Climate

Climate Model from your Laptop with MAGICC7 and Pymagicc

Until recently, I thought that all climate models were incredible complex and only run on massive supercomputers. However, through my Master’s, I became accustomed to the fact that many models, especially early models, are quite simple. This might be initially surprising, but when combined with the fact that the science behind climate change dates back over 100 years, it seems clear that simpler models could still be powerful.

MAGICC

One of the most common simplified models is called MAGICC. The acronym stands for “Model for the Assessment of Greenhouse Gas Induced Climate Change” and focuses on modeling the carbon cycle. It is commonly used by IPCC, other decision makers, and scientists when they need to quickly but accurately create climate projections.

Logo image used by MAGICC

More specific information about the model can be found on the website and wiki. As of this blog post, binaries of the current version, MAGICC7, are available for download via a form on the site. The previous version, MAGICC6, is also available but via the wiki. Once downloaded, you will need to extract the contents and ensure the binaries have executable permissions.

Pymagicc

However, being a avid python user, I immediately went to see if there was a port. Thankfully, I was greeted by pymagicc. The team at OpenSCM graciously wrote a python wrapper for the binary. However, there are a few caveats. The python package by default uses a Windows version of MAGICC6 as that was previously the only binary released by the MAGICC team. This means that on any other operating system, you are forced to use an emulator like wine. While wine is very powerful, I wanted to use the macOS-native MAGICC7 binary.

Once again, the OpenSCM team has gone above and beyond by already creating a tutorial notebook for MAGICC7. In short, we can provide the binary path directly to pymagicc and it will use that instead of the built-in version.

Basic Usage

I’m only a beginner with using this model, but here I will demonstrate how easy it is to use the model to make projections based on different emission scenarios RCPs. Let’s begin by installing the required pip packages.

# !pip install matplotlib scmdata pymagicc

Now we can import pymagicc, along with some data and plotting libraries.

import matplotlib.pyplot as plt

import pymagicc
import scmdata
from pymagicc import rcps

/Users/seangtkelley/miniconda3/envs/climate-model-pymagicc/lib/python3.10/site-packages/scmdata/database/_database.py:9: TqdmExperimentalWarning: Using `tqdm.autonotebook.tqdm` in notebook mode. Use `tqdm.tqdm` instead to force console mode (e.g. in jupyter console)
  import tqdm.autonotebook as tqdman

Next, we will point pymagicc to the MAGICC7 binary. We can do this using the MAGICC_EXECUTABLE_7 environment variable. Jupyter notebooks provide an easy way to set environment variables using the %env command.

%env MAGICC_EXECUTABLE_7=/Users/seangtkelley/Source/blog-posts/climate-model-magicc-python/magicc-v7.5.3/bin/magicc-darwin-arm64

env: MAGICC_EXECUTABLE_7=/Users/seangtkelley/Source/blog-posts/climate-model-magicc-python/magicc-v7.5.3/bin/magicc-darwin-arm64

If you encounter an error complaining about libgfortran, you likely also need to specify a path for the this library.

On MacOS, this will likely be be in your gcc install, but you can find its full path with the following command: find /path/to/search -name "libgfortran*.dylib"

%env DYLD_LIBRARY_PATH=/opt/homebrew/Cellar/gcc/14.2.0_1/lib/gcc/current:$DYLD_LIBRARY_PATH

env: DYLD_LIBRARY_PATH=/opt/homebrew/Cellar/gcc/14.2.0_1/lib/gcc/current:$DYLD_LIBRARY_PATH

Finally, we can run the model! We will group the RCPs by their specific scenario and run the model on each.

NOTE: I have set out_temperature=1 which is required to instruct MAGICC to output temperature data. This value (along with many other configuration values) can also be set in MAGCFG_USER.CFG in the run folder. More information can be found in the tutorial notebook and this issue.

rcps["scenario"].unique()

array(['RCP26', 'RCP45', 'RCP60', 'RCP85'], dtype=object)

results = []
for scen in rcps.groupby("scenario"):
    results_scen = pymagicc.run(scen, magicc_version=7, out_temperature=1)
    results.append(results_scen)

results = scmdata.run_append(results)

