevertheless, the use of a wider halo introduces itself some improvements for some kernels like for the MUSCL advection scheme which shows a gain of ~23% in the execution time comparing the original version and the new one with halo extended to 2 lines and the communication moved outside the computing region.

The current work has been performed according to the NEMO development strategy plan, defined by the NEMO Consortium, which establish the priorities of the design strategies to reduce the bottlenecks to the scalability and the time to solution.

 

Acknowledgments

This work is co-funded by the EU H2020 IS-ENES project Phase 3 (ISENES3) under Grant Agreement number 824084.

How to cite: Epicoco, I., Mocavero, S., Mele, F., D'Anca, A., and Aloisio, G.: New communication strategies in NEMO, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-4762, https://doi.org/10.5194/egusphere-egu21-4762, 2021.

15:52–15:54
|
EGU21-10970
Maicon Faria, Mario Acosta, Miguel Castrillo, Stella V. Paronuzzi Ticco, Sergi Palomas, David Vicente Dorca, and kim Serradell Maronda

This work makes part of an effort to make NEMO capable of taking advantage of modern accelerators. To achieve this objective we focus on port routines in NEMO that have a small impact on code maintenance and the higher possible overall time footprint reductions. Our candidates to port were the diagnostic routines, specifically diahsb (heat, salt, volume budgets) and diawri (Ocean variables) diagnostics. These two diagnostics correspond to 5% of the NEMO's runtime each on our test cases. Both can be executed in an asynchronous fashion allowing overlap between diagnostic GPU and other NEMO routines CPU computations.
We report a methodology to port runtime diagnostics execution on NEMO to GPU using CUDA Fortran and OpenACC. Both synchronous and asynchronous are implemented on diahsb and diawri diagnostics. Associated time step and stream interleave are proposed to allow the overlap of CPU execution of NEMO and data communication between CPU, and GPU.

In the case of constraint computational resources and high-resolution grids, synchronous implementation of diahsb and diawri show up to 3.5x speed-up. With asynchronous implementation we achieve a higher speed-up from 2.7x to 5x with diahsb in the study cases. The results for this diagnostic optimization point out that the asynchronous approach is profitable even in the case where plenty of computational resources are available and the number of MPI ranks is in the threshold of parallel effectiveness for a given computational workload. For diawri on the other hand, the results of the asynchronous implementation depart from the diahsb. In the diawri diagnostic module there are 30 times more datasets demanding pinned memory to overlap communication between CPU and GPU with CPU execution. Pinned memory attribute limits data management of datasets allocated on main memory, therefore makes possible to the GPU access to main memory, overlapping CPU computation. The result is a scenario where the improvement from offloading the diagnostic computation impacts on NEMO CPU general execution. Our main hypothesis is that the amount of pinned memory used decreases the performance on runtime data management, this is confirmed by the 7% increase of the L3 data cache misses in the study case. Although the necessity of evaluating the amount of datasets needed for asynchronous communication on a diagnostic port, the payout of asynchronous diagnostic may be worth given the higher speed-up values that we can achieve with this technique. This work proves that models such as NEMO, developed only for CPU architectures, can port some of their computation to accelerators. Additionally, this work explains a successful and simple way to implement an asynchronous approach, where CPU and GPU are working in parallel, but without modifying the CPU code itself, since the diagnostics are extracted as kernels for the GPU and the CPU is yet working in the simulation.

How to cite: Faria, M., Acosta, M., Castrillo, M., V. Paronuzzi Ticco, S., Palomas, S., Vicente Dorca, D., and Serradell Maronda, K.: Porting NEMO diagnostics to GPU accelerators, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-10970, https://doi.org/10.5194/egusphere-egu21-10970, 2021.

