Suggesting new realm "Ocean Waves"

[sol1105]

In the Ocean and Sea Ice Thematic Author group for the CMIP7 Data Request, it was suggested to add a new realm for "Ocean Waves". Some points to justify the addition of this new realm are listed in the following:

• Ocean waves interact with multiple components of the Earth system, including:
  • Ocean: They affect the mixed layer depth (MLD), turbulence kinetic energy (TKE), SST, ocean currents, surface fluxes such as momentum, latent and sensible heat, and CO2 exchange. Ocean waves also alter the ocean albedo.
  • Atmosphere: They impact the planetary boundary layer (PBL) vertical structure, influence fluxes between the ocean and atmosphere, and contribute to the generation of marine aerosols.
  • Sea Ice: Ocean waves impact sea ice fracturing, melting, and displacement.
  • Storm Surge & Coastal Erosion: Waves increase storm surge heights by causing water run-up, contributing to coastal erosion, and amplifying the reach of marine waters inland.
• Ocean-Atmosphere feedbacks: Ocean surface waves are the result of ocean-atmosphere interaction, impacting both, atmosphere and ocean, during their generation, propagation and dissipation phases. A new model realm might be needed to deal with this double "parenting".
• Increased amount and complexity of variables: The coupling between ocean waves and other system components adds complexity, and the number of wave-related variables is expected to increase. To better manage this complexity, i.e. organizing the wave related processes without overloading the atmosphere and ocean realms, and so better align with CMIP standards, a dedicated ocnWave model realm would be helpful.
• User-friendly separation: From the perspective of users, it would be practical to have a separate realm for ocean waves. This would simplify the process of finding and using wave-related data without having to navigate the broader ocean model realm.

The following revision paper by Babanin et al. (2023) can also be consulted for the discussion: https://doi.org/10.1029/2023JC019633

[andrelanfer]

It is really important to have wave data properly set into CMIP simulations. Ocean surface waves impacts everything at ocean-atmosphere interface. It is a need!

[durack1]

Hi @sol1105 and @andrelanfer thanks for raising this issue - and glad we've got a discussion open.

The current (CMIP6/CMIP6Plus-era) realms are included below

"realm": {
    "aerosol": "Aerosol",
    "atmos": "Atmosphere",
    "atmosChem": "Atmospheric Chemistry",
    "land": "Land Surface and Subsurface",
    "landIce": "Land Ice",
    "ocean": "Ocean",
    "ocnBgchem": "Ocean Biogeochemistry",
    "seaIce": "Sea Ice"
  },

A couple of questions, which would be useful to answer from a modelling group perspective, and model configuration perspective as we are trying to build an infrastructure and nomenclature that captures broad model suite features. Questions 1) What native grid does an ocean wave component use? 2) What output/request grid will an ocean wave use? 3) Are ocean wave models an offline or online component of the model? (e.g. COWCLIPx was offline sims using existing CMIPx output)

[andrelanfer]

Hi @sol1105 and @andrelanfer thanks for raising this issue - and glad we've got a discussion open.

The current (CMIP6/CMIP6Plus-era) realms are included below

"realm": {
    "aerosol": "Aerosol",
    "atmos": "Atmosphere",
    "atmosChem": "Atmospheric Chemistry",
    "land": "Land Surface and Subsurface",
    "landIce": "Land Ice",
    "ocean": "Ocean",
    "ocnBgchem": "Ocean Biogeochemistry",
    "seaIce": "Sea Ice"
  },

A couple of questions, which would be useful to answer from a modelling group perspective, and model configuration perspective as we are trying to build an infrastructure and nomenclature that captures broad model suite features. Questions

1. What native grid does an ocean wave component use?
2. What output/request grid will an ocean wave use?
3. Are ocean wave models an offline or online component of the model? (e.g. COWCLIPx was offline sims using existing CMIPx output)

Hi @durack1 Thanks to enhance this discussion. You pointed out very uselful questions. I will try to address the better I can. 1) wave model uses a single layer grid, and might be a regular grid ( with a certain resolution) or a non structured grid ( relative new feature). 2) every modeling center might think different, but for me, in order , to facilitated for the users, I think a regular grid should be fine. 3) wave models can be run as offline or coupled. I think this might be decided by each modelling center. Offline as done by COWCLIP, is possible. But also single coupling with ocean component, or single coupling with atmosphere component, or also single couplibg with sea-ice model. Can also be coupled with all of those. Online coupling allow the waves fields to affect and interact with the other components field, impacting the solutions of thoses, by changing surface characteristics. So each modelling center should decides their approaches.

