This page describes the setup details and gives an overview of Isca, a framework of idealized GCMs of increasing complexity.

Status (caa 24 Oct 2019): Installed and successfully compiled.  Held-Suarez model has been run, but not tested.  Aquaplanet run has been run, results look okay.

Overview

Isca (see https://execlim.github.io/IscaWebsite/) has been installed and tested.  It offers many different options, such as:

• Held-Suarez Dry Dynamical Core
• Radiative Schemes:
  • Grey atmosphere w/ optical depth (Frierson et al., 2006)
  • Grey atmosphere w/ moisture feedback (Bryne and O’Gorman, 2013)
  • 2+1-band radiation (Geen et al., 2016)
  • Rapid radiative transfer model (Mlawer et al., 1997)
  • SOCRATES (Manners et al., 2017) - not available yet on Odyssey
• Convective Schemes:

  • no convective scheme

  • Simplified Betts-Miller convection (Frierson et al., 2007)

  • Full Betts-Miller convection (Betts and Miller, 1986)

  • Relaxed Arakawa-Schubert (Moorthi and Suarez, 1992)

• Land Options:
  • Aquaplanet
  • Idealised / Realistic Continents (not available for Held-Suarez option)

Code Locations

• Source: /n/holylfs/INTERNAL_REPOS/CLIMATE_MODELS/isca
• Compiled Code: /n/scratchlfs/kuang_lab/isca

Setup Instructions

In order to get the code working, you need to install Isca into your Anaconda environment in Odyssey/Cannon

1. cd ./src/extra/python
2. pip install -r requirements.txt
3. pip install -e .

In your ~/.bashrc file, the following needs to be set (I have modified this for use on Odyssey/Cannon):

export GFDL_BASE=/n/holylfs/INTERNAL_REPOS/CLIMATE_MODELS/isca
export GFDL_ENV=fasrc
export GFDL_WORK=(WHEREVER YOU PUT THE EXECUTABLE) (Note: I recommend somewhere on /n/holyscratch01/)
export GFDL_DATA=(WHEREVER YOU WANT TO OUTPUT TO)
#export GFDL_SOCRATES="" (I haven't set this up yet)

(NB: I strongly recommend that you set your experiments up in a separate folder.  Since (if you have progressed this far) you have already installed Isca into the Python module, you can literally run Isca from anywhere since you've exported the environment in the .bashrc file.  So for example, I have the experiments set up in my home directory ~/Isca/isca_exp, with all the relevant input files and .nml files inside this directory, and submit the batch job in this directory, then the server will run the compiled code in the temp directory in GFDL_WORK.)

(NB2: Multiple people can likely run the compiled code simultaneously at any given point in time.  The caveat is that if the code is modified, any people who's jobs are currently running and are dependent on that compiled code will likely break.  Please contact me (nathanaelwong@fas.harvard.edu) if you want to modify the code, and we can discuss further.)