15:54–15:56
|
EGU21-9078
Nicolas Ducousso, Florian Lemarié, Gurvan Madec, and Laurent Debreu

The NEMO ocean model is currently based on the Leapfrog scheme that provides a good combination between simplicity and efficiency for low-resolution global simulations. However, this scheme is subject to difficulties that question its relevance at high-resolution : the necessary damping of its computational mode, e.g. via a Robert-Asselin filter, affect stability and increases amplitude and phase errors of the physical mode ; because it is unconditionally unstable for diffusive processes, monotonicity or positive-definiteness comes at a substantial cost and complication. The evolution toward a 2-level time stepping algorithm based on Runge-Kutta schemes is studied. Special attention is given to how to articulate a mode-splitting technique to handle the fast dynamics associated with the free surface. Linear stability analyses of several Runge-Kutta based, split-explicit algorithms are performed and the most promising ones are identified. They allow a good compromise between robustness, stability and accuracy for integration of internal gravity waves, Coriolis and advection processes. Idealized test-cases illustrate the benefits associated to the revised time-stepping compared to the original Leapfrog.

How to cite: Ducousso, N., Lemarié, F., Madec, G., and Debreu, L.: Design of Runge-Kutta based split-explicit time integration algorithms for the NEMO ocean model, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-9078, https://doi.org/10.5194/egusphere-egu21-9078, 2021.

15:56–15:58
|
EGU21-1296
|
ECS
Amy Young and Mike Bell

Terrain following coordinates allow for better representation of physics at the sea-bed than traditional z-coordinates but result in numerical discretisation errors in the calculation of the horizontal pressure gradient (HPG) which manifest as spurious currents.  As of NEMO r4.0.4, there were two HPG schemes available for use with terrain following coordinates, the traditional 2nd order sco scheme and the 3rd order prj scheme.  The prj scheme, while highly accurate in the ocean interior, shows unphysical behaviour at the sea-bed for steeply sloping bathymetry.  A task in the IMMERSE project was set up to identify, implement and test promising HPG schemes suitable for general vertical coordinates that are accurate, robust and physically consistent.  As part of this task, the 3rd-order accurate density Jacobian scheme (djc) as proposed by Shchepetkin and McWilliams (2003) has now been implemented in the NEMO trunk (as a rewrite of the previously existing but non-operational djc scheme).  Idealised testing has shown this scheme to be significantly more accurate than the sco scheme, and more robust than the prj scheme in coping with steeply sloping bathymetry.  Initial results from applying the djc scheme in a challenging realistic configuration (the AMM7 with hybrid s-z-coordinates and non-uniform vertical discretisation) show a reduction in spurious currents with respect to the sco scheme.  The prj scheme is highly sensitive to the rmax (maximum permitted slope) criterion.  In cases where the bathymetry is so steep that a velocity-point may lie multiple levels below one of its neighbouring tracer-points, the nature of the prj near-bed HPG calculation leads to sudden spin-ups of spurious velocities which can exceed those of the djc scheme in the longer-term.  Performance-wise, the djc scheme is 3 times slower than the sco scheme, but less expensive than the prj.  Further work is planned to reduce the memory footprint.  In addition to continued testing of the djc scheme, further work will look at alternative formulation (finite volume) HPG schemes, and high order variants.

This work is distributed under the Creative Commons Attribution 4.0 License. This licence does not affect the Crown copyright work, which is re-usable under the Open Government Licence (OGL). The Creative Commons Attribution 4.0 License and the OGL are interoperable and do not conflict with, reduce or limit each other.

How to cite: Young, A. and Bell, M.: Exploring horizontal pressure gradient (HPG) schemes for general vertical coordinates., EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-1296, https://doi.org/10.5194/egusphere-egu21-1296, 2021.

15:58–16:00
|
EGU21-4152
|
ECS
Diego Bruciaferri, James Harle, Anthony Wise, Enda O'Dea, and Jeff Polton

The choice of the vertical coordinate system is the single most important factor affecting the quality of ocean model simulations (e.g. Griffies et al. 2000). This is especially true in regions such as the European North-West Shelf (NWS), where complex ocean dynamics result from the combination of a variety of multi-scale physical processes.

