AGNI

Time-stepped radiative-convective solver designed for integration into a coupled atmosphere-interior code.

AGNI relies on SOCRATES (2306) for calculating the radiative-transfer. It makes use of the Julia interface to SOCRATES as written by Stuart Daines (see their branch here). SOCRATES is setup here to include shortwave irradiation from the star, Rayleigh scattering, and continuum absorption / CIA.

Surface boundary conditions are intended to be set by an interior model, so AGNI won't work as well for cooler planets. The model also includes a parameterised conductive 'skin' with a prescribed thickness and conductivity, allowing the surface temperature to be calculated according to the required conductive flux. Two convection parameterisations are included: convective adjustment (directly manipulating the temperature arrays), and mixing length theory (calculating convective energy fluxes). Results are optionally plotted (and animated), and may be saved as NetCDF or CSV files. Integration is primarily via a first-order Euler method with various acceleration options, however it is also possible to activate a high-order stiff integrator with automatic switching (Sundials CVODE Adams-Moulton).

Pronounced: ag-nee. Named after the fire deity of Hinduism.

Repository structure

README.md - This file
LICENSE.txt - License for use and re-use
doc/ - Other documentation
out/ - Output files
res/ - Resources
src/ - AGNI source code
socrates/ - Directory containing SOCRATES and associated files (subject to the license therein)
agni.jl - AGNI executable for debugging
agni_cli.jl - AGNI executable with command-line interface
demo_steamrun.jl- Script to demonstrate the pure-steam runaway greenhouse effect
demo_earth.jl - Script to demonstrate solving for Earth's temperature structure

Requirements

Julia (version 1.9.1 or later)
Python (version 3.10 or later)
NumPy and SciPy
gfortran
netcdf-fortran
make
OpenMP

Supported platforms

MacOS (ARM and x86-64)
Ubuntu (x86-64)

Installation instructions

$ cd socrates
$ cp ../res/Mk_cmd_PLAT ./make/Mk_cmd where PLAT is your platform
$ ./build_code
$ source set_rad_env
$ cd julia
$ julia
julia> ]
(@v1.9) pkg> add OffsetArrays
(@v1.9) pkg> add Revise
(@v1.9) pkg> add PCHIPInterpolation
(@v1.9) pkg> add LaTeXStrings
(@v1.9) pkg> add Plots
(@v1.9) pkg> add NCDatasets
(@v1.9) pkg> add DataStructures
(@v1.9) pkg> add Glob
(@v1.9) pkg> add ArgParse
(@v1.9) pkg> add SciMLBase
(@v1.9) pkg> add SteadyStateDiffEq
(@v1.9) pkg> add Sundials
(@v1.9) pkg> activate .
Press backspace
julia> cd("src")
julia> include("generate_wrappers.jl")
julia> exit()
$ cd lib
$ make
$ cd ../../..
You should end up in the root directory of the repository.

Running the code

For the command line interface, run $ ./agni_cli.jl (pass --help for help).
To debug the program, run $ ./agni.jl in the root directory of the repository.
To demo the steam runaway greenhouse effect, run $ ./demo_steamrun.jl.

Example outputs

Pure steam runaway greenhouse.

Calculating fluxes with SOCRATES, without solving for RCE.

Solving for RCE with accelerated time-stepping. (Outdated plot).

An attempt at solving for Earth's temperature profile. The ~100 K accuracy here wouldn't be good enough for Earth system modelling, but it's probably good enough for magma ocean modelling.