Session 2: Lessons learnt at CMIP5-6

This session will take place from 11:45 am -1 pm (Tuesday 28 February 2023). Use this topic for questions and to capture discussions and thoughts.

Session 2 Speakers:

Australia’s CMIP6 efforts (Rachel Law, CSIRO, 10+5 min)

The challenges of model development and participation in CMIP (Martin Dix - ACCESS-NRI, 10+5 min)

High resolution models (Christian Jakob, Monash University, 10+5 min)

Net-zero and climate adaptation (Will Howard, Climate Change Authority, Virtual, 10+5 min)

Downscaling and the NPCP (Michael Grose, CSIRO, 10+5 min)

Question for @RachelLaw :

I couldn’t see anything about support for the experiments after publishing. An example, there is a problem with the post-processing of a CM2 experiment that means the heat budget doesn’t close which has not been resolved, and it isn’t clear who is responsible for fixing this.

3 Likes

Thanks for the answer @RachelLaw. I definitely see the benefit of having CMIP data shared widely on the ESGF, but if the data is wrong, and known to be so, then I’d question the scientific utility in that case.

At the very least IMO it should be flagged in the metadata that there are issues with the heat budget diagnostics.

It sounds like we’re under-resourcing the on-going support and documentation/publishing of CMIP submissions. This is something that should be addressed in the CMIP7 planning.

2 Likes

Interesting questions from @Christian_Jakob

Science questions (Marotzke et al 2017):

  • How is the weather going to change? [Global models at high resolution]
  • Where does the carbon go? [ESMs]
  • How does the climate change affect liveability? [City scale models]

User Needs (providing the best information possible as we will never have all that users want)

  • Planning for Adaptation
  • Planning for mitigation implementation

Technological capabilities

  • Computer says no

Pandora Hope asked:

My Q to Christian and the community - do we feel have responsibility to run these high-res models for the Global South/tropics, rather than just Australia?

2 Likes

@Christian_Jakob If the very best climate science is required to meet our existential challenge, and we accept that open source code is critical for excellent scientific modelling (as ECMWF recently seemed to suggest), is it ethical for the MetOffice to keep the UM on a closed licence?

1 Like

How does the Climate Change Authority interface with other areas of Government? Is there a unified Government approach and position that is being coordinated?

1 Like

Yes we do, but we do not have the capacity to do so. We probably never will on our own, so we will need to build strong international efforts in this space to drive the science forward and make it service ready as soon a possible. It’s a tall order!

1 Like

Good question, but it’s not (just) about open source code. It’s about resourcing modelling at a level where it can be carried out professionally and with the latest science integrated regularly. The MetOffice (closed source) does this much better than many open source modelling centres that contribute to CMIP. Given the trillion dollar decisions made based on the model results, we ought to be able to fund their development and application in a more sustained way. My view is that in the medium to long term climate modelling must be made operational, just like NWP.

1 Like

Thanks for the talks so far today, got to dash to another meeting now. Back online at 2pm. (PS:
Loving the hive @Aidan ! )

1 Like

From Michael Grose:
What do projections want from CMIP?​
Data required for downstream analysis and models – e.g., variables to do TC tracking​
Models that evaluate well – current climate, past trends, paleoclimate – especially for Australia, processes like ENSO ​
A balanced ensemble – avoid the ‘hot model’ problem, maximise independence ​
Minimum number of models, largest ensemble size from each model, highest resolution practical​
Exploration of the probability space, including tails​
Useful experiments, plausible scenarios – including more on carbon​
Manageable data and code workflows​
The ability to use it in conjunction with other tools to generate reliable knowledge​