As part of the Copernicus Marine Environment Monitoring Service, the Met Office runs an operational coupled ocean-wave forecasting system of the NWS. The ocean model employed is a regional implementation of NEMO hydrodynamic code (Madec 2017), further developed by both the Met Office and the National Oceanography Centre under the umbrella of the Joint Marine Modelling Programme (JMMP). Here we describe the work of the JMMP group in assessing the impact of different vertical coordinate systems on the accuracy of the solution of the free-running NWS ocean model. 

Five different vertical discretization schemes are compared: i) geopotential z-levels with partial steps, ii) s-levels following a smooth version of the bottom topography using either the Song & Haidvogel (1994) or iii) the Siddorn & Furner (2013) stretching functions, iv) the hybrid Harle et al. (2013) s-z with partial step scheme, and v) the multi-envelope s-coordinate system of Bruciaferri et al. (2018). Three different type of numerical experiments with increasing level of complexity are conducted: i) an idealised test for horizontal pressure gradient errors (HPGE), ii) a barotropic simulation forced only by the astronomical tides (TIDE) and iii) a fully baroclinic simulation using realistic initial condition and external forcing (REAL).      

Numerical results of the HPGE test show that s-levels models develop the highest spurious currents (order of cm/s),  the multi-enveloping method allows relatively reduction of the error of pure s-levels grids while z-levels with partial steps or the hybrid s-z scheme are affected by the smallest error (order of mm/s).  The TIDE experiment reveals some differences between the models for amplitude and phase of the major tidal components. Preliminary results of the REAL experiment show that models differing only in the vertical discretization schemes broadly represent the same general ocean dynamics, although presenting non-trivial differences in the active tracers and flow fields especially in the proximity of the shelf-break.        

Song, Y. & Haidvogel, D.B., 1994. A semi-implicit ocean circulation model using a generalized topography-following coordinate system. Journal of Computational Physics 115, 228–244

Griffies, S.M. et al. 2000. Developments in ocean climate modelling. Ocean Modelling 2, 123–192, 10.1016/S1463-5003(00)00014-7

Siddorn, J.R. & Furner, R., 2013. An analytical stretching function that combines the best attributes of geopotential and terrain-following vertical coordinates. Ocean Modelling 66, 1–13, 10.1016/j.ocemod.2013.02.001

Harle, J.D. et al. 2013. Report on role of biophysical interactions on basin-scale C and N budgets. Deliverable 6.5, European Basin-scale Analysis, Synthesis and Integration (EURO-BASIN) Project, http://eurobasin.dtuaqua.dk/eurobasin/documents/deliverables/D6.5%20Report%20on%20role%20of%20biophysical%20interactions%20on%20C%20N%20budget.pdf

Madec G. et al. (2017). NEMO ocean engine. Notes Du Pôle De Modélisation De L'institut Pierre-simon Laplace (IPSL). http://doi.org/10.5281/zenodo.3248739

Bruciaferri, D. et al. 2018. A multi-envelope vertical coordinate system for numerical ocean modelling. Ocean Dynamics, 68 (10), 1239-1258, 10.1007/s10236-018-1189-x

How to cite: Bruciaferri, D., Harle, J., Wise, A., O'Dea, E., and Polton, J.: The impact of the vertical discretization scheme on the accuracy of a model of the European north-west shelf, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-4152, https://doi.org/10.5194/egusphere-egu21-4152, 2021.

16:00–16:02
|
EGU21-2234
Yves Morel, Rachid Benshila, Benoit Tranchant, Jerome Chanut, Brian Arbic, Damien Allain, Florent Lyard, Loren Carrere, and Ariane Koch-Larrouy

This study proposes a new methodology for implementing the barotropic tide in an ocean general circulation model (OGCM). The assumptions underlying this methodology are that the best barotropic tide solutions are computed by specialized models and that the fields that should be accurately reproduced by the OGCM are the transport fields from the specialized tide model. The target/reference solution for the OGCM is thus the projection of the tide model on the OGCM grid, for each tidal harmonic.

