Joint evaluation

Joint evaluation of the CMIP6 downscaling within EURO-CORDEX.

This repository is supposed to help in organizing the EURO-CORDEX joint evaluation, e.g., it holds meta data like tables, and catalogs, but also README files and issues regarding evaluation. If you have any topics, questions or suggestions, please don't hesitate to open a new issue! There is also the possibility to join a team to stay in touch easily if you want.

For a broader overview, you can also have a look at the project board.

There is also a repository (evaltools) that might hold useful code used in the evaluation analysis. This is separated for better maintainability.

xkcd comic used under CC BY-NC 2.5 license

Data exchange

Some storage space on the existing jsc-cordex data exchange infrastructure at Jülich Supercomputing Centre (JSC) is available and may be used to provide some intermediate, temporary, limited storage for the first joint analyses publications of CORDEX-CMIP6 simulations before data will be stored longer-term on ESGF-related storage at the respective centres.

Access to storate at JSC is decribed here (private repo until this is released to the community).

The variables requested for this joint evaluation, along their corresponding metadata, are available in this repository, under the file dreq_EUR_joint_evaluation.csv. This file can be explored using the search box in Github; e.g. search for "Overview" to filter the variables currently requested for the overview evaluation work. This file can also be downloaded here, to use it in your data uploading scripts.

import pandas as pd

pd.read_csv("https://raw.githubusercontent.com/euro-cordex/joint-evaluation/refs/heads/main/dreq_EUR_joint_evaluation.csv")

   out_name frequency  units                       long_name        standard_name            cell_methods   priority comment
0       clt       mon      %    Total Cloud Cover Percentage  cloud_area_fraction        area: time: mean     Trends     NaN
1      hurs       1hr      %  Near-Surface Relative Humidity    relative_humidity  area: mean time: point        FWI     NaN
2      hurs       day      %  Near-Surface Relative Humidity    relative_humidity        area: time: mean        FWI     NaN
3   hus1000       6hr      1               Specific Humidity    specific_humidity  area: mean time: point  AtmRivers     NaN
4    hus200       6hr      1               Specific Humidity    specific_humidity  area: mean time: point  AtmRivers     NaN
..      ...       ...    ...                             ...                  ...                     ...        ...     ...
63    va700       6hr  m s-1                  Northward Wind       northward_wind  area: mean time: point  AtmRivers     NaN
64    va850       6hr  m s-1                  Northward Wind       northward_wind  area: mean time: point  AtmRivers     NaN
65    va925       6hr  m s-1                  Northward Wind       northward_wind  area: mean time: point  AtmRivers     NaN
66      vas       mon  m s-1     Northward Near-Surface Wind       northward_wind        area: time: mean     Trends     NaN
67    zg500       mon      m             Geopotential Height  geopotential_height        area: time: mean     Trends     NaN

Catalog

A catalog of available data at JSC-CORDEX is available from this repository. For example,

import intake

cat = intake.open_esm_datastore("https://raw.githubusercontent.com/euro-cordex/joint-evaluation/refs/heads/main/CORDEX-CMIP6.json")
cat.keys()

gives

['CORDEX.EUR-12.GERICS.ERA5.evaluation.r1i1p1f1.REMO2020.v1.mon.hurs.v20240529',
 'CORDEX.EUR-12.GERICS.ERA5.evaluation.r1i1p1f1.REMO2020.v1.mon.pr.v20240529',
 'CORDEX.EUR-12.GERICS.ERA5.evaluation.r1i1p1f1.REMO2020.v1.mon.prsn.v20240529',
 'CORDEX.EUR-12.GERICS.ERA5.evaluation.r1i1p1f1.REMO2020.v1.mon.ps.v20240529',
 'CORDEX.EUR-12.GERICS.ERA5.evaluation.r1i1p1f1.REMO2020.v1.mon.tas.v20240529',
 'CORDEX.EUR-12.GERICS.ERA5.evaluation.r1i1p1f1.REMO2020.v1.mon.tasmax.v20240529',
 'CORDEX.EUR-12.GERICS.ERA5.evaluation.r1i1p1f1.REMO2020.v1.mon.tasmin.v20240529',
 'CORDEX.EUR-12.GERICS.ERA5.evaluation.r1i1p1f1.REMO2020.v1.mon.uas.v20240529',
 'CORDEX.EUR-12.GERICS.ERA5.evaluation.r1i1p1f1.REMO2020.v1.mon.vas.v20240529']

You can filter and load datasets using, e.g.,:

dataset_dict = cat.search(variable_id=["tas", "orog", "sftlf"]).to_dataset_dict()

and you will get a dictionary of datasets back. Note that this, for now, will only work when you are logged in to jsc-cordex since datasets reside on the local filesystem. For more info about intake-esm, please also see the documentation.

Starting Jupyter Lab

If you want to work interactively on jsc-cordex, you can use jupyterlab via ssh (right now, jsc-cordex is not available from Jupyter-JSC).

Note

You need to activate a virtual environemt, e.g., using conda activate base, or any other environment in which you have installed your requirements including conda install juypterlab. The current base environment only contains some basic requirements, usually, you might want to setup a dedicated environment for yourself. Please consult the README.

Once you are logged in to jsc-cordex and set up, you can start the jupyter server (without a browser) like this:

jupyter lab --no-browser

Note the port in the URL, e.g. http://localhost:8888 (the port can be different if several servers are running) and start an ssh tunnel with port forwarding on your local computer:

ssh -N -L 8000:localhost:8888 jsc-cordex

The jupyterlab should then be available in your local browser at https://localhost:8000/. The login token can also be found in the URL on the jsc-terminal. Please don't forget to kill your server once you are finished. It will be killed automatically if you close the terminal in which you started the server.

Plots

These plots show some preliminary results from available data at jsc-cordex and show regional and yearly means in the PRUDENCE regions. These plots give an overview of what data is available and are updated once new data comes in.