/Users/seangtkelley/miniconda3/envs/climate-model-pymagicc/lib/python3.10/site-packages/pymagicc/io/scen7.py:43: UserWarning: MAGICC6 RCP region naming (R5*) is not compatible with MAGICC7, automatically renaming to MAGICC7 compatible regions (R5.2*)
  warnings.warn(warn_msg)
/Users/seangtkelley/miniconda3/envs/climate-model-pymagicc/lib/python3.10/site-packages/pymagicc/io/scen7.py:43: UserWarning: MAGICC6 RCP region naming (R5*) is not compatible with MAGICC7, automatically renaming to MAGICC7 compatible regions (R5.2*)
  warnings.warn(warn_msg)
/Users/seangtkelley/miniconda3/envs/climate-model-pymagicc/lib/python3.10/site-packages/pymagicc/core.py:436: UserWarning: magicc logged a ERROR message. Check the 'stderr' key of the result's `metadata` attribute.
  warnings.warn(
/Users/seangtkelley/miniconda3/envs/climate-model-pymagicc/lib/python3.10/site-packages/pymagicc/io/scen7.py:43: UserWarning: MAGICC6 RCP region naming (R5*) is not compatible with MAGICC7, automatically renaming to MAGICC7 compatible regions (R5.2*)
  warnings.warn(warn_msg)
/Users/seangtkelley/miniconda3/envs/climate-model-pymagicc/lib/python3.10/site-packages/pymagicc/io/scen7.py:43: UserWarning: MAGICC6 RCP region naming (R5*) is not compatible with MAGICC7, automatically renaming to MAGICC7 compatible regions (R5.2*)
  warnings.warn(warn_msg)
/Users/seangtkelley/miniconda3/envs/climate-model-pymagicc/lib/python3.10/site-packages/pymagicc/core.py:436: UserWarning: magicc logged a ERROR message. Check the 'stderr' key of the result's `metadata` attribute.
  warnings.warn(
/Users/seangtkelley/miniconda3/envs/climate-model-pymagicc/lib/python3.10/site-packages/pymagicc/io/scen7.py:43: UserWarning: MAGICC6 RCP region naming (R5*) is not compatible with MAGICC7, automatically renaming to MAGICC7 compatible regions (R5.2*)
  warnings.warn(warn_msg)
/Users/seangtkelley/miniconda3/envs/climate-model-pymagicc/lib/python3.10/site-packages/pymagicc/io/scen7.py:43: UserWarning: MAGICC6 RCP region naming (R5*) is not compatible with MAGICC7, automatically renaming to MAGICC7 compatible regions (R5.2*)
  warnings.warn(warn_msg)
/Users/seangtkelley/miniconda3/envs/climate-model-pymagicc/lib/python3.10/site-packages/pymagicc/core.py:436: UserWarning: magicc logged a ERROR message. Check the 'stderr' key of the result's `metadata` attribute.
  warnings.warn(
/Users/seangtkelley/miniconda3/envs/climate-model-pymagicc/lib/python3.10/site-packages/pymagicc/io/scen7.py:43: UserWarning: MAGICC6 RCP region naming (R5*) is not compatible with MAGICC7, automatically renaming to MAGICC7 compatible regions (R5.2*)
  warnings.warn(warn_msg)
/Users/seangtkelley/miniconda3/envs/climate-model-pymagicc/lib/python3.10/site-packages/pymagicc/io/scen7.py:43: UserWarning: MAGICC6 RCP region naming (R5*) is not compatible with MAGICC7, automatically renaming to MAGICC7 compatible regions (R5.2*)
  warnings.warn(warn_msg)
/Users/seangtkelley/miniconda3/envs/climate-model-pymagicc/lib/python3.10/site-packages/pymagicc/core.py:436: UserWarning: magicc logged a ERROR message. Check the 'stderr' key of the result's `metadata` attribute.
  warnings.warn(

From what I understand, the warnings here are expected since the data available in the pymagicc library is designed for MAGICC6 so it needs to be converted.

Either way, with the results, we can now plot the global average surface temperature relative to the pre-industrial mean.

temperature_rel_to_1850_1900 = (
    results
    .filter(variable="Surface Temperature", region="World")
    .relative_to_ref_period_mean(year=range(1850, 1900 + 1))
)

temperature_rel_to_1850_1900.lineplot()
plt.title("Global Mean Temperature Projection")
plt.ylabel("°C over pre-industrial (1850-1900 mean)")
plt.show()

/Users/seangtkelley/miniconda3/envs/climate-model-pymagicc/lib/python3.10/site-packages/scmdata/run.py:905: FutureWarning: Dtype inference on a pandas object (Series, Index, ExtensionArray) is deprecated. The Index constructor will keep the original dtype in the future. Call `infer_objects` on the result to get the old behavior.
  df.columns = pd.Index(columns, name="time")
/Users/seangtkelley/miniconda3/envs/climate-model-pymagicc/lib/python3.10/site-packages/scmdata/run.py:905: FutureWarning: Dtype inference on a pandas object (Series, Index, ExtensionArray) is deprecated. The Index constructor will keep the original dtype in the future. Call `infer_objects` on the result to get the old behavior.
  df.columns = pd.Index(columns, name="time")
/Users/seangtkelley/miniconda3/envs/climate-model-pymagicc/lib/python3.10/site-packages/scmdata/run.py:905: FutureWarning: Dtype inference on a pandas object (Series, Index, ExtensionArray) is deprecated. The Index constructor will keep the original dtype in the future. Call `infer_objects` on the result to get the old behavior.
  df.columns = pd.Index(columns, name="time")
/Users/seangtkelley/miniconda3/envs/climate-model-pymagicc/lib/python3.10/site-packages/scmdata/plotting.py:81: FutureWarning: 

The `ci` parameter is deprecated. Use `errorbar='sd'` for the same effect.

  ax = sns.lineplot(data=plt_df, **kwargs)

png

Conclusion

That’s it! In just a few lines of code, we have run state of the art climate model to project global temperatures through the next 75 years. MAGICC is obviously much more powerful than this basic example, but I hope it’s now clear that climate models can be a lot more accessible than supercomputers and petabytes of data.

If you have climate model or python knowledge, please considering supporting pymagicc as it’s completely open source and a wonderful tool for those who need efficient but accurate climate modeling in the world’s most popular programming language.

By Sean Kelley
On April 25, 2025