The proposed methodology involves little change of the OGCM modeland yields almost exactly the reference solution, with a cost that is belowmost of the current methodologies. It relies on the modification of the tidepotential, or more accurately, on the replacement of all terms associatedwith the tide (tide potential, self attraction and loading, tide dissipation,  ...) by a general tide forcing term in the barotropic momentum equationwhich is calculated from the –known- reference solution.

The tide forcing terms can be tricky to calculate as they depend on details of the OGCM numerical schemes (for both temporal and spatial operators). A general procedure, automatically adapting the chosen schemes, is proposed for their calculation, so that the procedure is independent of the model.  

Tests with academic configurations are first proposed to validate the methodology and its implementation, and the OGCM is chosen to be the NEMO (Nucleus for European Modelling of the Ocean) model.

A global ¼° configuration with realistic bathymetry and with FES tide solutions (Finite Element Solution) are then performed. Current tests show that when FES solutions are crudely interpolated on the NEMO grid, the methodology exactly reproduces the FES fluxes, but the associates NEMO SSH is very noisy in regions where FES has high resolution. This problem is currently addressed. To get rid of this problem, fluxes must be carefully integrated along each grid cell, so that the reproduced SSH is exactly an average of the FES SSH within the NEMO grid cell. Hopefully, we will be able to present final –clean- solutions at the conference.

How to cite: Morel, Y., Benshila, R., Tranchant, B., Chanut, J., Arbic, B., Allain, D., Lyard, F., Carrere, L., and Koch-Larrouy, A.: Implementation of the barotropic tide in oceanic circulation models. Current tests with the NEMO model , EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-2234, https://doi.org/10.5194/egusphere-egu21-2234, 2021.

16:02–16:04
|
EGU21-13489
Jerome Chanut, James Harle, Tim Graham, and Laurent Debreu

The NEMO platform possesses a versatile block-structured refinement capacity thanks to the AGRIF library. It is however restricted up to versions 4.0x, to the horizontal direction only. In the present work, we explain how we extended the nesting capabilities to the vertical direction, a feature which can appear, in some circumstances, as beneficial as refining the horizontal grid.

Doing so is not a new concept per se, except that we consider here the general case of child and parent grids with possibly different vertical coordinate systems, hence not logically defined from each other as in previous works. This enables connecting together for instance z (geopotential), s (terrain following) or eventually ALE (Arbitrary Lagrangian Eulerian) coordinate systems. In any cases, two-way exchanges are enabled, which is the other novel aspect tackled here.  

Considering the vertical nesting procedure itself, we describe the use of high order conservative and monotone polynomial reconstruction operators to remap from parent to child grids and vice versa. Test cases showing the feasibility of the approach are presented, with particular attention on the connection of s and z grids in the context of gravity flow modelling. This work can be considered as a preliminary step towards the application of the vertical nesting concept over major overflow regions in global realistic configurations. The numerical representation of these areas is indeed known to be particularly sensitive to the vertical coordinate formulation. More generally, this work illustrates the typical methodology from the development to the validation of a new feature in the NEMO model.

How to cite: Chanut, J., Harle, J., Graham, T., and Debreu, L.: Two-way nesting ocean models with different vertical coordinates, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-13489, https://doi.org/10.5194/egusphere-egu21-13489, 2021.

16:04–16:06
|
EGU21-2782
Alex Megann, Jerome Chanut, and Dave Storkey

The eddy-permitting 1/4° resolution in NEMO has been known to suffer from significant numerical diapycnal mixing. This arises from truncations in the advection scheme, which causes spurious mixing of tracers where there are transient vertical motions from internal tides and near-inertial waves, as well as from computational modes associated with partly-resolved mesoscale features. Suppressing the near-gridscale noise by increasing the viscosity has been shown to offer a useful reduction in that contribution to numerical mixing, but does not have a significant effect on tides and inertial waves.

