10
$\begingroup$

I am interested in earth system models, land surface models and climate prediction models. I found a lot of online information on different models, including CIMP5, and the Earth System Modeling Framework. I also found a lot of articles using these models.

I would like to run and experiment with these models on my own computer. However, I could not find any clear installation guides or code examples on how to run these models. I would like to have a step by step instruction how to install the model, how to include the input data, how to run the model and how to view the output. There must be decent tutorials somewhere out there. I am particularly interested in models using Normalized Difference Vegetation Index (NDVI) and Leaf area index (LAI) data.

For the time being it should be something that can be run on a desktop computer or in a virtual machine, but I also have access to more powerful clusters.

Improve this question
$\endgroup$
5
  • 3
    $\begingroup$ The "professional" climate models are ofter difficult to set up when one did never do it before. The compilation itself is not the actual the problem for a user, who is experienced in working in the terminal on Linux workstations. It takes some time to compile dependencies like netCDF4. Getting input data and choosing parameterization for different processes is quite a lot of work and might be quite difficult. $\endgroup$
    – daniel.heydebreck
    Commented Mar 21, 2018 at 11:36
  • $\begingroup$ For the beginning it might be an option to first try a toy model like the PlanetSimulator. $\endgroup$
    – daniel.heydebreck
    Commented Mar 21, 2018 at 11:37
  • $\begingroup$ Are you working at an academic institution? In which country do you work? If you do not work at an academic institution it might be difficult to obtain the necessary input data. Often, each institution has its own model or at least its own model setup. It might be reasonable, to choose a model and than contact the respective institution directly. $\endgroup$
    – daniel.heydebreck
    Commented Mar 21, 2018 at 11:39
  • 2
    $\begingroup$ I'm not sure precisely for what you want to do, but in general, climate models are run on super-computers. A desktop might be to little computing power. $\endgroup$
    – userLTK
    Commented Mar 21, 2018 at 12:19
  • 1
    $\begingroup$ Yes I am working at an academic institution. I am aware that most models run on supercomputers, but I know that some can also be run on a regular desktop. Compiling on linux is no problem for me. I have done that before. What I am really looking for is a a good guide on how to install/compile a commonly used earth system model. $\endgroup$
    – mace
    Commented Mar 21, 2018 at 12:58

3 Answers 3

Reset to default
9
$\begingroup$

As alluded to in some of the comments, all of the CMIP5 models will have been run on supercomputers, and it takes a lot of effort to get one of them running on a new platform, even for a team who know both the model and platform well already. In addition to that, many of those models are closed-source and are not generally available to an individual researcher. However, there are several older models such as MITgcm and EdGCM that it might well be possible to get running on a cluster. It's something I've thought about but never quite got around to. One group have even been running CESM on Amazon's EC2 service.

Depending on what aspect of leaf phenology you're interested in, you may need to run quite long simulations (e.g., several centuries), which is prohibitive with a full ESM. You can make a lot of progress in this area by running just the land surface components. This is several orders of magnitude simpler than running an ESM: the source codes are often freely available where their host ESM code is not, and they can be run globally on desktop machines or clusters. Some example land surface models you might want to look into:

Each of these will require a set of forcing boundary conditions, and there are several commonly used by the land community:

If you go down the land surface route (and I suggest that's a good place to start), you should probably just pick a model and ask it's community directly about the details of getting it running. In my experience, several of them have very good tutorials especially for people with a bit of Linux experience.

Improve this answer
$\endgroup$
0
7
$\begingroup$

We are moving all of CESM to open source - in particular here is the latest version of the Community Terrestrial Systems Model (CTSM) https://github.com/ESCOMP/ctsm

Formerly known as CLM, CTSM includes capability to run on linux or MacOS systems.

Improve this answer
$\endgroup$
1
  • $\begingroup$ Thanks Jim. Great to see the documentation in there! $\endgroup$
    – mace
    Commented Mar 24, 2018 at 10:17
3
$\begingroup$

You can take a look at the Predictive Ecosystem Analyzer (PEcAn) - it is a statistical framework and workflow system that supports parameterization and analysis of a number of land surface, crop, and ecosystem models. It supports a half-dozen models and support for another half-dozen or so is planned.

The mission of the project is to "Develop and promote accessible tools for reproducible ecosystem modeling and forecasting" You can learn more, find a demo, tutorials, documentation, and a virtual machine at the project home page: pecanproject.org.

You can also find us on github.com/pecanproject and meet the community on Slack.

disclaimer: I am a PI and developer on the project.

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.