Do CMIP5 emergent constraints on the large scale circulation still work in CMIP6?

CMIP6 Hackathon CMIP6 Project Proposals
1

Do CMIP5 emergent constraints on the large scale circulation still work in CMIP6?

Scientific Motivation

In the CMIP5 archive a number of emergent constraints were proposed that linked future projected large scale circulation changes to present day climatology. If these emergent constraints are correct, then they can help us to constrain future climate projections. As advocated for by Hall et al (2019), doi.org/10.1038/s41558-019-0436-6 a primary indicator of a true emergent constraint is whether it survives an out of sample test. The CMIP6 archive provides us with a whole new set of simulations and models on which to determine whether the CMIP5 emergent constraints work to explain the distribution of in the CMIP6 models and vice-versa. An example of an emergent constraint on the large scale circulation, first proposed by Kidston and Gerber (2010) doi:10.1029/2010GL042873 and elaborated on by Simpson and Polvani (2016) doi:10.1002/2016GL067989 relates the magnitude of the poleward shift of the mid-latitude westerlies to the present day climatology during the winter (shown in the figure)

simpsonandpolvani
simpsonandpolvani.jpeg885Ă—633 38.7 KB

Other emergent constraints were proposed in CMIP5 for circulation change over the Pacific/North American sector with opposite conclusions. Simpson et al (2014) doi.org/10.1038/NCLIMATE2783 related the mangitude of meridional wind changes in western North America to the amplitude of climatological intermediate-scale stationary waves across the Pacific and concluded that many models are likely overestimating the amplitude of the meridional wind change over North America. This meridional wind change is connected to precipitation change over California and they concluded based on this emergent constraint that the real world is unlikely to become as wet in the winter as the CMIP5 ensemble mean suggests. In contrast Allen and Luptowitz DOI: 10.1038/ncomms16055 propose that the magnitude of circulation induced precipitation changes over California relate to a models fidelity in capturing ENSO teleconnections and that the models that have a better simulation of ENSO teleconnections exhibit more wetting over California, suggesting the precipitation increase over California may be higher than models suggest.

Proposed Hacking

During the hackathon, we will devise python notebooks to calculate each of the quantities used in the above emergent constraints for present day and future simulations. We will assess whether the emergent constraints determined from CMIP5 are still present in CMIP6 and will asses whether the constraint devised from the CMIP5 spread can be used to predict CMIP6 and vice-versa.

Anticipated Data Needs

Monthly ua, va, sst, precip for historical and future scenarios for both CMIP5 and CMIP6

Anticipated Software Tools

Not sure. Maybe someone with more python expertise can suggest. But we will want to do a fourier decomposition, quadratic fits and linear fits.

Desired Collaborators

People who have an interest in the large scale circulation and future hydroclimate change.

5 Likes
2

Thanks @islasimpson for a great project idea! A couple of questions

  • Your project involves CMIP5 data as well. You obviously know how to get it and work with it, but should we be planning to support less-experienced users in accessing CMIP5?
  • For any sort of Fourier analysis on xarray data, we developed the xrft python package; it might be helpful here
  • meta question about this forum: Did you experience a limitation in the number of links you were allowed to post? This might be a feature of discourse, designed to prevent spam. Can you try writing a reply post with all the links you might have wanted to include? I think the restriction only applies to your first post.
3

Thanks for the suggestions @rabernat.

  • For CMIP5. It seems that on cheyenne, the CMIP5 data request form is still available… https://www2.cisl.ucar.edu/resources/cmip-analysis-platform/request-cmip-data. So, maybe it’s not a big deal to get some CMIP5 data there. I don’t know about the cloud business. I guess that depends on what @naomi-henderson thinks about that. I already have what is needed for this project, so I figured it wouldn’t be a big deal if only the CMIP6 data were supported. I recall one project idea from the applications that also wanted to use CMIP5. In fact, I think they also wanted to use CMIP3.

  • Thanks for the Fourier analysis package info. I am completely new to python. Probably more new than most of our participants so I am quite clueless…I have been trying - it’s just hard to find the time.

  • I did experience a limitation in the number of links. It said new users are only allowed two links. But it also said that when I tried to create a reply post with all the links included.

4

@islasimpson, @rabernat

You obviously know how to get it and work with it, but should we be planning to support less-experienced users in accessing CMIP5?

For CMIP5 data hosted on GLADE, intake-esm might be a useful, convenient tool for less-experienced users, and advanced users as well:

import intake
col = intake.open_esm_metadatastore(collection_name="GLADE-CMIP5")

cat = col.search(variable=['hfls'], frequency='mon', modeling_realm='atmos',
                institute=['CCCma', 'CNRM-CERFACS'])

dsets = cat.to_xarray(decode_times=True, chunks={'time': 50})

I’m happy to help out with this one.

5

Intake-esm does work for CMIP6 data as well:

import intake
col = intake.open_esm_metadatastore(collection_name="GLADE-CMIP6")

cat = col.search(variable_id=['hfls'], table_id=['Amon'], 
                 experiment_id=['1pctCO2', 'histSST'],
                 source_id=['CanESM5', 'IPSL-CM6A-LR'])

dsets = cat.to_xarray(chunks={'time': 100})

More in-depth tutorial: https://intake-esm.readthedocs.io/en/latest/notebooks/tutorial.html

6

@islasimpson - I like this project. I am curious if you are specifically interested in constraints on the large scale circulation and hydroclimate or if you’d also be interested in other constraints (e.g., climate feedbacks)?

7

Hi @pochedley, Glad you’re interested! I’m definitely interested to consider other constraints too. The more the merrier! I just suggested those ones as I am most familiar with those. But it would be a good idea to consider climate feedbacks constraints too.

8

Is there an “Allocated Project Code” I can use for requesting CMIP5 data to be downloaded at https://www2.cisl.ucar.edu/resources/cmip-analysis-platform/request-cmip-data? I also don’t have a username yet so can’t request CMIP5 data. Any suggestions?

I could put together a script to scrape the monthly fields I think I’ll need for my project and then just run it once I get access on Cheyenne? Will I have access to a few 100 Gb of disk space?

9

Hi, I am interested in this project!
I am also keen on seeking emergent constraints to narrow uncertainties in future extreme precipitation. Constraints for cloud feedback are also in my interest. Looking forward to learning from you! :slight_smile:

1 Like
10

@hdrake1 I think if you use the google form that was sent out to the group to request the data then the data request on cheyenne will be taken care of without a need for the project code.

1 Like
11

@vitacd Great! Looking forward to working with you. Be sure to request any data you’d be interested in looking at through the provided google form.

1 Like
12

@islasimpson Thank you – this worked for me for CMIP6 but the CMIP5 form required an allocation code. It’s not a problem though, I think it will be more than sufficient to just work with CMIP6 data during the hackathon and bring the CMIP5 data in later / download necessary files while we’re there.

13

@hdrake1 It’s probably worthwhile trying to get the data before hand. If you send me along what you’d need, I may already have it or can fill in the CMIP5 form as I have a project code. Thanks!