JAX-RRTMGP: JAX-based RRTMGP Radiative Transfer

July 21, 2025 ยท View on GitHub

JAX-RRTMGP is a JAX-based implementation of the RRTMGP (Rapid Radiative Transfer Model for General circulation models - Parallel) radiative transfer scheme. This package provides fast, differentiable radiative transfer calculations for atmospheric modeling applications.

RRTMGP is a correlated k-distribution model for computing optical depths, source functions, and fluxes for longwave and shortwave radiation in planetary atmospheres. This JAX implementation enables automatic differentiation and efficient execution on GPUs and TPUs.

Features

  • JAX-native implementation: Full compatibility with JAX transformations (jit, grad, vmap, pmap)
  • GPU/TPU acceleration: Efficient execution on modern accelerators
  • Automatic differentiation: Enable gradient-based optimization and sensitivity analysis [IN PROGRESS]
  • Longwave and shortwave radiation: Complete radiative transfer calculations
  • Gas and cloud optics: Support for molecular absorption and cloud scattering/absorption
  • RRTMGP data files included: Pre-computed optical property lookup tables
  • Comprehensive testing: Extensive test suite with reference data

Installation

It is recommended to install in a virtual environment.

git clone https://github.com/climate-analytics-lab/jax-rrtmgp.git
python3 -m pip install -e jax-rrtmgp

Basic Usage

import jax
import jax.numpy as jnp
from rrtmgp import rrtmgp

# Set up atmospheric state
temperature = jnp.array([...])  # Temperature profile [K]
pressure = jnp.array([...])     # Pressure profile [Pa] 
vmr_h2o = jnp.array([...])      # Water vapor volume mixing ratio

# Initialize RRTMGP
config = rrtmgp.get_default_config()
rrtmgp_state = rrtmgp.initialize(config, pressure, temperature, vmr_h2o)

# Compute radiative fluxes
fluxes_lw = rrtmgp.compute_longwave_fluxes(rrtmgp_state, ...)
fluxes_sw = rrtmgp.compute_shortwave_fluxes(rrtmgp_state, ...)

Package Structure

  • rrtmgp: Main RRTMGP interface
  • rrtmgp.optics: Gas and cloud optical property calculations
  • rrtmgp.rte: Radiative transfer equation solvers
  • rrtmgp.optics.rrtmgp_data: Pre-computed lookup tables

Dependencies

  • JAX: Automatic differentiation and JIT compilation
  • NumPy: Numerical computing
  • NetCDF4: Reading RRTMGP data files

Testing

Run the test suite with:

pytest rrtmgp/

Or run individual test modules:

python rrtmgp/optics/gas_optics_test.py
python rrtmgp/rte/two_stream_test.py

License

Licensed under the Apache License, Version 2.0. This implementation is based on the JAX port of the original RRTMGP Fortran code by Eli Mlawer and Robert Pincus, originally included in swirl-jatmos.

Citation

If you use this software, please cite:

  1. The original RRTMGP paper: Pincus, R., Mlawer, E. J., and Delamere, J. S.: Balancing accuracy, efficiency, and flexibility in radiation calculations for dynamical models, J. Adv. Model. Earth Syst., 11, 3074-3089, 2019.

  2. This JAX implementation: [Add citation when published]