The z~ scheme replaces eulerian vertical tracer advection across the vertical coordinate surfaces, on time scales less than a few days, with displacements of the coordinate surfaces themselves, in a manner more consistent with the nearly adiabatic nature of near-inertial gravity waves and tides. This has been shown to give substantial reduction in numerical mixing in an idealised configuration, but has yet to be fully evaluated in a global ocean domain. It is shown, using a new prototype eORCA025 global NEMO configuration, that z~ with the default filter timescales reduces the effective diapycnal diffusivity and temperature drifts by only about 10%. Preliminary results will be presented for the sensitivity of the numerical mixing to the z~ timescale and other parameters. The application of z~ to a tidally-forced simulation will also be discussed.

How to cite: Megann, A., Chanut, J., and Storkey, D.: Evaluation of the z-tilde vertical coordinate in a 1/4° global NEMO, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-2782, https://doi.org/10.5194/egusphere-egu21-2782, 2021.

Sub session "Parametrisations and Community building" - Chair Claire Lévy, co-chair: Julie Deshayes
16:06–16:08
|
EGU21-10069
|
ECS
Marcus Falls, Martí Galí Tàpias, Raffaele Bernardello, and Miguel Castrillo

When working with Earth system models, a considerable challenge that arises is the need to establish the set of parameter values that ensure the optimal model performance in terms of how they reflect real-world observed data. Given that each additional parameter under investigation increases the dimensional space of the problem by one, simple brute-force sensitivity tests can quickly become too computationally strenuous. In addition, the complexity of the model and interactions between parameters mean that testing parameters on an individual basis has the potential to miss key information. As such, this work argues the need of the development of a tool that can give an estimation of parameters. Specifically it proposes the use of a Biased Random Key Genetic Algorithm (BRKGA). This method is tested using the one dimensional configuration of PISCES, the biogeochemical component of NEMO, a global ocean model. A test case of particulate organic carbon in the North Atlantic down to 1000m depth is examined. In this case, two tests are run, one where each of the model outputs are compared to the model outputs with default parameters, and another where they are compared with 3 sets of observed data from their respective regions, which is followed by a cross reference of the results. The results of these analyses provide evidence that this approach is robust and consistent, and also that it provides indication of the sensitivity of parameters on variables of interest. Given the deviation of the optimal set of parameters from the default, further analyses using observed data in other locations is recommended to establish the validity of the parameters.

How to cite: Falls, M., Galí Tàpias, M., Bernardello, R., and Castrillo, M.: Use of Genetic Algorithms for Ocean Model Parameter Optimisation, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-10069, https://doi.org/10.5194/egusphere-egu21-10069, 2021.

16:08–16:10
|
EGU21-12418
|
ECS
Stephan Juricke, Sergey Danilov, Marcel Oliver, Nikolay Koldunov, Dmitry Sidorenko, and Dmitry Sein

Capturing mesoscale eddy dynamics is crucial for accurate simulations of the large-scale ocean currents as well as oceanic and climate variability. Eddy-mean flow interactions affect the position, strength and variations of mean currents and eddies are important drivers of oceanic heat transport and atmosphere-ocean-coupling. However, simulations at eddy-permitting resolutions are substantially underestimating eddy variability and eddy kinetic energy many times over. Such eddy-permitting simulations will be in use for years to come, both in coupled and uncoupled climate simulations. We present a set of kinetic energy backscatter schemes with different complexity as alternative momentum closures that can alleviate some eddy related biases such as biases in the mean currents, in sea surface height variability and in temperature and salinity. The complexity of the schemes reflects in their computational costs, the related simulation improvements and their adaptability to different resolutions. However, all schemes outperform classical viscous closures and are computationally less expensive than a related necessary resolution increase to achieve similar results. While the backscatter schemes are implemented in the ocean model FESOM2, the concepts can be adjusted to any ocean model including NEMO.

