Reference datasets

[JavierDiezSierra]

@larsbuntemeyer @gnikulin @SvenKotlarski @kgoergen @ClaasTeichmann @jesusff

The figures below show how the spatial and regional bias plots look when including an additional reference dataset (in this case, CERRA-Land for precipitation). Which other datasets do you think would be interesting to include?

[kgoergen]

Hi @JavierDiezSierra I understand you are looking for pan-EU, not national products or national products composites, reference data and not necessarily E-OBS; what about EMO (https://essd.copernicus.org/articles/14/3249/2022/); another dataset might be EURADCLIM (but this may be more for km-scale runs) (https://essd.copernicus.org/articles/15/1441/2023/)

[gnikulin]

EURADCLIM are too short, only covers 2013-2022. EMO-1 (about 1.5km, 1990-2022) is recommended instead of EMO-5 (not supported anymore, 1990-2019) and potentially may be used. However, I'm not sure that our focus is on a detailed intercomparison of observational gridded products for Europe such a study needs a separate paper.

[gnikulin]

As a minimum set of gridded products I would include E-OBS, CERRA(-Land) and ERA5.

[JavierDiezSierra]

Great! Then I'll include E-OBS, CERRA (-land for precipitation), and ERA5. I was thinking if another source of reference dataset would be interesting, such as radar or satellite data, since now we have one gridded observational dataset and two reanalysis products. But as you mentioned, @gnikulin, we can leave that for a separate paper.

[gnikulin]

I'm not aware about long-term radar- or satellite-based gridded precipitation datasets for Europe, may be they exist :-). A common practice is to calibrate such datasets using ground-based observations so they are not really independent.

[kgoergen]

To the best of my understanding EURADCLIM is the long-term radar data product (based on OPERA). I thought you wanted to look "beyond" ERA5 and E-OBS. :-)

[gnikulin]

Of course it's very good to look "beyond" our traditional ERA5 and E-OBS, completely agree. Under "long-term" I usually assume a 20- or 30-year period. Indeed, EURADCLIM is based on OPERA. As I mentioned EURADCLIM only covers 2013-2022 (https://dataplatform.knmi.nl/dataset/rad-opera-24h-rainfall-accumulation-euradclim-2-0) while we use 1989-2008 to compare CORDEX-CMIP5 and CORDEX-CMIP6. EURADCLIM is still can be used for some analysis. Potentially, EMO-1 can be utilised if we skip 1989 (https://jeodpp.jrc.ec.europa.eu/ftp/jrc-opendata/CEMS-EFAS/meteorological_forcings/EMO-1arcmin/pr/), although it includes 4 regional (national) hi-res observational products that may lead to spatial inhomogeneity.