Python package for process-oriented climate modeling¶
climlab is a flexible engine for process-oriented climate modeling.
It is based on a very general concept of a model as a collection of individual,
climlab defines a base class called
can contain an arbitrarily complex tree of sub-processes (each also some
Process). Every climate process (radiative, dynamical,
physical, turbulent, convective, chemical, etc.) can be simulated as a stand-alone
process model given appropriate input, or as a sub-process of a more complex model.
New classes of model can easily be defined and run interactively by putting together an
appropriate collection of sub-processes.
climlab has out-of-the-box support and documented examples for
- Radiative and radiative-convective column models, with various radiation schemes:
RRTMG (a widely used radiative transfer code)
CAM3 (from the NCAR GCM)
Simplified band-averaged models (4 bands each in longwave and shortwave)
- Convection schemes:
Emanuel moist convection scheme
Hard convective adjustment (to constant lapse rate or to moist adiabat)
1D Advection-Diffusion solvers
Moist and dry Energy Balance Models
Flexible insolation including: - Seasonal and annual-mean models - Arbitrary orbital parameters
Boundary layer scheme including sensible and latent heat fluxes
- Arbitrary combinations of the above, for example:
2D latitude-pressure models with radiation, horizontally-varying meridional diffusion, and fixed relative humidity
Installing pre-built binaries with conda (Mac OSX, Linux, and Windows)¶
You can install
climlab and all its dependencies with:
conda install -c conda-forge climlab
Or (recommended) add
conda-forge to your conda channels with:
conda config --add channels conda-forge
and then simply do:
conda install climlab
Binaries are available for OSX, Linux, and Windows.
Installing from source¶
If you do not use conda, you can install
climlab from source with:
pip install climlab
(which will download the latest stable release from the pypi repository and trigger the build process.)
Alternatively, clone the source code repository with:
git clone https://github.com/brian-rose/climlab.git
and, from the
climlab directory, do:
python -m pip install . --no-deps -vv
You will need a Fortran compiler on your system. The build has been tested with both gcc/gfortran and ifort (Linux)
Installing from source without a Fortran compiler¶
Many parts of
climlab are written in pure Python and should work on any system.
Fortran builds are necessary for the RRTMG and CAM3 radiation schemes and for the Emanuel convection scheme.
If you follow the instructions for installing from source (above) without a valid Fortran compiler,
you should find that you can still:
and use most of the package. You will see warning messages about the missing components.
These are handled automatically if you install with conda.
Python 2.7, 3.6, 3.7
Recommended for full functionality¶
numba >=0.43.1 (used for acceleration of some components)
Note that there is a bug in previous numba versions that caused a hanging condition in climlab under Python 3.
Complete development environment¶
To build from source and develop new code you will need some additional pieces
gfortran (OSX or linux) or flang (Windows)
pytest (to run the automated tests, important if you are developing new code)
Anaconda Python is highly recommended and will provide everything you need. See “Installing pre-built binaries with conda” above.
Documentation and Examples¶
Full user manual is available here.
climlab/courseware/ also contains a collection of Jupyter notebooks (
*.ipynb) used for teaching some basics of climate science, and documenting use of the
These are self-describing, and should all run out-of-the-box once the package is installed, e.g:
jupyter notebook Insolation.ipynb
- Version 0.7.5 (released July 2019)
Bug fixes and improvements to continuous integration
- Version 0.7.4 (released June 2019)
New flexible solver for 1D advection-diffusion processes on non-uniform grids, along with some bug fixes.
- Version 0.7.3 (released April 2019)
Bug fix and changes to continuous integration for Python 2.7 compatibility
- Version 0.7.2 (released April 2019)
Improvements to surface flux processes, a new data management strategy, and improved documentation.
climlab.surface.SensibleHeatFluxare now documented, more consistent with the climlab API, and have new optional
resistanceparameters to reduce the fluxes (e.g. for modeling stomatal resistance)
climlab.surface.LatentHeatFluxnow produces the diagnostic
precipitationin the same units.
PRECIPdiagnostic (mm/day) in
climlab.convection.EmanuelConvectionis removed. This is a BREAKING CHANGE.
Data files have been removed from the climlab source repository. All data is now accessible remotely. climlab will attempt to download and cache data files upon first use.
climlab.convection.ConvectiveAdjustementis now accelerated with
numbaif it is available (optional)
- Version 0.7.1 (released January 2019)
Deeper xarray integration, include one breaking change to
climlab.solar.orbital.OrbitalTable, Python 3.7 compatibility, and minor enhancements.
climlab.utils.attr_dict.AttrDictand replaced with AttrDict package (a new dependency)
xarrayinput and output capabilities for
xarray.Datasetobjects containing the orbital data.
lookup_parameter()method was removed in favor of using built-in xarray interpolation.
climlab.process.ExternalForcing()for arbitrary externally defined tendencies for state variables.
