Modeling

Modeling

Ocean, sea-ice, and biogeochemical (BGC) model components are tested extensively and exercised in studies of the global, regional, and coastal ocean and of the impacts on Earth system state, variability, and change.

In collaboration with the broader model development community, the following ocean models have been used in the Community Earth System Model (CESM):

MOM6

MOM6 is a mass-conserving, C-grid model with generalized vertical coordinates known as Arbitrary Lagrangian-Eulerian coordinates (ALE), primarily developed at the Geophysical Fluid Dynamics Laboratory (GFDL). It evolved from earlier versions of MOM, which had regular z or Z* vertical grids (Pacanowski and Griffies ) as well as the GFDL-GOLD isopycnal model (Hallberg and Adcroft).

Since CESM2 was released, members of the Oceanography Section, in collaboration with GFDL and other members of the CESM community, have been implementing MOM6 in CESM for the next major release (CESM3). This non-trivial process has involved adapting CESM to deal with a mass-conserving and C-grid model, helping the users with the ALE coordinates (output is provided on native and regular vertical coordinates), optimizing and incorporating new parameterizations, and handling diagnostics.  As of summer 2023, target configurations include a nominal ⅔° grid with parameterized eddies, ¼°, and 1/12°.

MOM6 is also being adapted for idealized configurations and regional configurations.

MOM6 has been available as an optional component since the CESM2.2 release.

GFDL MOM6 guide

NCAR GitHub pages:

• https://github.com › ESCOMP › MOM_interface
• https://github.com › NCAR › mom6-tools
• https://github.com › NCAR › regional_cesm_mom6

POP

The Parallel Ocean Program (POP) model was a collaboration between Los Alamos National Laboratory (LANL, DOE), NCAR, and the community. It evolved from the Bryan-Cox-Semtner model types. POP is a global, volume-conserving, B-grid model (Smith and Gent 2010, Danabasoglu et al., 2012) that has been used in two main grid configurations: The standard resolution POP2 is on a nominal 1° grid with 60 vertical levels, and 10m grid spacing in the upper 100m.  High resolution POP has nominal 0.1º horizontal resolution (decreasing from about 11km at the Equator to 2.5km at high latitudes) and uses a tripolar grid with two poles in the northern hemisphere over North America and Asia to overcome the north pole singularity, and 62 vertical levels, ranging from 10m spacing in the upper 200m to 250m spacing in the deep ocean, with the use of partial bottom cells.

The main parameterizations are of vertical mixing (K-Profile-Parameterization-KPP, Large et al. 1994, currently implemented in CVMIX), mesoscale eddies (used in the lower resolution configuration, Gent and McWilliams, 1990, Danabasoglu et al., 1995, Gent et. al., 1995), submesoscale eddies (Fox-Kemper, et. al., 2008), and dense overflows (Danabasoglu et al., 2010). 

POP has been used for a wide array of research within CCSM3, CCSM4, CESM1 and CESM2 (e.g., Large and Yeager 2009, Gent et al. 2010, Hurrell et al. 2013, Tsujino et al., 2020, Danabasoglu et al., 2020). Studies have used ocean-ice simulations forced by a best estimate of observed atmospheric state as well as fully coupled simulations.

Ocean BGC/MARBL

The Oceanography Section helped develop the Biogeochemical model (BGC) in collaboration with Keith Moore and colleagues and has recently implemented the Marine Biogeochemistry Library (MARBL) framework in CESM.

CICE

The sea-ice model and ocean model need to be developed synergistically. For example, currently the sea-ice model runs on the same grid as the ocean model (in CESM2 both models used a B-grid but complexity was added when C-grid MOM6 was introduced). Time stepping should also be coordinated. https://github.com/CICE-Consortium/About-Us  

Other models

Other models used in the Oceanography Section include the Regional Ocean Modelling System (ROMS, Shchepetkin and McWilliams 2005, Haidvogel et al. 2008, https://www.myroms.org/), both in stand-alone and in coupled configurations (e.g. the Regional Community Earth System Model, RCESM, Fu et al., 2021,  https://github.com/ihesp/rcesm1).

Ocean Forcing Datasets

Two ocean-forcing datasets have been used: CORE (Large and Yeager, 2004, 2009), and JRA55-do (Tsujino et al., 2018). CORE utilized wind fields at 10m and air temperature, humidity, and air density at 2m, from NCEP/NCAR reanalysis, together with surface radiation derived from satellite data and a radiative transfer model (Zhang et al., 2004) and precipitation derived from multiple datasets and observation types (Large and Yeager, 2009). It further included adjustments of these fields towards other reference datasets to help remove or reduce known deficiencies of the original dataset (for example, NCEP/NSF NCAR winds were adjusted towards QuiKSCAT wind estimates to increase too-weak wind speeds in the deep Tropics and elsewhere, Large and Yeager, 2009). In contrast, JRA55-do is based on fields from the JRA55 reanalysis (Kobayashi et. al., 2015):  some of the advantages of this reanalysis are the higher horizontal resolution (grid spacing around 0.5 compared to 2 for NCEP-NCAR), 3-hour output (vs 6 hours), and ingestion of a larger number of observations (satellite, in-situ). The modified dataset, referred to as JRA55-do (Tsujino et al., 2018), followed a similar adjustment method to CORE but with some differences; full details are given in Tsujino et. al., 2018. CORE covers the period 1948-2009, while JRA55-do has been produced from 1958 to the present.