ISIMIP Repository


Most ISIMIP datasets are published in the ISIMIP Repository located at data.isimip.org. The data is organized by simulation round, e.g. ISIMIP3a, and data products:

  • Input data - the data used to drive the impact models, exactly as described in the ISIMIP simulation protocols.
  • Output data - the climate-impacts simulation data provided by participating modelling groups, corresponding to the output variables described in the ISIMIP simulation protocol.
  • Secondary input data - data that have been derived from or are related to the input data, but not described in the simulation protocol (e.g. with a differnt spatial or temporal resolution).
  • Secondary output data - simulation data that goes beyond the simulation protocol or data that have been derived by processing the ISIMIP output data, but do not correspond to output variables described in the simulation protocol.

Submitted output data are only published after an embargo period and after explicit permission by the modelling group is given (see also: Workflow for contributions to ISIMIP).

Currently, we are in the beta stage of the development of the portal. While more features will be added to the data portal, we encourage you to use the portal for searching and downloading. Please don't hesitate to report bugs and pitfalls that need to be fixed. Your feedback is important for us!

Note: The ISIMIP data used to be publicly available via an ESGF service, but that was decommissioned in May, 2022.

Search the repository

Using the advanced search interface you can search the repository in several ways:

  • You can enter a search query, e.g. "ISIMIP2b rcp60soc dis", and click Search to filter the results using this string. Please note that this search is fuzzy, meaning that an input like "rcp61soc" will also find "rcp60soc". For short strings or strings that differ only little, this can lead to unexpected behavior: a search for "ISIMIP2a" will also find "ISIMIP2b".
  • You can restrict the search using the tree in the sidebar. By clicking on the check boxes you can select a more and more specific set of results, much like in a file system explorer.
  • You can switch the sidebar to facets and select exact specifiers for the different identifiers, e.g. "Simulation round: ISIMIP2a" or "Sector: Water (global)".

Please note that these restrictions are combined to provide you the datasets you are looking for. In the box Search constraints all current constraints are visible and can be deselected using the red cross. Furthermore, a click on reset will reset the whole interface.

If you don't find the data you're looking for, please don't hesitate to get in touch.

Search results

In the ISIMIP repository, files for the same variable, but different time steps, are combined to datasets, e.g.

  • gswp3-w5e5_counterclim_tas_global_daily_1901_1910.nc
  • gswp3-w5e5_counterclim_tas_global_daily_1911_1920.nc
  • ...

are combined to the dataset: "gswp3-w5e5_counterclim_tas_global_daily". A search as described above, will result in a display of a number of these kind of datasets. If more that 12 datasets are found, the list can be navigated using the pagination in the top right of the page.

For each dataset, the stored attributes (e.g. the file size or the specifiers from the protocol) and the different files of the dataset can be viewed. Each file of the dataset can be downloaded individually or all files can be downloaded at once (this might take a long time).

A click on the title of the dataset or the title of a file will open a separate page for this dataset or file. These landing pages have a unique URL which identifies the dataset or file and which we will keep available even if we should de-publish the actual file in the future.

Using the Configure download link opens the mask service for this dataset (more information below).

Finally, a file list can be downloaded. This plain text file can be used for a bulk download with tools like wget (see below).

Configure download

In order to minimize the amount of data you need to download, the repository includes a masking service which can be used to remove all values of a file except a region (a country, a bounding box or just land data). This operation can also be performed on all files of a dataset at once. A link to this interface is included in the search result and on every dateset/file landing page.

Please note that the masking of NetCDF files takes a considerable amount of time. Depending on the size of the dataset, it can take tens of minutes to create the download. It is only possible to mask global files.

Bulk data download using wget

In order to download a large amount of files, usage of the wget tool is advised. If not already installed on you machine, you should be able to install it using your distibution (on Linux) or using homebrew (macOS). For Windows 10 , we recommend to run the wget scripts by installing the Windows Subsystem for Linux.

Once installed, a downloaded file list can be used with wget using (the "-c" (continue) flag ensures that already downloaded file are not downloaded again and the "-i" (input file) tells wget to use the downloaded list of files):

wget -ci FILE_LIST.txt

If you allready know what files you want, you can use the path of the file directly, e.g.:

wget -c https://files.isimip.org/ISIMIP3b/InputData/climate/atmosphere_composition/co2/historical/co2_historical_annual_1850_2014.txt

You can also download a complete directory using "--mirror --no-parent", e.g.:

wget -c --mirror --no-parent https://files.isimip.org/ISIMIP3b/InputData/climate/atmosphere_composition/

Using the API

The API of the repository can be accessed at data.isimip.org/api/v1. The important endpoints are "datasets" and "files". Both accept similar options, therefor we will give examples for "datasets" only. A GET request to

returns a paginated list of datasets. The response is a JSON object with "count" (count of all datasets), "next" (link to the next page), "previous" (link to the previous page) and "results" (the list of the datasets).

Since the API is also used by the search interface described above, searches can be done in a similar way. A few examples:

The API can be used for scripted search and download. For Python, we created a small client library with some example Jupyter notebooks.