New input option
ozone_file=Nonefor radiation components, sets ozone to zero.
Tested on Python 3.7. Builds will be available through conda-forge.
- Version 0.7.0 (released July 2018)
New functionality, improved documentation, and a few breaking changes to the API.
Major new functionality includes convective adjustment to the moist adiabat and moist EBMs with diffusion on moist static energy gradients.
climlab.convection.ConvectiveAdjustementnow allows non-constant critical lapse rates, stored in input parameter
New switches to implement automatic adjustment to dry and moist adiabats (pseudoadiabat)
climlab.EBM()and its daughter classes are significantly reorganized to better respect CLIMLAB principles:
Essentially all the computations are done by subprocesses
SW radiation is now handled by
Diffusion and its diagnostics now handled by
Diffusivity can be altered at any time by the user, e.g. during timestepping
Diffusivity input value
climlab.dynamics.MeridionalDiffusionis now specified in physical units of m2/s instead of (1/s). This is consistent with its parent class
A new class
climlab.dynamics.MeridionalMoistDiffusionfor the moist EBM (diffusion down moist static energy gradient)
Tests that require compiled code are now marked with
pytest.mark.compiledfor easy exclusion during local development
Under-the-hood changes include
Internal changes to the timestepping; the
compute()method of every subprocess is now called explicitly.
compute()now always returns tendency dictionaries
- Version 0.6.5 (released April 2018)
Some improved documentation, associated with publication of a meta-description paper in JOSS.
- Version 0.6.4 (released February 2018)
Some bug fixes and a new
climlab.couple()method to simplify creating complete models from components.
- Version 0.6.3 (released February 2018)
Under-the-hood improvements to the Fortran builds which enable successful builds on a wider variety of platforms (incluing Windows/Python3).
- Version 0.6.2 (released February 2018)
Introduces the Emanuel moist convection scheme, support for asynchonous coupling, and internal optimzations.
- Version 0.6.1 (released January 2018)
Provides basic integration with xarray (convenience methods for converting climlab objects into
- Version 0.6.0 (released December 2017)
Provides full Python 3 compatibility, updated documentation, and minor enhancements and bug fixes.
- Version 0.5.5 (released early April 2017)
Finally provides easy binary distrbution with conda
- Version 0.5.2 (released late March 2017)
Many under-the-hood improvements to the build procedure, which should make it much easier to get climlab installed on user machines. Binary distribution with conda is coming soon!
- Version 0.5 (released March 2017)
Bug fixes and full functionality for the RRTMG radiation module, an improved common API for all radiation modules, and better documentation.
- Version 0.4.2 (released January 2017)
Introduces the RRTMG radiation scheme, a much-improved build process for the Fortran extension, and numerous enhancements and simplifications to the API.
- Version 0.4 (released October 2016)
Includes comprehensive documentation, an automated test suite, support for latitude-longitude grids, and numerous small enhancements and bug fixes.
- Version 0.3 (released February 2016)
Includes many internal changes and some backwards-incompatible changes (hopefully simplifications) to the public API. It also includes the CAM3 radiation module.
- Version 0.2 (released January 2015)
The package and its API was completely redesigned around a truly object-oriented modeling framework in January 2015.
It was used extensively for a graduate-level climate modeling course in Spring 2015: http://www.atmos.albany.edu/facstaff/brose/classes/ATM623_Spring2015/
Many more examples are found in the online lecture notes for that course: http://nbviewer.jupyter.org/github/brian-rose/ClimateModeling_courseware/blob/master/index.ipynb
- Version 0.1
The first versions of the code and notebooks were originally developed in winter / spring 2014 in support of an undergraduate course at the University at Albany.
See the original course webpage at http://www.atmos.albany.edu/facstaff/brose/classes/ENV480_Spring2014/
The documentation was first created by Moritz Kreuzer (Potsdam Institut for Climate Impact Research) as part of a thesis project in Spring 2016.
Contact and Bug Reports¶
Users are strongly encouraged to submit bug reports and feature requests on github at https://github.com/brian-rose/climlab
This code is freely available under the MIT license. See the accompanying LICENSE file.
climlab is partially supported by the National Science Foundation under award AGS-1455071 to Brian Rose.
Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.