ECMWF Open Data (Meteorology Forecasts)

I figured this might be an interesting place to ask about this. Has anyone here used the data from ECMWF Open Data? I just stumbled into it but the website/documentation is a bit confusing to me. It’s not clear what is or isn’t available and how to access it, as they describe different accesses for different use cases. Also not clear if the “historical forecasts” are available openly as well.
I was thinking about using it as additional features to predict other stuff.

So, if you have experience with it, how is the interaction with the API? At what time is the data made available for medium/long range forecasts? Is it a JSON api? Etc etc…

1 Like

Yes, I am using ECMWF data for my work and the data are open to download. I would recommend to get an account at the Copernicus Climate Data Store and then you can download the data through python API requests. Historical forecasts are available e.g. here Copernicus Climate Data Store | Copernicus Climate Data Store you can click on the downloads tab, select a few variables/years and then generate an API request that generates a small python script to download the data you selected.

You can choose to download the data either in Grib or NetCDF format and we we have julia packages to read these:

3 Likes

Thank you Fabian. I guess for daily updates it would be as according to https://confluence.ecmwf.int/display/DAC/ECMWF+open+data%3A+real-time+forecasts+from+IFS+and+AIFS#ECMWFopendata:realtimeforecastsfromIFSandAIFS-Wheretofindthereal-timeopendata and for the historical forecasts to get them from Copernicus.

Any other documentation links you might have at hand on how the files are organized, etc?

I think this link describes the data model for both NetCDF and grib,
https://cds.climate.copernicus.eu/toolbox/doc/how-to/15_how_to_understand_the_common_data_model/15_how_to_understand_the_common_data_model.html

in the end most the data is stored as 3-dimensional arrays where the dimensions correspond to longitude, latitude and time.

2 Likes

Thanks, I really appreciate the help!