ERA-5-Land-T data availability

I went through ERA-5-Land documentation (ERA5-Land: data documentation - Copernicus Knowledge Base - ECMWF Confluence Wiki) and it seems that ERA5-Land-T (updated every ~5 days version of the dataset) is replaced with ERA5-Land (updated with 2-3 months lag) when it’s available: “Initial release data, i.e. data with just a few days behind real time, is called ERA5-Land-T. Both for the CDS and MARS, daily updates for ERA5-Land-T are available about 5 days behind real time and monthly mean updates are available about 5 days after the end of the month. The daily updates for ERA5-Land-T data on the CDS occur at no fixed time during the day. For the CDS, ERA5-Land-T data for a month is overwritten by the final ERA5-Land data about two months after the month in question.”

  1. Does it mean then that, as we make predictions on historical data in the Final Price Stage, what we download is actually ERA-5-Land (not ERA-5-Land-T) and we have to adjust to the 3 months lag? Could you confirm that if issue date is 2022-04-08, then the closest date of ERA-5-Land that could be used is 2022-01-08 (3 months difference)?

  2. Still, it is also stated in the Data format section of the documentation that “For GRIB format, ERA5-Land-T data can be identified by the key expver=0005 in the GRIB header. Consolidated ERA5-Land data is identified by the key expver=0001. For netCDF data requests which return just ERA5-Land or just ERA5-Land-T data, there is no means of differentiating between ERA5-Land and ERA5-Land-T data in the resulting netCDF files.”
    I’m not sure if adding expver=0005 to the query actually downloads all historical ERA5-Land-T data, as I’ve been struggling with processing data from GRIB format. Was someone successful in that and could confirm that it indeed contains all historical data only from ERA5-Land-T?

@progin Thanks for your patience waiting for a response on this.

Both ERA5-Land and ERA5-Land-T are considered approved for use in the competition.

To simplify usage, it’s okay to treat them as a single data source. This means that for the Final Prize Stage, you can use the historical ERA5-Land as if it were ERA5-Land-T data.

It’s possible that differences in data quality between ERA5-Land-T and ERA5-Land may impact quality of predictions if using models in a real operational setting where the actual data available in near-real-time is ERA5-Land-T. This would be a good topic to discuss in your final model report, especially if you are able to reference any studies on the differences between ERA5-Land and ERA5-Land-T.

Hope that helps. Let me know if you have any further questions.

1 Like