[taylor13]

For addressing research questions, do you think users will want the wave model output to be reported on the same grid as the ocean variables or the atmosphere variables? Or would they prefer the wave data be reported on an entirely different grid? Or won't they care? Depending on resolution and regirdding method, wave data that has been regridded from its native grid would likely become smoothed a bit. For impact studies will that be a problem?

Do you think it is more common for the wave model to share the same native grid as the ocean or the atmosphere, or is it more common that the waves are calculated on an entirely different grid?

[andrelanfer]

Hi Taylor,

That's a quite good question. For research questions each user might want differently depends on what aspects they will focus on, i guess. For modelling centers, since to run the wave model needs variables,as input, from different models, ( wind at 10m from atmosphere model, and ice cover from sea-ice model, that some models shares the oceanic grid with), and might use surface currents and SST from ocean model, and also t2m from atmosphere.

I think an independent grid that optimises thoses regriddings for input should work fine. Em qua., 25 de set. de 2024 12:19, taylor13 @.***> escreveu:

For addressing research questions, do you think users will want the wave model output to be reported on the same grid as the ocean variables or the atmosphere variables? Or would they prefer the wave data be reported on an entirely different grid? Or won't they care? Depending on resolution and regirdding method, wave data that has been regridded from its native grid would likely become smoothed a bit. For impact studies will that be a problem?

Do you think it is more common for the wave model to share the same native grid as the ocean or the atmosphere, or is it more common that the waves are calculated on an entirely different grid?

— Reply to this email directly, view it on GitHub https://github.com/WCRP-CMIP/CMIP7_CVs/issues/3#issuecomment-2374395519, or unsubscribe https://github.com/notifications/unsubscribe-auth/AK7F2JLLN7XWJI6PWV46YT3ZYLH7DAVCNFSM6AAAAABO2CQ5H2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGNZUGM4TKNJRHE . You are receiving this because you were mentioned.Message ID: @.***>

[durack1]

@andrelanfer @taylor13 this is indeed a good discussion. Stating up front, we ideally want CMIP7 output to be AS USEABLE as possible, and so at the outset want to consider data usage (and access), along with making it easy for modelling groups to generate these data, and ESGF and other data providers to manage it. We also want the infrastructure to facilitate groups to document their model configurations (and data), and the realm and how this maps onto model components across groups is a key consideration - and a topic that the CMIP Model Documentation Task Team has been working very hard on (and I hope they chime in here too).

We've learnt that modelling groups that do not regrid their ocean output (both in horizontal and vertical coords) often leads to TBs of data contributed that sees little use. To avoid this problem altogether with a new wave quantity(ies), I would recommend that the most useable format was generated, and for this to match either the ocean (my preference) or the atmosphere grid where this data and its use would be most tightly connected.

Following this suggestion, it does raise the query whether a new realm is needed. My instinct here is to reduce complexity as much as we possibly can, whilst ensuring that by keeping things simple we don't complicate by breaking a logic linking variables, grids and realms together. A realm is just one way to organize output, and if this is not preserved in filenames and directory structures, the importance of a realm reduces somewhat - we would ensure that such information was contained in file metadata, the richer this resource is the better

[andrelanfer]

Hi Paul, You are right, it is a very good discussion. Since wave models does not share same grid with atm or ocn, the formats might be different as well. I personally prefer an ocean's format, but am pretty sure that others might think in a different way. In that way, to keep it easier I suggested a new realm, even this mean a litle bit more work and freedon degree. Best, Andre

Em qua., 25 de set. de 2024 19:21, Paul J. Durack @.***> escreveu:

@andrelanfer https://github.com/andrelanfer @taylor13 https://github.com/taylor13 this is indeed a good discussion, as we ideally want CMIP7 output to be AS USEABLE as possible. We've learnt that modelling groups that do not regrid their ocean output (both in horizontal and vertical coords) often leads to TBs of data contributed that sees little use. To avoid this problem altogether with a new wave quantity(ies), I would recommend that the most useable format was generated, and for this to match either the ocean (my preference) or the atmosphere grid where this data and its use would be most tightly connected.

Following this suggestion, it does raise the query whether a new realm is needed. My instinct here is to reduce complexity as much as we possibly can, whilst ensuring that by keeping things simple we don't complicate by breaking a logic linking variables, grids and realms together.

— Reply to this email directly, view it on GitHub https://github.com/WCRP-CMIP/CMIP7_CVs/issues/3#issuecomment-2375368947, or unsubscribe https://github.com/notifications/unsubscribe-auth/AK7F2JMKL5XNYUJIEA7VUI3ZYMZNDAVCNFSM6AAAAABO2CQ5H2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGNZVGM3DQOJUG4 . You are receiving this because you were mentioned.Message ID: @.***>