You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This dataset has the same structure as the Euro-Cordex dataset. The domain is North America instead of Europe.
A pangeo-forge recipe for Euro-Cordex will likely be quite similar to the recipe for NA-Cordex, provided the target Zarr store structure seems appropriate.
How are the source files organized? (e.g. one file per day):
one file per variable+simulator+gridresolution+bias-correction+scenario combination
How are the source files accessed (e.g. FTP):
via Globus on NCAR's Glade filesystem, Glade location: /glade/collections/cdg/data/cordex/data/
Any special steps required to access the data (e.g. password required):
According to @rabernat, Globus 5 allows anonymous access to Glade
Transformation / Alignment / Merging
One Zarr store per variable+gridresolution+bias-correction combination. In other words, data are aggregated over simulators and scenarios. Each scenario has a time range and a rate of carbon emission; for example, one scenario covers the years 2006-2100, with the RCP 8.5 emission levels.
The GitHub repo with Pangeo-based python code for creating Zarr stores by hand is in the following GitHub location:
Although a single notebook is used to create these Zarr files, it reads in parameters from config.yaml to produce a particular subset of Zarr stores, as it was simply too difficult to handle edge cases easily. That is, certain Zarr stores required special chunking strategies to deal with smaller or larger number of simulation runs, for example. There were many such edge cases to deal with in different dimensions of the dataset.
Glade location for NetCDF data:
/glade/collections/cdg/data/cordex/data/
Output Dataset
Instructions for accessing the hand-created versions of the Target Output Zarr stores can found here:
Source Dataset
This dataset has the same structure as the Euro-Cordex dataset. The domain is North America instead of Europe.
A pangeo-forge recipe for Euro-Cordex will likely be quite similar to the recipe for NA-Cordex, provided the target Zarr store structure seems appropriate.
Transformation / Alignment / Merging
One Zarr store per variable+gridresolution+bias-correction combination. In other words, data are aggregated over simulators and scenarios. Each scenario has a time range and a rate of carbon emission; for example, one scenario covers the years 2006-2100, with the RCP 8.5 emission levels.
The GitHub repo with Pangeo-based python code for creating Zarr stores by hand is in the following GitHub location:
Although a single notebook is used to create these Zarr files, it reads in parameters from config.yaml to produce a particular subset of Zarr stores, as it was simply too difficult to handle edge cases easily. That is, certain Zarr stores required special chunking strategies to deal with smaller or larger number of simulation runs, for example. There were many such edge cases to deal with in different dimensions of the dataset.
Glade location for NetCDF data:
Output Dataset
Instructions for accessing the hand-created versions of the Target Output Zarr stores can found here:
The text was updated successfully, but these errors were encountered: