O0.6 (coupler) Hierarchies paper

[LenkaNovak]

The Climate Modeling Alliance

Software Design Issue 📜

Purpose

This showcases the uniqueness of our framework to bridge gaps in classical hierarchical modeling (e.g. Jeevanjee et al., 2017. Although clouds in idealized GCMs now do exist, this framework allows ease of use and practical computational resources.

While we could design several types of hierarchies, this issue focuses on the elegant hierarchies for process investigation of global large scale circulations. The separation of hierarchies into the mid-latitude and tropical branches is beyond the scope.

In a separate issue / paper we can focus on the hierarchy useful for model development of a coupled system (SCM, dry LES, moist LES, (dry GCM), moist GCM).

Cost/Benefits/Risks

b: demonstrate the utility of the modularity of ClimaESM to the atmospheric community, increasing potential user base

People and Personnel

Components

• Cases presented:
  • dry HS - parametrized convection and radiation
  • moist HS - parametrized moist convection and radiation
  • Frierson (B-M scheme) - simple convection no clouds and gray radiation (perhaps not necessary?)
  • aquaplanet with diagnostic EDMF and allsky radiation - clouds and coupled 1M/0M
  • convection resolving AGCM
• case study: seasonal cycle presented in all models

Results and Deliverables

• JAMES paper
• set of demonstrable experiments that showcase the ease of use of the AGCM hierarchy

Task Breakdown And Schedule

Minimal working case

Beta Give feedback

1. compile a list of showcased simulations
2. draft paper structure + intro - Overleaf
3. scope out the most performant setups Use 4xA100 to achieve > 1SYPD for target EDMF AMIP #740
4. adapt AMIP driver to accommodate for simplest setup (i.e. dry HS)
5. Generalize atmos_init to dry and no radiation cases #754
   +0

Diagnostics

Beta Give feedback

1. climate (slices, zonal means, anims): u, T, mass\_psi, (qt)
2. storm tracks diagnostics: vt and egr timeseries + composites
3. vT-egr phase space plot
4. add additional diagnostics to CAD extensions #756
   🍃 leaf +1
   
5. add hierarchy data visualizations #814
   +0

Implement / run the remaining simulations with comparable results to the benchmark papers:

Simulations suite (all config/toml/parsed args/driver setups reduced to one run script) -

Beta Give feedback

1. Hierarchies: Dry Held-Suarez driver #759
   🍃 leaf +1
   
2. Hierarchies: Moist Held-Suarez driver #760
   🍃 leaf +1
   
3. Hierarchies: Cloud-less aquaplanet driver #761
   🍃 leaf +1
   
4. Hierarchies: Cloudy aquaplanet driver #762
   🍃 leaf +1
   
5. Hierarchies: Cloudy slabplanet (AMIP minus data) #812
   🍃 leaf +1
   

paper

Beta Give feedback

1. initial intro and discussion
2. all cases described for generality-based hierarchy
3. all cases described for process-based hierarchy
4. convert to JAMES format
5. finalize and clean up (references, check eqns, acknewledgements, etc)

What happens next

• 1. all drivers are prepared but come of the configs need to be fine tuned (see this issue). Hopefully increasing the resolution will help with most of the stability issues.
• 2. plotting scripts for clean plots of the climatology are ready to be used in experiments/ClimaEarth/hierarchy/climate_plots.jl
• 3. Contingent on 1., if we get long enough time series of large scale heat fluxes (vT) and Eady growth rate (egr), we can include phase plots in the paper. These can be generated using [this script]
• 4. After 1.-3. we can clean up the rest of the paper as specified in the last task list above.

Timelines

• draft paper body + implement all code infrastructure by May 24
• stabilize experiments by July 30
• submission by August 10

SDI Revision Log

CC