README.md

seager19

Model run results: https://wandb.ai/sdat2/seager19

Docker image for gfortran/gcc/cdf/conda: https://hub.docker.com/repository/docker/sdat2/seager19

Purpose

A repository to contain, analyze, and expand upon the model from:

Seager et al. 2019, Nature Climate Change, Strengthening Tropical Pacific Zonal Sea Surface Temperature Gradient Consistent with Rising Greenhouse Gases

https://doi.org/10.1038/s41558-019-0505-x

Summary of their paper

• There is a west-east warm-to-cold contrast in the Pacific.

• State-of-the-art GCMs predict weakening contrast with GHG concentration.

• In observations, it has increased.

• Their simple linear equatorial coupled model agrees with observations when forced with ECMWF reanalysis product.

• When forced with CMIP5 ensemble relative humidity and windspeed their model can reproduce the bias, highlighting these fields as the mechanism through which it is created.

Citation

The citation for their paper is:

@article{seager2019strengthening,
  title={Strengthening tropical Pacific zonal sea surface temperature gradient consistent with rising greenhouse gases},
  author={Seager, Richard and Cane, Mark and Henderson, Naomi and Lee, Dong-Eun and Abernathey, Ryan and Zhang, Honghai},
  journal={Nature Climate Change},
  volume={9},
  number={7},
  pages={517--522},
  year={2019},
  url={https://doi.org/10.1038/s41558-019-0505-x},
  publisher={Nature Publishing Group}
}

Podcast

The paper is discussed in a podcast available at Deep Convection Season 1 Episode 5:

Specifically the time period 33:00-44:30.

Code

The code and data were taken from a Columbia University website:

This website is blocked from some IP addresses, such as the Cambridge VPN (probably my fault), and so you may need to try a different connection method if it doesn't load.

Setup

Get the Github repository

    git clone https://github.com/sdat2/seager19.git

    cd seager19

If you have root access and can install Linux packages

    # Enforcing these versions is necessary to be able to use 70s fortran.
    # If you would rather not change your version, you might be better off
    # Using the docker image instead.

    sudo apt-get install gfortran-4.8

    sudo apt-get install gcc-4.8

    sudo apt-get install --fix-missing libnetcdf-dev libnetcdff-dev

    sudo apt-get install cloc  # brew install cloc for mac

If you want to make the docker environment yourself

    docker build . -t sdat2/seager19:g4.8

    docker push sdat2/seager19:g4.8

If you need to install the singularity environment

    
    # on jasmin you would need to make a new tmp directory in your home directory
    # to be able to install the quite large singularity environment e.g:

    mkdir /home/users/sithom/tmp

    # and then call the command to create the singularity object

    TMPDIR=/home/users/sithom/tmp SINGULARITY_CACHEDIR=/home/users/sithom/tmp singularity pull docker://sdat2/seager19:g4.8

    # and then to launch the envirnonment

    singularity run seager19_g4.8.sif

Making the environment and testing it works (either in singularity or not)

    # in case conda isn't activated:

    conda init bash

    source ~/.bash_profile

    # activate the environment

    conda activate ./env/

    # test that the environment and package work before using:

    make test        # Also downloads the data if needed.

Add optional features

    make jupyter_pro  # adds timing etc.

    make vscode_pro  # sets up nice autodoc, pylint etc.

    make jupyter_dark  # dark mode for jupyter notebooks

    # and to reverse this:

    make jupyter_light

Examples of running the model

    # default run:

    python src/main.py name=cd_2.25 coup.c_d=2.25e-3

    # Sweep through different levels of Raleigh friction

    # values near the value in the paper
    python src/main.py atm.eps_days=1.05,0.95,0.85,0.75,0.65,0.55 -m

    python src/main.py atm.eps_days=1.7,1.8,1.85 -m 

    python src/main.py atm.eps_days=0.7,0.8 -m 


    # values from other papers quoted in the introduction to Romps (2014) "Raleigh Damping in the Free Troposphere"
    python src/main.py atm.eps_days=1.25,1.8,2,2.5,3,5,10 -m

    python src/main.py atm.eps_days=2.1,1.75,1.9 -m

    # Seager91, Matsuno66, Yu97, Gill80, Chang82, Sugiyama09, Wu00
    # breaks at 5, 10: Sugiyama09, Wu00. Breaks through tau.
    # looks pretty strange at 3: Chang82.

    python src/main.py atm.k_days=7,8,9,10 -m

    # uncoupled run without syncing:
    python src/main.py name=it_1 coup.iterations=1 coup.c_d=2.25e-3 wandb=false

To look at the commands to relicate the paper figures, see replicate.md.

Other handy commands for development of repo

Make a notebook with helpful magic functions for dark mode

    make notebook name=your-notebook-name

New python script

    make py name=src/to-new.py

Update docs of the src directory

    make autodoc

Check where a variable is referenced in the ocean model

   grep -R "f1prime" ocean/SRC

Get CMIP5 multimodel means from Columbia

http://kage.ldeo.columbia.edu:81/SOURCES/.LDEO/.ClimateGroup/.PROJECTS/.IPCC/.CMIP5/.MultiModelMeans/.MMM-v2.3/.historical/.Surface/data.cdf

http://kage.ldeo.columbia.edu:81/SOURCES/.LDEO/.ClimateGroup/.PROJECTS/.IPCC/.CMIP5/.MultiModelMeans/.MMM-v2.3/.rcp85/.Surface/data.cdf

CMIP6/CMIP5 bias mechanism data

https://docs.google.com/spreadsheets/d/1QrCLil7uHMRJECOoSL18uk2mvBwMLxqesXbZpUWh3ko/edit?usp=sharing