How to cite: Juricke, S., Danilov, S., Oliver, M., Koldunov, N., Sidorenko, D., and Sein, D.: Advancing the simulation of mesoscale eddies: Backscatter schemes in eddy-permitting ocean simulations, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-12418, https://doi.org/10.5194/egusphere-egu21-12418, 2021.

16:10–16:12
|
EGU21-12028
|
ECS
Katherine Hutchinson, Julie Deshayes, and Pierre Mathiot

Currently, none of the global 1° ocean-climate coupled models used for the Coupled Model Intercomparison Project (CMIP) explicitly simulate sub-ice shelf cavity circulation. This circulation plays a critical role in global ocean overturning as it transforms salty water formed at the surface in Antarctica into the parent waters of Antarctic Bottom Water (AABW). A challenge that the ocean-climate modelling community faces is the inclusion of these ocean-ice shelf interactions in global ocean 1° resolution models, so as to explicitly simulate dense water production and export. Choices regarding various numerical schemes and parameterizations need to be made, but in testing sensitivity to these choices and feedback effects of biases, large super-computing costs associated with running a global configuration are incurred. To address this we present an adapted configuration of the Ice Shelf-Ocean Model Intercomparison Project (ISOMIP), named ISOMIP+K, as the default idealised ISOMIP+ setup is not appropriate for modelling the deep, cold Antarctic cavities responsible for forming the dense parent waters of AABW. ISOMIP+K is currently adapted for the NEMO ocean model, motivated by the fact that this model is used for 6 of the climate groups participating in CMIP. We present results from ISOMIP+K configurations for Filchner-Ronne, Larsen-C and Ross ice shelves, which are important for dense water formation and large enough to be resolved, albeit coarsely, in a global 1° Earth System Model. This adapted ISOMIP+K test case, which is now far from idealized, is used to test the effect of initial conditions, the choice of values for lateral diffusion of momentum, mixing, drag coefficients and bathymetry on key indicators describing melt, sub-ice shelf circulation and dense water export. As opposed to regional high resolution Southern Ocean configurations, the ISOMIP+K configurations are designed so that the lessons learnt are directly transferable to a global ocean configuration where each choice made is backed-up by extensive, yet affordable, testing.

How to cite: Hutchinson, K., Deshayes, J., and Mathiot, P.: Navigating the challenges of explicitly including ocean-ice shelf interactions in a global ocean model using an adapted ISOMIP+ configuration as a fit-for-purpose tool, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-12028, https://doi.org/10.5194/egusphere-egu21-12028, 2021.

16:12–16:22
|
EGU21-11707
|
solicited
Julie Deshayes

When comparing realistic simulations produced by two ocean general circulation models, differences may emerge from alternative choices in boundary conditions and forcings, which alters our capacity to identify the actual differences between the two models (in the equations solved, the discretization schemes employed and/or the parameterizations introduced). The use of idealised test cases (idealized configurations with analytical boundary conditions and forcings, resolving a given set of equations) has proven efficient to reveal numerical bugs, determine advantages and pitfalls of certain numerical choices, and highlight remaining challenges. I propose to review historical progress enabled by the use of idealised test cases, and promote their utilization when assessing ocean dynamics as represented by an ocean model. For the latter, I would illustrate my talk using illustrations from my own research activities using NEMO in various contexts. I also see idealised test cases as a promising training tool for inexperienced ocean modellers, and an efficient solution to enlarge collaboration with experts in adjacent disciplines, such as mathematics, fluid dynamics and computer sciences.

How to cite: Deshayes, J.: On the use of idealised test cases for ocean model development, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-11707, https://doi.org/10.5194/egusphere-egu21-11707, 2021.

16:22–17:00